>>from around the globe. It's the Cube with digital coverage of AWS reinvent 2020 sponsored by Intel, AWS and our community partners. Welcome back to the Cubes Live coverage of AWS reinvent 2020. It's virtual this year. We're not in person because of the pandemic. We're doing the remote Cube Cube Virtual were the Cube virtual. I'm your host, John for here with Andy Jassy, the CEO of Amazon Web services, in for his annual at the end of the show comes on the Cube. This year, it's virtual Andy. Good to see you remotely in Seattle or in Palo Alto. Uh, Dave couldn't make it in a personal conflict, but he says, Hello, great to see you. >>Great to see you as well, John. It's an annual tradition. On the last day of reinvent. I wish we were doing it in person, but I'm glad at least were able to do it. Virtually >>the good news is, I know you could arrested last night normally at reinvent you just like we're all both losing our voice at the end of the show. At least me more than you, your and we're just at the end of like okay, Relief. It happens here. It's different. It's been three weeks has been virtual. Um, you guys had a unique format this year went much better than I expected. It would go on because I was pretty skeptical about these long, um, multiple days or weeks events. You guys did a good job of timing it out and creating these activations and with key news, starting with your keynote on December 1st. Now, at the end of the three weeks, um, tell me, are you surprised by the results? Can you give us, Ah, a feeling for how you think everything went? What's what's your take So far as we close out reinvented >>Well, I think it's going really well. I mean, we always gnome or a Z get past, reinvent and you start, you know, collecting all the feedback. But we've been watching all the metrics and you know, there's trade offs. Of course, now I think all of us giving our druthers would be together in Las Vegas, and I think it's hard to replace that feeling of being with people and the excitement of learning about things together and and making decisions together after you see different sessions that you're gonna make big changes in your company and for your customer experience. And yeah, and there's a community peace. And there's, you know, this from being there. There's a concert. The answer. I think people like being with one another. But, you know, I think this was the best that any of us could imagine doing doing a virtual event. And we had to really reinvent, reinvent and all the pieces to it. And now I think that some of the positive trade offs are they. You get a lot mawr engagement than you would normally get in person So normally. Last year, with about 65,000 people in Las Vegas this year, we had 530,000 people registered to reinvent and over 300,000 participate in some fashion. All the sessions had a lot more people who are participating just because you remove the constraints of of travel in costs, and so there are trade offs. I think we prefer being together, but I think it's been a really good community event, um, in learning event for for our customers, and we've been really pleased with it so >>far. No doubt I would totally agree with you. I think a lot of people like, Hey, I love to walk the floor and discover Harry and Sarah Davis moments of finding an exhibit her and the exhibit hall or or attending a session or going to a party, bumping into friends and seeing making new friends. But I think one of the things I want to get your reaction to it. So I think this is comes up. And, you know, we've been doing a lot of Q virtual for the past year, and and everyone pretty much agrees that when we go back, it's gonna be a hybrid world in the sense of events as well as cloud. You know that. But you know, I think one of the things that I noticed this year with reinvent is it almost was a democratization of reinvent. So you really had to reinvent the format. You had 300,000 plus people attend 500 pending email addresses, but now you've got a different kind of beehive community. So you're a bar raiser thinker. It's with the culture of Amazon. So I gotta ask you do the economics does this new kind of extra epiphany impact you and how you raise the bar to keep the best of the face to face when it comes back. And then if you keep the virtual any thoughts on how to leverage this and kind of get more open, it was free. You guys made it free this year and people did show up. >>Yeah, it's a really good question, and it's probably a question will be better equipped to answer in a month or two after we kind of debrief we always do after reading that we spend. Actually, I really enjoy the meeting because the team, the Collective A. W s team, works so hard in this event. There's so many months across everything. All the product teams, um, you know, all the marketing folks, all the event folks, and I think they do a terrific job with it. And we we do about 2.5 3 hour debrief on everything we did, things that we thought was really well the things that we thought we could do better and all the feedback we get from our community and so I wouldn't be surprised if we didn't find things from what we tried this year that we incorporate into what we do when we're back to being a person again. You know, of course, none of us really know when we'll be back in person again. Re event happens to fall on the time of the year, which is early December. And so you with with a lot of people seemingly able to get vaccinated, probably by you know, they'd spring early summer. You could kind of imagine that we might be able to reinvent in person next year. We'll have to see e think we all hope we will. But I'm sure there are a number of pieces that we will take from this and incorporate into what we do in person. And you know, then it's just a matter of how far you go. >>Fingers crossed and you know it's a hybrid world for the Cube two and reinvent and clouds. Let's get into the announcement. I want to get your your take as you look back now. I mean, how many announcements is you guys have me and a lot of announcements this year. Which ones did you like? Which one did you think were jumping off the page, which ones resonated the most or had impact. Can you share kind of just some stats on e mean how many announcements launches you did this >>year? But we had about 100 50 different new services and features that we announced over the last three weeks and reinvent And there, you know the question you're asking. I could easily spend another three hours like my Kino. You know, answering you all the ones that I like thought were important. You know, I think that, you know, some of the ones I think that really stood out for people. I think first on the compute side, I just think the, um the excitement around what we're doing with chips, um, is very clear. I think what we've done with gravitas to our generalized compute to give people 40% better price performance and they could find in the latest generation X 86 processors is just It's a huge deal. If you could save 40% price performance on computer, you get a lot more done for less on. Then you know some of the chip work we're doing in machine learning with inferential on the inference chips that we built And then what? We announced the trainee, um, on the machine learning training ship. People are very excited about the chip announcements. I think also, people on the container side is people are moving to smaller and smaller units of compute. I think people were very taken with the notion of E. K s and D. C s anywhere so they can run whatever container orchestration framework they're running in A. W s also on premises. To make it easier, Thio manage their deployments and containers. I think data stores was another space where I think people realize how much more data they're dealing with today. And we gave a couple statistics and the keynote that I think are kind of astonishing that, you know, every every hour today, people are creating mawr content that there was in an entire year, 20 years ago or the people expect more data to be created. The next three years in the prior 30 years combined these air astonishing numbers and it requires a brand new reinvention of data stores. And so I think people are very excited about Block Express, which is the first sand in the cloud and there really excited about Aurora in general, but then Aurora surveillance V two that allow you to scale up to hundreds of thousands of transactions per second and saved about 90% of supervision or people very excited about that. I think machine learning. You know, uh, Sage Maker has just been a game changer and the ease with which everyday developers and data scientists can build, train, tune into play machine learning models. And so we just keep knocking out things that are hard for people. Last year we launched the first i D for Machine Learning, the stage maker studio. This year, if you look at things that we announced, like Data Wrangler, which changes you know the process of Data Prep, which is one of the most time consuming pieces in machine learning or our feature store or the first see, I see deeper machine learning with pipelines or clarify, which allow you to have explain ability in your models. Those are big deals to people who are trying to build machine learning models, and you know that I'd say probably the last thing that we hear over and over again is really just the excitement around Connect, which is our call center service, which is just growing unbelievably fast and just, you know, the the fact that it's so easy to get started and so easy to scale so much more cost effective with, you know, built from the ground up on the cloud and with machine learning and ai embedded. And then adding some of the capabilities to give agents the right information, the right time about customers and products and real time capabilities for supervisors. Throw when calls were kind of going off the rails and to be ableto thio, stop the the contact before it becomes something, it hurts. The brand is there. Those are all big deals that people have been excited about. >>I think the connecting as I want to just jump on that for a second because I think when we first met many, many years ago, star eighth reinvent. You know the trends are always the same. You guys do a great job. Slew of announcements. You keep raising the bar. But one of the things that you mentioned to me when we talked about the origination of a W S was you were doing some stuff for Amazon proper, and you had a, you know, bootstrap team and you're solving your own problems, getting some scar tissue, the affiliate thing, all these examples. The trend is you guys tend to do stuff for yourself and then re factor it into potentially opportunities for your customers. And you're working backwards. All that good stuff. We'll get into that next section. But this year, more than ever, I think with the pandemic connect, you got chime, you got workspaces. This acceleration of you guys being pretty nimble on exposing these services. I mean, connect was a call center. It's an internal thing that you guys had been using. You re factored that for customer consumption. You see that kind of china? But you're not competing with Zoom. You're offering a service toe bundle in. Is this mawr relevant? Now, as you guys get bigger with more of these services because you're still big now you're still serving yourself. What? That seems to be a big trend now, coming out of the pandemic. Can you comment on um, >>yeah, It's a good question, John. And you know we do. We do a bunch of both. Frankly, you know, there there's some services where our customers. We're trying to solve certain problems and they tell us about those problems and then we build new services for him. So you know a good example that was red shift, which is our data warehouse and service, you know, two or three very large customers of ours. When we went to spend time with them and asked them what we could do to help them further, they just said, I wish I had a data warehousing service for the cloud that was built in the AWS style way. Um and they were really fed up with what they were using. Same thing was true with relation databases where people were just fed up with the old guard commercial, great commercial, great databases of Oracle and Sequel Server. And they hated the pricing and the proprietary nature of them and the punitive licensing. And they they wanted to move to these open engines like my sequel and post dress. But to get the same performance is the commercial great databases hard? So we solve that problem with them. With Aurora, which is our fastest growing service in our history, continues to be so there's sometimes when customers articulate a need, and we don't have a service that we've been running internally. But we way listen, and we have a very strong and innovative group of builders here where we build it for customers. And then there are other cases where customers say and connect with a great example of this. Connect with an example where some of our customers like into it. And Capital One said, You know, we need something for our contact center and customer service, and people weren't very happy with what they were using in that space. And they said, You, you've had to build something just to manage your retail business last 15, 20 years Can't you find a way to generalize that expose it? And when you have enough customers tell you that there's something that they want to use that you have experienced building. You start to think about it, and it's never a simple. It's just taking that technology and exposing it because it's often built, um, internally and you do a number of things to optimize it internally. But we have a way of building services and Amazon, where we do this working backwards process that you're referring to, where We build everything with the press release and frequently asked questions document, and we imagine that we're building it to be externalized even if it's an internal feature. But our feature for our retail business, it's only gonna be used as part of some other service that you never imagine Externalizing to third party developers. We always try and build it that way, and we always try to have well documented, hardened AP eyes so that other teams can use it without having to coordinate with those teams. And so it makes it easier for us to think about Externalizing it because we're a good part of the way there and we connect we. That's what we did way generalized it way built it from the ground up on top of the cloud. And then we embedded a bunch of AI and it so that people could do a number of things that would have taken him, you know, months to do with big development teams that they could really point, click and do so. We really try to do both. >>I think that's a great example of some of the scale benefits is worth calling out because that was a consistent theme this past year, The people we've reported on interviewed that Connect really was a lifeline for many during the pandemic and way >>have 5000 different customers who started using connect during the pandemic alone. Where they, you know, overnight they had to basically deal with having a a call center remotely. And so they picked up connect and they spun up call center remotely, and they didn't really quickly. And you know, it's that along with workspaces, which are virtual desktops in the cloud and things like Chime and some of our partners, Exume have really been lifelines for people. Thio have business continuity during a tandem. >>I think there's gonna be a whole set of new services that are gonna emerge You talked about in your keynote. We talked about it prior to the event where you know, if this pandemic hit with that five years ago, when there wasn't the advancements in, say, videoconferencing, it'd be a whole different world. And I think the whole world can see on full display that having integrated video communications and other cool things is gonna have a productivity benefit. And that's kind >>of could you imagine what the world would have been like the last nine months and we didn't have competent videoconferencing. I mean, just think about how different it would have been. And I think that all of these all of these capabilities today are kind of the occult 1.5 capabilities where, by the way, thank God for them. We've we've all been able to be productive because of them. But there's so early stage, they're all going to get evolved. I'm so significantly, I mean, even just today, you know, I was spending some time with with our team thinking about when we start to come back to the office and bigger numbers. And we do meetings with our remote partners, how we think about where the center of gravity should be and who should be on video conferencing and whether they should be allowed to kind of video conference in conference rooms, which are really hard to see them. We're only on their laptops, which are easier and what technology doesn't mean that you want in the conference rooms on both sides of the table, and how do you actually have it so that people who are remote could see which side of the table. I mean, all this stuff is yet to be invented. It will be very primitive for the next couple few years, even just interrupting one another in video conferencing people. When you do it, the sound counsel cancels each other out. So people don't really cut each other off and rip on one another. Same way, like all that, all that technology is going to get involved over time. It's a tremendous >>I could just see people fighting for the mute button. You know, that's power on these meetings. You know, Chuck on our team. All kidding aside, he was excited. We talked about Enron Kelly on your team, who runs product marketing on for your app side as well as computer networking storage. We're gonna do a green room app for the Q because you know, we're doing so many remote videos. We just did 112 here for reinvent one of things that people like is this idea of kind of being ready and kind of prepped. So again, this is a use case. We never would have thought off if there wasn't a pandemic. So and I think these are the kinds of innovation, thinking that seems small but works well when you start thinking about how easy it could be to say to integrate a chime through this sdk So this is the kind of things, that kind thing. So so with that, I want to get into your leadership principles because, you know, if you're a startup or a big company trying to reinvent, you're looking at the eight leadership principles you laid out, which were, um don't be afraid to reinvent. Acknowledge you can't fight gravity. Talent is hungry to reinvent solving real customer problems. Speed don't complex. If I use the platform with the broader set of tools, which is more a plug for you guys on cloud pull everything together with top down goals. Okay, great. How >>do you >>take those leadership principles and apply them broadly to companies and start ups? Because I think start ups in the garage are also gonna be there going. I'm going to jump on this wave. I'm inspired by the sea change. I'm gonna build something new or an enterprise. I'm gonna I'm gonna innovate. How do you How do you see these eight principles translating? >>Well, I think they're applicable to every company of every size and every industry and organization. Frankly, also, public sector organizations. I think in many ways startups have an advantage. And, you know, these were really keys to how to build a reinvention culture. And startups have an advantage because just by their very nature, they are inventive. You know, you can't you can't start a company that's a direct copy of somebody else that is an inventive where you have no chance. So startups already have, you know, a group of people that feel insurgent, and they wanted their passionate about certain customer experience. They want to invent it, and they know that they they only have so much time. Thio build something before money runs out and you know they have a number of those built in advantages. But I think larger companies are often where you see struggles and building a reinvention and invention culture and I've probably had in the last three weeks is part of reinvent probably about 40 different customer meetings with, you know, probably 75 different companies were accomplished in those or so and and I think that I met with a lot of leaders of companies where I think these reinvention principles really resonated, and I think they're they're battling with them and, you know, I think that it starts with the leaders if you, you know, when you have big companies that have been doing things a certain way for a long period of time, there's a fair bit of inertia that sets in and a lot of times not ill intended. It's just a big group of people in the middle who've been doing things a certain way for a long time and aren't that keen to change sometimes because it means ripping up something that they that they built and they remember how hard they worked on it. And sometimes it's because they don't know what it means for themselves. And you know, it takes the leadership team deciding that we are going to change. And usually that means they have to be able to have access to what's really happening in their business, what's really happening in their products in the market. But what customers really think of it and what they need to change and then having the courage and the energy, frankly, to pick the company up and push him to change because you're gonna have to fight a lot of inertia. So it always starts with the leaders. And in addition to having access that truth and deciding to make the change, you've gotta also set aggressive top down goal. The force of the organization moved faster than otherwise would and that also, sometimes leaders decide they're gonna want to change and they say they're going to change and they don't really set the goal. And they were kind of lessons and kind of doesn't listen. You know, we have a term the principal we have inside Amazon when we talk about the difference between good intentions and mechanisms and good intentions is saying we need to change and we need to invent, reinvent who we are and everyone has the right intentions. But nothing happens. Ah, mechanism, as opposed to good intention, is saying like Capital One did. We're going to reinvent our consumer digital banking platform in the next 18 months, and we're gonna meet every couple of weeks to see where we are into problem solved, like that's a mechanism. It's much harder to escape getting that done. Then somebody just saying we're going to reinvent, not checking on it, you know? And so, you know, I think that starts with the leaders. And then I think that you gotta have the right talent. You gotta have people who are excited about inventing, as opposed to really, Justin, what they built over a number of years, and yet at the same time, you're gonna make sure you don't hire people who were just building things that they're interested in. They went where they think the tech is cool as opposed to what customers want. And then I think you've got to Really You gotta build speed into your culture. And I think in some ways this is the very biggest challenge for a lot of enterprises. And I just I speak to so many leaders who kind of resigned themselves to moving slowly because they say you don't understand my like, companies big and the culture just move slow with regulator. There are a lot of reasons people will give you on why they have to move slow. But, you know, moving with speed is a choice. It's not something that your preordained with or not it is absolutely a leadership choice. And it can't happen overnight. You can't flip a switch and make it happen, but you can build a bunch of things into your culture first, starting with people. Understand that you are gonna move fast and then building an opportunity for people. Experiment quickly and reward people who experiment and to figure out the difference between one way doors and two way doors and things that are too way doors, letting people move quick and try things. You have to build that muscle or when it really comes, time to reinvent you won't have. >>That's a great point in the muscle on that's that's critical. You know, one of things I want to bring up. You brought on your keynote and you talk to me privately about it is you gave attribute in a way to Clay Christensen, who you called out on your keynote. Who was a professor at Harvard. Um, and he was you impressed by him and and you quoted him and he was He was your professor there, Um, your competitive person and you know, companies have strategy departments, and competitive strategy is not necessarily departments of mindset, and you were kind of brought this out in a zone undertone in your talk, we're saying you've got to be competitive in the sense of you got to survive and you've got to thrive. And you're kind of talking about rebuilding and building and, you know, Clay Christians. Innovative dilemma. Famous book is a mother, mother teachings around metrics and strategy and prescriptions. If he were alive today and he was with us, what would he be talking about? Because, you know, you have kind of stuck in the middle. Strategy was not Clay Christensen thing, but, you know, companies have to decide who they are. Their first principles face the truth. Some of the things you mentioned, what would we be talking with him about if we were talking about the innovator's dilemma with respect to, say, cloud and and some of the key decisions that have to be made right now? >>Well, then, Clay Christensen on it. Sounds like you read some of these books on. Guy had the fortunate, um, you know, being able to sit in classes that he taught. And also I got a chance. Thio, meet with him a couple of times after I graduated. Um, school, you know, kind of as more of a professional sorts. You can call me that. And, uh, he he was so thoughtful. He wasn't just thoughtful about innovation. He was thoughtful about how to get product market fit. And he was thoughtful about what your priorities in life were and how to build families. And, I mean, he really was one of the most thoughtful, innovative, um, you know, forward thinking, uh, strategist, I had the opportunity Thio encounter and that I've read, and so I'm very appreciative of having the opportunity Thio learn from him. And a lot of I mean, I think that he would probably be continuing to talk about a lot of the principles which I happen to think are evergreen that he he taught and there's it relates to the cloud. I think that one of the things that quite talked all the time about in all kinds of industries is that disruption always happens at the low end. It always happens with products that seem like they're not sophisticated enough. Don't do enough. And people always pooh pooh them because they say they won't do these things. And we learned this. I mean, I watched in the beginning of it of us. When we lost just three, we had so many people try and compare it Thio things like e m. C. And of course, it was very different than EMC. Um, but it was much simpler, but And it and it did a certain set of activities incredibly well at 1 1/100 of the price that's disrupted, you know, like 1 1/100 of the price. You find that builders, um, find a lot of utility for products like that. And so, you know, I think that it always starts with simple needs and products that aren't fully developed. That overtime continue to move their way up. Thio addressing Maura, Maura the market. And that's what we did with is what we've done with all our services. That's three and easy to and party ass and roar and things like that. And I think that there are lots of lessons is still apply. I think if you look at, um, containers and how that's changing what compute looks like, I think if you look at event driven, serverless compute in Lambda. Lambda is a great example of of really ah, derivative plays teaching, which is we knew when we were building Lambda that as people became excited about that programming model it would cannibalize easy to in our core compute service. And there are a lot of companies that won't do that. And for us we were trying to build a business that outlasts all of us. And that's you know, it's successful over a long period of time, and the the best way I know to do that is to listen to what customers We're trying to solve an event on their behalf, even if it means in the short term you may cannibalize yourself. And so that's what we always think about is, you know, wherever we see an opportunity to provide a better customer experience, even if it means in the short term, make cannibalism revenue leg lambda with complete with easy to our over our surveillance with provisions or are we're going to do it because we're gonna take the long view, and we believe that we serve customers well over a long period of time. We have a chance to do >>that. It's a cannibalize yourself and have someone else do it to you, right? That's that's the philosophy. Alright, fine. I know you've got tight for time. We got a you got a hard stop, But let's talk about the vaccine because you know, you brought up in the keynote carrier was a featured thing. And look at the news headlines. Now you got the shots being administered. You're starting to see, um, hashtag going around. I got my shot. So, you know, there's a There's a really Momenta. Mit's an uplifting vibe here. Amazon's involved in this and you talked about it. Can you share the innovation? There can just give us an update and what's come out of that and this supply chain factor. The cold chain. You guys were pretty instrumental in that share your your thoughts. >>We've been really excited and privileged partner with companies who are really trying to change what's possible for all of us. And I think you know it started with some of the companies producing vaccines. If you look at what we do with Moderna, where they built their digital manufacturing sweet on top of us in supply chain, where they used us for computing, storage and data warehousing and machine learning, and and on top of AWS they built, they're Cove in 19 vaccine candidate in 42 days when it normally takes 20 months. I mean, that is a total game changer. It's a game changer for all of us and getting the vaccine faster. But also, you just think about what that means for healthcare moving forward, it zits very exciting. And, yeah, I love what carriers doing. Kariya is building this product on top of AWS called links, which is giving them end and visibility over the transportation and in temperature of of the culture and everything they're delivering. And so it, uh, it changes what happens not only for food, ways and spoilage, but if you think about how much of the vaccine they're gonna actually transport to people and where several these vaccines need the right temperature control, it's it's a big deal. And what you know, I think there are a great example to what carrier is where. You know, if you think about the theme of this ring and then I talked about in my keynote, if you want to survive as an organization over a long period of time, you're gonna have to reinvent yourself. You're gonna have to probably do it. Multiple times over and the key to reinventing his first building, the right reinvention culture. And we talk about some of those principles earlier, but you also have to be aware of the technology that's available that allows you to do that. If you look at Carrier, they have built a very, very strong reinvention culture. And then, if you look at how they're leveraging, compute and storage and I o. T at the edge and machine learning, they know what's available, and they're using that technology to reinvent what's what's possible, and we're gonna all benefit because of >>it. All right. Well, Andy, you guys were reinventing the virtual space. Three weeks, it went off. Well, congratulations. Great to go along for the ride with the cube virtual. And again. Thank you for, um, keeping the show alive over there. Reinvent. Um, thanks for your team to for including the Cube. We really appreciate the Cube virtual being involved. Thank you. >>It's my pleasure. And thanks for having me, John and, uh, look forward to seeing you soon. >>All right? Take care. Have a hockey game in real life. When? When we get back, Andy Jesse, the CEO of a W s here to really wrap up. Reinvent here for Cuba, Virtual as well as the show. Today is the last day of the program. It will be online for the rest of the year and then into next month there's another wave coming, of course. Check out all the coverage. Come, come back, It's It's It's online. It's all free Cube Cube stuff is there on the Cube Channel. Silicon angle dot com For all the top stories, cube dot net tons of content on Twitter. Hashtag reinvent. You'll see all the commentary. Thanks for watching the Cube Virtual. I'm John Feehery.

>>don't talk mhm, >>Okay, thing is my presentation on coherent nonlinear dynamics and combinatorial optimization. This is going to be a talk to introduce an approach we're taking to the analysis of the performance of coherent using machines. So let me start with a brief introduction to easing optimization. The easing model represents a set of interacting magnetic moments or spins the total energy given by the expression shown at the bottom left of this slide. Here, the signal variables are meditate binary values. The Matrix element J. I. J. Represents the interaction, strength and signed between any pair of spins. I. J and A Chive represents a possible local magnetic field acting on each thing. The easing ground state problem is to find an assignment of binary spin values that achieves the lowest possible value of total energy. And an instance of the easing problem is specified by giving numerical values for the Matrix J in Vector H. Although the easy model originates in physics, we understand the ground state problem to correspond to what would be called quadratic binary optimization in the field of operations research and in fact, in terms of computational complexity theory, it could be established that the easing ground state problem is np complete. Qualitatively speaking, this makes the easing problem a representative sort of hard optimization problem, for which it is expected that the runtime required by any computational algorithm to find exact solutions should, as anatomically scale exponentially with the number of spends and for worst case instances at each end. Of course, there's no reason to believe that the problem instances that actually arrives in practical optimization scenarios are going to be worst case instances. And it's also not generally the case in practical optimization scenarios that we demand absolute optimum solutions. Usually we're more interested in just getting the best solution we can within an affordable cost, where costs may be measured in terms of time, service fees and or energy required for a computation. This focuses great interest on so called heuristic algorithms for the easing problem in other NP complete problems which generally get very good but not guaranteed optimum solutions and run much faster than algorithms that are designed to find absolute Optima. To get some feeling for present day numbers, we can consider the famous traveling salesman problem for which extensive compilations of benchmarking data may be found online. A recent study found that the best known TSP solver required median run times across the Library of Problem instances That scaled is a very steep route exponential for end up to approximately 4500. This gives some indication of the change in runtime scaling for generic as opposed the worst case problem instances. Some of the instances considered in this study were taken from a public library of T SPS derived from real world Veil aside design data. This feels I TSP Library includes instances within ranging from 131 to 744,710 instances from this library with end between 6880 13,584 were first solved just a few years ago in 2017 requiring days of run time and a 48 core to King hurts cluster, while instances with and greater than or equal to 14,233 remain unsolved exactly by any means. Approximate solutions, however, have been found by heuristic methods for all instances in the VLS i TSP library with, for example, a solution within 0.14% of a no lower bound, having been discovered, for instance, with an equal 19,289 requiring approximately two days of run time on a single core of 2.4 gigahertz. Now, if we simple mindedly extrapolate the root exponential scaling from the study up to an equal 4500, we might expect that an exact solver would require something more like a year of run time on the 48 core cluster used for the N equals 13,580 for instance, which shows how much a very small concession on the quality of the solution makes it possible to tackle much larger instances with much lower cost. At the extreme end, the largest TSP ever solved exactly has an equal 85,900. This is an instance derived from 19 eighties VLSI design, and it's required 136 CPU. Years of computation normalized to a single cord, 2.4 gigahertz. But the 24 larger so called world TSP benchmark instance within equals 1,904,711 has been solved approximately within ophthalmology. Gap bounded below 0.474%. Coming back to the general. Practical concerns have applied optimization. We may note that a recent meta study analyzed the performance of no fewer than 37 heuristic algorithms for Max cut and quadratic pioneer optimization problems and found the performance sort and found that different heuristics work best for different problem instances selected from a large scale heterogeneous test bed with some evidence but cryptic structure in terms of what types of problem instances were best solved by any given heuristic. Indeed, their their reasons to believe that these results from Mexico and quadratic binary optimization reflected general principle of performance complementarity among heuristic optimization algorithms in the practice of solving heart optimization problems there. The cerise is a critical pre processing issue of trying to guess which of a number of available good heuristic algorithms should be chosen to tackle a given problem. Instance, assuming that any one of them would incur high costs to run on a large problem, instances incidence, making an astute choice of heuristic is a crucial part of maximizing overall performance. Unfortunately, we still have very little conceptual insight about what makes a specific problem instance, good or bad for any given heuristic optimization algorithm. This has certainly been pinpointed by researchers in the field is a circumstance that must be addressed. So adding this all up, we see that a critical frontier for cutting edge academic research involves both the development of novel heuristic algorithms that deliver better performance, with lower cost on classes of problem instances that are underserved by existing approaches, as well as fundamental research to provide deep conceptual insight into what makes a given problem in, since easy or hard for such algorithms. In fact, these days, as we talk about the end of Moore's law and speculate about a so called second quantum revolution, it's natural to talk not only about novel algorithms for conventional CPUs but also about highly customized special purpose hardware architectures on which we may run entirely unconventional algorithms for combinatorial optimization such as easing problem. So against that backdrop, I'd like to use my remaining time to introduce our work on analysis of coherent using machine architectures and associate ID optimization algorithms. These machines, in general, are a novel class of information processing architectures for solving combinatorial optimization problems by embedding them in the dynamics of analog, physical or cyber physical systems, in contrast to both MAWR traditional engineering approaches that build using machines using conventional electron ICS and more radical proposals that would require large scale quantum entanglement. The emerging paradigm of coherent easing machines leverages coherent nonlinear dynamics in photonic or Opto electronic platforms to enable near term construction of large scale prototypes that leverage post Simoes information dynamics, the general structure of of current CM systems has shown in the figure on the right. The role of the easing spins is played by a train of optical pulses circulating around a fiber optical storage ring. A beam splitter inserted in the ring is used to periodically sample the amplitude of every optical pulse, and the measurement results are continually read into a refugee A, which uses them to compute perturbations to be applied to each pulse by a synchronized optical injections. These perturbations, air engineered to implement the spin, spin coupling and local magnetic field terms of the easing Hamiltonian, corresponding to a linear part of the CME Dynamics, a synchronously pumped parametric amplifier denoted here as PPL and Wave Guide adds a crucial nonlinear component to the CIA and Dynamics as well. In the basic CM algorithm, the pump power starts very low and has gradually increased at low pump powers. The amplitude of the easing spin pulses behaviors continuous, complex variables. Who Israel parts which can be positive or negative, play the role of play the role of soft or perhaps mean field spins once the pump, our crosses the threshold for parametric self oscillation. In the optical fiber ring, however, the attitudes of the easing spin pulses become effectively Qantas ized into binary values while the pump power is being ramped up. The F P J subsystem continuously applies its measurement based feedback. Implementation of the using Hamiltonian terms, the interplay of the linear rised using dynamics implemented by the F P G A and the threshold conversation dynamics provided by the sink pumped Parametric amplifier result in the final state of the optical optical pulse amplitude at the end of the pump ramp that could be read as a binary strain, giving a proposed solution of the easing ground state problem. This method of solving easing problem seems quite different from a conventional algorithm that runs entirely on a digital computer as a crucial aspect of the computation is performed physically by the analog, continuous, coherent, nonlinear dynamics of the optical degrees of freedom. In our efforts to analyze CIA and performance, we have therefore turned to the tools of dynamical systems theory, namely, a study of modifications, the evolution of critical points and apologies of hetero clinic orbits and basins of attraction. We conjecture that such analysis can provide fundamental insight into what makes certain optimization instances hard or easy for coherent using machines and hope that our approach can lead to both improvements of the course, the AM algorithm and a pre processing rubric for rapidly assessing the CME suitability of new instances. Okay, to provide a bit of intuition about how this all works, it may help to consider the threshold dynamics of just one or two optical parametric oscillators in the CME architecture just described. We can think of each of the pulse time slots circulating around the fiber ring, as are presenting an independent Opio. We can think of a single Opio degree of freedom as a single, resonant optical node that experiences linear dissipation, do toe out coupling loss and gain in a pump. Nonlinear crystal has shown in the diagram on the upper left of this slide as the pump power is increased from zero. As in the CME algorithm, the non linear game is initially to low toe overcome linear dissipation, and the Opio field remains in a near vacuum state at a critical threshold. Value gain. Equal participation in the Popeo undergoes a sort of lazing transition, and the study states of the OPIO above this threshold are essentially coherent states. There are actually two possible values of the Opio career in amplitude and any given above threshold pump power which are equal in magnitude but opposite in phase when the OPI across the special diet basically chooses one of the two possible phases randomly, resulting in the generation of a single bit of information. If we consider to uncoupled, Opio has shown in the upper right diagram pumped it exactly the same power at all times. Then, as the pump power has increased through threshold, each Opio will independently choose the phase and thus to random bits are generated for any number of uncoupled. Oppose the threshold power per opio is unchanged from the single Opio case. Now, however, consider a scenario in which the two appeals air, coupled to each other by a mutual injection of their out coupled fields has shown in the diagram on the lower right. One can imagine that depending on the sign of the coupling parameter Alfa, when one Opio is lazing, it will inject a perturbation into the other that may interfere either constructively or destructively, with the feel that it is trying to generate by its own lazing process. As a result, when came easily showed that for Alfa positive, there's an effective ferro magnetic coupling between the two Opio fields and their collective oscillation threshold is lowered from that of the independent Opio case. But on Lee for the two collective oscillation modes in which the two Opio phases are the same for Alfa Negative, the collective oscillation threshold is lowered on Lee for the configurations in which the Opio phases air opposite. So then, looking at how Alfa is related to the J. I. J matrix of the easing spin coupling Hamiltonian, it follows that we could use this simplistic to a p o. C. I am to solve the ground state problem of a fair magnetic or anti ferro magnetic ankles to easing model simply by increasing the pump power from zero and observing what phase relation occurs as the two appeals first start delays. Clearly, we can imagine generalizing this story toe larger, and however the story doesn't stay is clean and simple for all larger problem instances. And to find a more complicated example, we only need to go to n equals four for some choices of J J for n equals, for the story remains simple. Like the n equals two case. The figure on the upper left of this slide shows the energy of various critical points for a non frustrated and equals, for instance, in which the first bifurcated critical point that is the one that I forget to the lowest pump value a. Uh, this first bifurcated critical point flows as symptomatically into the lowest energy easing solution and the figure on the upper right. However, the first bifurcated critical point flows to a very good but sub optimal minimum at large pump power. The global minimum is actually given by a distinct critical critical point that first appears at a higher pump power and is not automatically connected to the origin. The basic C am algorithm is thus not able to find this global minimum. Such non ideal behaviors needs to become more confident. Larger end for the n equals 20 instance, showing the lower plots where the lower right plot is just a zoom into a region of the lower left lot. It can be seen that the global minimum corresponds to a critical point that first appears out of pump parameter, a around 0.16 at some distance from the idiomatic trajectory of the origin. That's curious to note that in both of these small and examples, however, the critical point corresponding to the global minimum appears relatively close to the idiomatic projector of the origin as compared to the most of the other local minima that appear. We're currently working to characterize the face portrait topology between the global minimum in the antibiotic trajectory of the origin, taking clues as to how the basic C am algorithm could be generalized to search for non idiomatic trajectories that jump to the global minimum during the pump ramp. Of course, n equals 20 is still too small to be of interest for practical optimization applications. But the advantage of beginning with the study of small instances is that we're able reliably to determine their global minima and to see how they relate to the 80 about trajectory of the origin in the basic C am algorithm. In the smaller and limit, we can also analyze fully quantum mechanical models of Syrian dynamics. But that's a topic for future talks. Um, existing large scale prototypes are pushing into the range of in equals 10 to the 4 10 to 5 to six. So our ultimate objective in theoretical analysis really has to be to try to say something about CIA and dynamics and regime of much larger in our initial approach to characterizing CIA and behavior in the large in regime relies on the use of random matrix theory, and this connects to prior research on spin classes, SK models and the tap equations etcetera. At present, we're focusing on statistical characterization of the CIA ingredient descent landscape, including the evolution of critical points in their Eigen value spectra. As the pump power is gradually increased. We're investigating, for example, whether there could be some way to exploit differences in the relative stability of the global minimum versus other local minima. We're also working to understand the deleterious or potentially beneficial effects of non ideologies, such as a symmetry in the implemented these and couplings. Looking one step ahead, we plan to move next in the direction of considering more realistic classes of problem instances such as quadratic, binary optimization with constraints. Eso In closing, I should acknowledge people who did the hard work on these things that I've shown eso. My group, including graduate students Ed winning, Daniel Wennberg, Tatsuya Nagamoto and Atsushi Yamamura, have been working in close collaboration with Syria Ganguly, Marty Fair and Amir Safarini Nini, all of us within the Department of Applied Physics at Stanford University. On also in collaboration with the Oshima Moto over at NTT 55 research labs, Onda should acknowledge funding support from the NSF by the Coherent Easing Machines Expedition in computing, also from NTT five research labs, Army Research Office and Exxon Mobil. Uh, that's it. Thanks very much. >>Mhm e >>t research and the Oshie for putting together this program and also the opportunity to speak here. My name is Al Gore ism or Andy and I'm from Caltech, and today I'm going to tell you about the work that we have been doing on networks off optical parametric oscillators and how we have been using them for icing machines and how we're pushing them toward Cornum photonics to acknowledge my team at Caltech, which is now eight graduate students and five researcher and postdocs as well as collaborators from all over the world, including entity research and also the funding from different places, including entity. So this talk is primarily about networks of resonate er's, and these networks are everywhere from nature. For instance, the brain, which is a network of oscillators all the way to optics and photonics and some of the biggest examples or metal materials, which is an array of small resonate er's. And we're recently the field of technological photonics, which is trying thio implement a lot of the technological behaviors of models in the condensed matter, physics in photonics and if you want to extend it even further, some of the implementations off quantum computing are technically networks of quantum oscillators. So we started thinking about these things in the context of icing machines, which is based on the icing problem, which is based on the icing model, which is the simple summation over the spins and spins can be their upward down and the couplings is given by the JJ. And the icing problem is, if you know J I J. What is the spin configuration that gives you the ground state? And this problem is shown to be an MP high problem. So it's computational e important because it's a representative of the MP problems on NPR. Problems are important because first, their heart and standard computers if you use a brute force algorithm and they're everywhere on the application side. That's why there is this demand for making a machine that can target these problems, and hopefully it can provide some meaningful computational benefit compared to the standard digital computers. So I've been building these icing machines based on this building block, which is a degenerate optical parametric. Oscillator on what it is is resonator with non linearity in it, and we pump these resonate er's and we generate the signal at half the frequency of the pump. One vote on a pump splits into two identical photons of signal, and they have some very interesting phase of frequency locking behaviors. And if you look at the phase locking behavior, you realize that you can actually have two possible phase states as the escalation result of these Opio which are off by pie, and that's one of the important characteristics of them. So I want to emphasize a little more on that and I have this mechanical analogy which are basically two simple pendulum. But there are parametric oscillators because I'm going to modulate the parameter of them in this video, which is the length of the string on by that modulation, which is that will make a pump. I'm gonna make a muscular. That'll make a signal which is half the frequency of the pump. And I have two of them to show you that they can acquire these face states so they're still facing frequency lock to the pump. But it can also lead in either the zero pie face states on. The idea is to use this binary phase to represent the binary icing spin. So each opio is going to represent spin, which can be either is your pie or up or down. And to implement the network of these resonate er's, we use the time off blood scheme, and the idea is that we put impulses in the cavity. These pulses air separated by the repetition period that you put in or t r. And you can think about these pulses in one resonator, xaz and temporarily separated synthetic resonate Er's if you want a couple of these resonator is to each other, and now you can introduce these delays, each of which is a multiple of TR. If you look at the shortest delay it couples resonator wanted to 2 to 3 and so on. If you look at the second delay, which is two times a rotation period, the couple's 123 and so on. And if you have and minus one delay lines, then you can have any potential couplings among these synthetic resonate er's. And if I can introduce these modulators in those delay lines so that I can strength, I can control the strength and the phase of these couplings at the right time. Then I can have a program will all toe all connected network in this time off like scheme, and the whole physical size of the system scales linearly with the number of pulses. So the idea of opium based icing machine is didn't having these o pos, each of them can be either zero pie and I can arbitrarily connect them to each other. And then I start with programming this machine to a given icing problem by just setting the couplings and setting the controllers in each of those delight lines. So now I have a network which represents an icing problem. Then the icing problem maps to finding the face state that satisfy maximum number of coupling constraints. And the way it happens is that the icing Hamiltonian maps to the linear loss of the network. And if I start adding gain by just putting pump into the network, then the OPI ohs are expected to oscillate in the lowest, lowest lost state. And, uh and we have been doing these in the past, uh, six or seven years and I'm just going to quickly show you the transition, especially what happened in the first implementation, which was using a free space optical system and then the guided wave implementation in 2016 and the measurement feedback idea which led to increasing the size and doing actual computation with these machines. So I just want to make this distinction here that, um, the first implementation was an all optical interaction. We also had an unequal 16 implementation. And then we transition to this measurement feedback idea, which I'll tell you quickly what it iss on. There's still a lot of ongoing work, especially on the entity side, to make larger machines using the measurement feedback. But I'm gonna mostly focused on the all optical networks and how we're using all optical networks to go beyond simulation of icing Hamiltonian both in the linear and non linear side and also how we're working on miniaturization of these Opio networks. So the first experiment, which was the four opium machine, it was a free space implementation and this is the actual picture off the machine and we implemented a small and it calls for Mexico problem on the machine. So one problem for one experiment and we ran the machine 1000 times, we looked at the state and we always saw it oscillate in one of these, um, ground states of the icing laboratoria. So then the measurement feedback idea was to replace those couplings and the controller with the simulator. So we basically simulated all those coherent interactions on on FB g. A. And we replicated the coherent pulse with respect to all those measurements. And then we injected it back into the cavity and on the near to you still remain. So it still is a non. They're dynamical system, but the linear side is all simulated. So there are lots of questions about if this system is preserving important information or not, or if it's gonna behave better. Computational wars. And that's still ah, lot of ongoing studies. But nevertheless, the reason that this implementation was very interesting is that you don't need the end minus one delight lines so you can just use one. Then you can implement a large machine, and then you can run several thousands of problems in the machine, and then you can compare the performance from the computational perspective Looks so I'm gonna split this idea of opium based icing machine into two parts. One is the linear part, which is if you take out the non linearity out of the resonator and just think about the connections. You can think about this as a simple matrix multiplication scheme. And that's basically what gives you the icing Hambletonian modeling. So the optical laws of this network corresponds to the icing Hamiltonian. And if I just want to show you the example of the n equals for experiment on all those face states and the history Graham that we saw, you can actually calculate the laws of each of those states because all those interferences in the beam splitters and the delay lines are going to give you a different losses. And then you will see that the ground states corresponds to the lowest laws of the actual optical network. If you add the non linearity, the simple way of thinking about what the non linearity does is that it provides to gain, and then you start bringing up the gain so that it hits the loss. Then you go through the game saturation or the threshold which is going to give you this phase bifurcation. So you go either to zero the pie face state. And the expectation is that Theis, the network oscillates in the lowest possible state, the lowest possible loss state. There are some challenges associated with this intensity Durban face transition, which I'm going to briefly talk about. I'm also going to tell you about other types of non aerodynamics that we're looking at on the non air side of these networks. So if you just think about the linear network, we're actually interested in looking at some technological behaviors in these networks. And the difference between looking at the technological behaviors and the icing uh, machine is that now, First of all, we're looking at the type of Hamilton Ian's that are a little different than the icing Hamilton. And one of the biggest difference is is that most of these technological Hamilton Ian's that require breaking the time reversal symmetry, meaning that you go from one spin to in the one side to another side and you get one phase. And if you go back where you get a different phase, and the other thing is that we're not just interested in finding the ground state, we're actually now interesting and looking at all sorts of states and looking at the dynamics and the behaviors of all these states in the network. So we started with the simplest implementation, of course, which is a one d chain of thes resonate, er's, which corresponds to a so called ssh model. In the technological work, we get the similar energy to los mapping and now we can actually look at the band structure on. This is an actual measurement that we get with this associate model and you see how it reasonably how How? Well, it actually follows the prediction and the theory. One of the interesting things about the time multiplexing implementation is that now you have the flexibility of changing the network as you are running the machine. And that's something unique about this time multiplex implementation so that we can actually look at the dynamics. And one example that we have looked at is we can actually go through the transition off going from top A logical to the to the standard nontrivial. I'm sorry to the trivial behavior of the network. You can then look at the edge states and you can also see the trivial and states and the technological at states actually showing up in this network. We have just recently implement on a two D, uh, network with Harper Hofstadter model and when you don't have the results here. But we're one of the other important characteristic of time multiplexing is that you can go to higher and higher dimensions and keeping that flexibility and dynamics, and we can also think about adding non linearity both in a classical and quantum regimes, which is going to give us a lot of exotic, no classical and quantum, non innate behaviors in these networks. Yeah, So I told you about the linear side. Mostly let me just switch gears and talk about the nonlinear side of the network. And the biggest thing that I talked about so far in the icing machine is this face transition that threshold. So the low threshold we have squeezed state in these. Oh, pios, if you increase the pump, we go through this intensity driven phase transition and then we got the face stays above threshold. And this is basically the mechanism off the computation in these O pos, which is through this phase transition below to above threshold. So one of the characteristics of this phase transition is that below threshold, you expect to see quantum states above threshold. You expect to see more classical states or coherent states, and that's basically corresponding to the intensity off the driving pump. So it's really hard to imagine that it can go above threshold. Or you can have this friends transition happen in the all in the quantum regime. And there are also some challenges associated with the intensity homogeneity off the network, which, for example, is if one opioid starts oscillating and then its intensity goes really high. Then it's going to ruin this collective decision making off the network because of the intensity driven face transition nature. So So the question is, can we look at other phase transitions? Can we utilize them for both computing? And also can we bring them to the quantum regime on? I'm going to specifically talk about the face transition in the spectral domain, which is the transition from the so called degenerate regime, which is what I mostly talked about to the non degenerate regime, which happens by just tuning the phase of the cavity. And what is interesting is that this phase transition corresponds to a distinct phase noise behavior. So in the degenerate regime, which we call it the order state, you're gonna have the phase being locked to the phase of the pump. As I talked about non degenerate regime. However, the phase is the phase is mostly dominated by the quantum diffusion. Off the off the phase, which is limited by the so called shallow towns limit, and you can see that transition from the general to non degenerate, which also has distinct symmetry differences. And this transition corresponds to a symmetry breaking in the non degenerate case. The signal can acquire any of those phases on the circle, so it has a you one symmetry. Okay, and if you go to the degenerate case, then that symmetry is broken and you only have zero pie face days I will look at. So now the question is can utilize this phase transition, which is a face driven phase transition, and can we use it for similar computational scheme? So that's one of the questions that were also thinking about. And it's not just this face transition is not just important for computing. It's also interesting from the sensing potentials and this face transition, you can easily bring it below threshold and just operated in the quantum regime. Either Gaussian or non Gaussian. If you make a network of Opio is now, we can see all sorts off more complicated and more interesting phase transitions in the spectral domain. One of them is the first order phase transition, which you get by just coupling to Opio, and that's a very abrupt face transition and compared to the to the single Opio phase transition. And if you do the couplings right, you can actually get a lot of non her mission dynamics and exceptional points, which are actually very interesting to explore both in the classical and quantum regime. And I should also mention that you can think about the cup links to be also nonlinear couplings. And that's another behavior that you can see, especially in the nonlinear in the non degenerate regime. So with that, I basically told you about these Opio networks, how we can think about the linear scheme and the linear behaviors and how we can think about the rich, nonlinear dynamics and non linear behaviors both in the classical and quantum regime. I want to switch gear and tell you a little bit about the miniaturization of these Opio networks. And of course, the motivation is if you look at the electron ICS and what we had 60 or 70 years ago with vacuum tube and how we transition from relatively small scale computers in the order of thousands of nonlinear elements to billions of non elements where we are now with the optics is probably very similar to 70 years ago, which is a table talk implementation. And the question is, how can we utilize nano photonics? I'm gonna just briefly show you the two directions on that which we're working on. One is based on lithium Diabate, and the other is based on even a smaller resonate er's could you? So the work on Nana Photonic lithium naive. It was started in collaboration with Harvard Marko Loncar, and also might affair at Stanford. And, uh, we could show that you can do the periodic polling in the phenomenon of it and get all sorts of very highly nonlinear processes happening in this net. Photonic periodically polls if, um Diabate. And now we're working on building. Opio was based on that kind of photonic the film Diabate. And these air some some examples of the devices that we have been building in the past few months, which I'm not gonna tell you more about. But the O. P. O. S. And the Opio Networks are in the works. And that's not the only way of making large networks. Um, but also I want to point out that The reason that these Nana photonic goblins are actually exciting is not just because you can make a large networks and it can make him compact in a in a small footprint. They also provide some opportunities in terms of the operation regime. On one of them is about making cat states and Opio, which is, can we have the quantum superposition of the zero pie states that I talked about and the Net a photonic within? I've It provides some opportunities to actually get closer to that regime because of the spatial temporal confinement that you can get in these wave guides. So we're doing some theory on that. We're confident that the type of non linearity two losses that it can get with these platforms are actually much higher than what you can get with other platform their existing platforms and to go even smaller. We have been asking the question off. What is the smallest possible Opio that you can make? Then you can think about really wavelength scale type, resonate er's and adding the chi to non linearity and see how and when you can get the Opio to operate. And recently, in collaboration with us see, we have been actually USC and Creole. We have demonstrated that you can use nano lasers and get some spin Hamilton and implementations on those networks. So if you can build the a P. O s, we know that there is a path for implementing Opio Networks on on such a nano scale. So we have looked at these calculations and we try to estimate the threshold of a pos. Let's say for me resonator and it turns out that it can actually be even lower than the type of bulk Pip Llano Pos that we have been building in the past 50 years or so. So we're working on the experiments and we're hoping that we can actually make even larger and larger scale Opio networks. So let me summarize the talk I told you about the opium networks and our work that has been going on on icing machines and the measurement feedback. And I told you about the ongoing work on the all optical implementations both on the linear side and also on the nonlinear behaviors. And I also told you a little bit about the efforts on miniaturization and going to the to the Nano scale. So with that, I would like Thio >>three from the University of Tokyo. Before I thought that would like to thank you showing all the stuff of entity for the invitation and the organization of this online meeting and also would like to say that it has been very exciting to see the growth of this new film lab. And I'm happy to share with you today of some of the recent works that have been done either by me or by character of Hong Kong. Honest Group indicates the title of my talk is a neuro more fic in silica simulator for the communities in machine. And here is the outline I would like to make the case that the simulation in digital Tektronix of the CME can be useful for the better understanding or improving its function principles by new job introducing some ideas from neural networks. This is what I will discuss in the first part and then it will show some proof of concept of the game and performance that can be obtained using dissimulation in the second part and the protection of the performance that can be achieved using a very large chaos simulator in the third part and finally talk about future plans. So first, let me start by comparing recently proposed izing machines using this table there is elected from recent natural tronics paper from the village Park hard people, and this comparison shows that there's always a trade off between energy efficiency, speed and scalability that depends on the physical implementation. So in red, here are the limitation of each of the servers hardware on, interestingly, the F p G, a based systems such as a producer, digital, another uh Toshiba beautification machine or a recently proposed restricted Bozeman machine, FPD A by a group in Berkeley. They offer a good compromise between speed and scalability. And this is why, despite the unique advantage that some of these older hardware have trust as the currency proposition in Fox, CBS or the energy efficiency off memory Sisters uh P. J. O are still an attractive platform for building large organizing machines in the near future. The reason for the good performance of Refugee A is not so much that they operate at the high frequency. No, there are particular in use, efficient, but rather that the physical wiring off its elements can be reconfigured in a way that limits the funding human bottleneck, larger, funny and phenols and the long propagation video information within the system. In this respect, the LPGA is They are interesting from the perspective off the physics off complex systems, but then the physics of the actions on the photos. So to put the performance of these various hardware and perspective, we can look at the competition of bringing the brain the brain complete, using billions of neurons using only 20 watts of power and operates. It's a very theoretically slow, if we can see and so this impressive characteristic, they motivate us to try to investigate. What kind of new inspired principles be useful for designing better izing machines? The idea of this research project in the future collaboration it's to temporary alleviates the limitations that are intrinsic to the realization of an optical cortex in machine shown in the top panel here. By designing a large care simulator in silicone in the bottom here that can be used for digesting the better organization principles of the CIA and this talk, I will talk about three neuro inspired principles that are the symmetry of connections, neural dynamics orphan chaotic because of symmetry, is interconnectivity the infrastructure? No. Next talks are not composed of the reputation of always the same types of non environments of the neurons, but there is a local structure that is repeated. So here's the schematic of the micro column in the cortex. And lastly, the Iraqi co organization of connectivity connectivity is organizing a tree structure in the brain. So here you see a representation of the Iraqi and organization of the monkey cerebral cortex. So how can these principles we used to improve the performance of the icing machines? And it's in sequence stimulation. So, first about the two of principles of the estimate Trian Rico structure. We know that the classical approximation of the car testing machine, which is the ground toe, the rate based on your networks. So in the case of the icing machines, uh, the okay, Scott approximation can be obtained using the trump active in your position, for example, so the times of both of the system they are, they can be described by the following ordinary differential equations on in which, in case of see, I am the X, I represent the in phase component of one GOP Oh, Theo f represents the monitor optical parts, the district optical Parametric amplification and some of the good I JoJo extra represent the coupling, which is done in the case of the measure of feedback coupling cm using oh, more than detection and refugee A and then injection off the cooking time and eso this dynamics in both cases of CNN in your networks, they can be written as the grand set of a potential function V, and this written here, and this potential functionally includes the rising Maccagnan. So this is why it's natural to use this type of, uh, dynamics to solve the icing problem in which the Omega I J or the eyes in coping and the H is the extension of the icing and attorney in India and expect so. Not that this potential function can only be defined if the Omega I j. R. A. Symmetric. So the well known problem of this approach is that this potential function V that we obtain is very non convicts at low temperature, and also one strategy is to gradually deformed this landscape, using so many in process. But there is no theorem. Unfortunately, that granted conventions to the global minimum of There's even Tony and using this approach. And so this is why we propose, uh, to introduce a macro structures of the system where one analog spin or one D O. P. O is replaced by a pair off one another spin and one error, according viable. And the addition of this chemical structure introduces a symmetry in the system, which in terms induces chaotic dynamics, a chaotic search rather than a learning process for searching for the ground state of the icing. Every 20 within this massacre structure the role of the er variable eyes to control the amplitude off the analog spins toe force. The amplitude of the expense toe become equal to certain target amplitude a uh and, uh, and this is done by modulating the strength off the icing complaints or see the the error variable E I multiply the icing complaint here in the dynamics off air d o p. O. On then the dynamics. The whole dynamics described by this coupled equations because the e I do not necessarily take away the same value for the different. I thesis introduces a symmetry in the system, which in turn creates security dynamics, which I'm sure here for solving certain current size off, um, escape problem, Uh, in which the X I are shown here and the i r from here and the value of the icing energy showing the bottom plots. You see this Celtics search that visit various local minima of the as Newtonian and eventually finds the global minimum? Um, it can be shown that this modulation off the target opportunity can be used to destabilize all the local minima off the icing evertonians so that we're gonna do not get stuck in any of them. On more over the other types of attractors I can eventually appear, such as limits I contractors, Okot contractors. They can also be destabilized using the motivation of the target and Batuta. And so we have proposed in the past two different moderation of the target amateur. The first one is a modulation that ensure the uh 100 reproduction rate of the system to become positive on this forbids the creation off any nontrivial tractors. And but in this work, I will talk about another moderation or arrested moderation which is given here. That works, uh, as well as this first uh, moderation, but is easy to be implemented on refugee. So this couple of the question that represent becoming the stimulation of the cortex in machine with some error correction they can be implemented especially efficiently on an F B. G. And here I show the time that it takes to simulate three system and also in red. You see, at the time that it takes to simulate the X I term the EI term, the dot product and the rising Hamiltonian for a system with 500 spins and Iraq Spain's equivalent to 500 g. O. P. S. So >>in >>f b d a. The nonlinear dynamics which, according to the digital optical Parametric amplification that the Opa off the CME can be computed in only 13 clock cycles at 300 yards. So which corresponds to about 0.1 microseconds. And this is Toby, uh, compared to what can be achieved in the measurements back O C. M. In which, if we want to get 500 timer chip Xia Pios with the one she got repetition rate through the obstacle nine narrative. Uh, then way would require 0.5 microseconds toe do this so the submission in F B J can be at least as fast as ah one g repression. Uh, replicate pulsed laser CIA Um, then the DOT product that appears in this differential equation can be completed in 43 clock cycles. That's to say, one microseconds at 15 years. So I pieced for pouring sizes that are larger than 500 speeds. The dot product becomes clearly the bottleneck, and this can be seen by looking at the the skating off the time the numbers of clock cycles a text to compute either the non in your optical parts or the dog products, respect to the problem size. And And if we had infinite amount of resources and PGA to simulate the dynamics, then the non illogical post can could be done in the old one. On the mattress Vector product could be done in the low carrot off, located off scales as a look at it off and and while the guide off end. Because computing the dot product involves assuming all the terms in the product, which is done by a nephew, GE by another tree, which heights scarce logarithmic any with the size of the system. But This is in the case if we had an infinite amount of resources on the LPGA food, but for dealing for larger problems off more than 100 spins. Usually we need to decompose the metrics into ah, smaller blocks with the block side that are not you here. And then the scaling becomes funny, non inner parts linear in the end, over you and for the products in the end of EU square eso typically for low NF pdf cheap PGA you the block size off this matrix is typically about 100. So clearly way want to make you as large as possible in order to maintain this scanning in a log event for the numbers of clock cycles needed to compute the product rather than this and square that occurs if we decompose the metrics into smaller blocks. But the difficulty in, uh, having this larger blocks eyes that having another tree very large Haider tree introduces a large finding and finance and long distance start a path within the refugee. So the solution to get higher performance for a simulator of the contest in machine eyes to get rid of this bottleneck for the dot product by increasing the size of this at the tree. And this can be done by organizing your critique the electrical components within the LPGA in order which is shown here in this, uh, right panel here in order to minimize the finding finance of the system and to minimize the long distance that a path in the in the fpt So I'm not going to the details of how this is implemented LPGA. But just to give you a idea off why the Iraqi Yahiko organization off the system becomes the extremely important toe get good performance for similar organizing machine. So instead of instead of getting into the details of the mpg implementation, I would like to give some few benchmark results off this simulator, uh, off the that that was used as a proof of concept for this idea which is can be found in this archive paper here and here. I should results for solving escape problems. Free connected person, randomly person minus one spring last problems and we sure, as we use as a metric the numbers of the mattress Victor products since it's the bottleneck of the computation, uh, to get the optimal solution of this escape problem with the Nina successful BT against the problem size here and and in red here, this propose FDJ implementation and in ah blue is the numbers of retrospective product that are necessary for the C. I am without error correction to solve this escape programs and in green here for noisy means in an evening which is, uh, behavior with similar to the Cartesian mission. Uh, and so clearly you see that the scaring off the numbers of matrix vector product necessary to solve this problem scales with a better exponents than this other approaches. So So So that's interesting feature of the system and next we can see what is the real time to solution to solve this SK instances eso in the last six years, the time institution in seconds to find a grand state of risk. Instances remain answers probability for different state of the art hardware. So in red is the F B g. A presentation proposing this paper and then the other curve represent Ah, brick a local search in in orange and silver lining in purple, for example. And so you see that the scaring off this purpose simulator is is rather good, and that for larger plant sizes we can get orders of magnitude faster than the state of the art approaches. Moreover, the relatively good scanning off the time to search in respect to problem size uh, they indicate that the FPD implementation would be faster than risk. Other recently proposed izing machine, such as the hope you know, natural complimented on memories distance that is very fast for small problem size in blue here, which is very fast for small problem size. But which scanning is not good on the same thing for the restricted Bosman machine. Implementing a PGA proposed by some group in Broken Recently Again, which is very fast for small parliament sizes but which canning is bad so that a dis worse than the proposed approach so that we can expect that for programs size is larger than 1000 spins. The proposed, of course, would be the faster one. Let me jump toe this other slide and another confirmation that the scheme scales well that you can find the maximum cut values off benchmark sets. The G sets better candidates that have been previously found by any other algorithms, so they are the best known could values to best of our knowledge. And, um or so which is shown in this paper table here in particular, the instances, uh, 14 and 15 of this G set can be We can find better converse than previously known, and we can find this can vary is 100 times faster than the state of the art algorithm and CP to do this which is a very common Kasich. It s not that getting this a good result on the G sets, they do not require ah, particular hard tuning of the parameters. So the tuning issuing here is very simple. It it just depends on the degree off connectivity within each graph. And so this good results on the set indicate that the proposed approach would be a good not only at solving escape problems in this problems, but all the types off graph sizing problems on Mexican province in communities. So given that the performance off the design depends on the height of this other tree, we can try to maximize the height of this other tree on a large F p g a onda and carefully routing the components within the P G A and and we can draw some projections of what type of performance we can achieve in the near future based on the, uh, implementation that we are currently working. So here you see projection for the time to solution way, then next property for solving this escape programs respect to the prime assize. And here, compared to different with such publicizing machines, particularly the digital. And, you know, 42 is shown in the green here, the green line without that's and, uh and we should two different, uh, hypothesis for this productions either that the time to solution scales as exponential off n or that the time of social skills as expression of square root off. So it seems, according to the data, that time solution scares more as an expression of square root of and also we can be sure on this and this production show that we probably can solve prime escape problem of science 2000 spins, uh, to find the rial ground state of this problem with 99 success ability in about 10 seconds, which is much faster than all the other proposed approaches. So one of the future plans for this current is in machine simulator. So the first thing is that we would like to make dissimulation closer to the rial, uh, GOP oh, optical system in particular for a first step to get closer to the system of a measurement back. See, I am. And to do this what is, uh, simulate Herbal on the p a is this quantum, uh, condoms Goshen model that is proposed described in this paper and proposed by people in the in the Entity group. And so the idea of this model is that instead of having the very simple or these and have shown previously, it includes paired all these that take into account on me the mean off the awesome leverage off the, uh, European face component, but also their violence s so that we can take into account more quantum effects off the g o p. O, such as the squeezing. And then we plan toe, make the simulator open access for the members to run their instances on the system. There will be a first version in September that will be just based on the simple common line access for the simulator and in which will have just a classic or approximation of the system. We don't know Sturm, binary weights and museum in term, but then will propose a second version that would extend the current arising machine to Iraq off F p g. A, in which we will add the more refined models truncated, ignoring the bottom Goshen model they just talked about on the support in which he valued waits for the rising problems and support the cement. So we will announce later when this is available and and far right is working >>hard comes from Universal down today in physics department, and I'd like to thank the organizers for their kind invitation to participate in this very interesting and promising workshop. Also like to say that I look forward to collaborations with with a file lab and Yoshi and collaborators on the topics of this world. So today I'll briefly talk about our attempt to understand the fundamental limits off another continues time computing, at least from the point off you off bullion satisfy ability, problem solving, using ordinary differential equations. But I think the issues that we raise, um, during this occasion actually apply to other other approaches on a log approaches as well and into other problems as well. I think everyone here knows what Dorien satisfy ability. Problems are, um, you have boolean variables. You have em clauses. Each of disjunction of collaterals literally is a variable, or it's, uh, negation. And the goal is to find an assignment to the variable, such that order clauses are true. This is a decision type problem from the MP class, which means you can checking polynomial time for satisfy ability off any assignment. And the three set is empty, complete with K three a larger, which means an efficient trees. That's over, uh, implies an efficient source for all the problems in the empty class, because all the problems in the empty class can be reduced in Polian on real time to reset. As a matter of fact, you can reduce the NP complete problems into each other. You can go from three set to set backing or two maximum dependent set, which is a set packing in graph theoretic notions or terms toe the icing graphs. A problem decision version. This is useful, and you're comparing different approaches, working on different kinds of problems when not all the closest can be satisfied. You're looking at the accusation version offset, uh called Max Set. And the goal here is to find assignment that satisfies the maximum number of clauses. And this is from the NPR class. In terms of applications. If we had inefficient sets over or np complete problems over, it was literally, positively influenced. Thousands off problems and applications in industry and and science. I'm not going to read this, but this this, of course, gives a strong motivation toe work on this kind of problems. Now our approach to set solving involves embedding the problem in a continuous space, and you use all the east to do that. So instead of working zeros and ones, we work with minus one across once, and we allow the corresponding variables toe change continuously between the two bounds. We formulate the problem with the help of a close metrics. If if a if a close, uh, does not contain a variable or its negation. The corresponding matrix element is zero. If it contains the variable in positive, for which one contains the variable in a gated for Mitt's negative one, and then we use this to formulate this products caused quote, close violation functions one for every clause, Uh, which really, continuously between zero and one. And they're zero if and only if the clause itself is true. Uh, then we form the define in order to define a dynamic such dynamics in this and dimensional hyper cube where the search happens and if they exist, solutions. They're sitting in some of the corners of this hyper cube. So we define this, uh, energy potential or landscape function shown here in a way that this is zero if and only if all the clauses all the kmc zero or the clauses off satisfied keeping these auxiliary variables a EMS always positive. And therefore, what you do here is a dynamics that is a essentially ingredient descend on this potential energy landscape. If you were to keep all the M's constant that it would get stuck in some local minimum. However, what we do here is we couple it with the dynamics we cooperated the clothes violation functions as shown here. And if he didn't have this am here just just the chaos. For example, you have essentially what case you have positive feedback. You have increasing variable. Uh, but in that case, you still get stuck would still behave will still find. So she is better than the constant version but still would get stuck only when you put here this a m which makes the dynamics in in this variable exponential like uh, only then it keeps searching until he finds a solution on deer is a reason for that. I'm not going toe talk about here, but essentially boils down toe performing a Grady and descend on a globally time barren landscape. And this is what works. Now I'm gonna talk about good or bad and maybe the ugly. Uh, this is, uh, this is What's good is that it's a hyperbolic dynamical system, which means that if you take any domain in the search space that doesn't have a solution in it or any socially than the number of trajectories in it decays exponentially quickly. And the decay rate is a characteristic in variant characteristic off the dynamics itself. Dynamical systems called the escape right the inverse off that is the time scale in which you find solutions by this by this dynamical system, and you can see here some song trajectories that are Kelty because it's it's no linear, but it's transient, chaotic. Give their sources, of course, because eventually knowledge to the solution. Now, in terms of performance here, what you show for a bunch off, um, constraint densities defined by M overran the ratio between closes toe variables for random, said Problems is random. Chris had problems, and they as its function off n And we look at money toward the wartime, the wall clock time and it behaves quite value behaves Azat party nominally until you actually he to reach the set on set transition where the hardest problems are found. But what's more interesting is if you monitor the continuous time t the performance in terms off the A narrow, continuous Time t because that seems to be a polynomial. And the way we show that is, we consider, uh, random case that random three set for a fixed constraint density Onda. We hear what you show here. Is that the right of the trash hold that it's really hard and, uh, the money through the fraction of problems that we have not been able to solve it. We select thousands of problems at that constraint ratio and resolve them without algorithm, and we monitor the fractional problems that have not yet been solved by continuous 90. And this, as you see these decays exponentially different. Educate rates for different system sizes, and in this spot shows that is dedicated behaves polynomial, or actually as a power law. So if you combine these two, you find that the time needed to solve all problems except maybe appear traction off them scales foreign or merely with the problem size. So you have paranormal, continuous time complexity. And this is also true for other types of very hard constraints and sexual problems such as exact cover, because you can always transform them into three set as we discussed before, Ramsey coloring and and on these problems, even algorithms like survey propagation will will fail. But this doesn't mean that P equals NP because what you have first of all, if you were toe implement these equations in a device whose behavior is described by these, uh, the keys. Then, of course, T the continue style variable becomes a physical work off. Time on that will be polynomial is scaling, but you have another other variables. Oxidative variables, which structured in an exponential manner. So if they represent currents or voltages in your realization and it would be an exponential cost Al Qaeda. But this is some kind of trade between time and energy, while I know how toe generate energy or I don't know how to generate time. But I know how to generate energy so it could use for it. But there's other issues as well, especially if you're trying toe do this son and digital machine but also happens. Problems happen appear. Other problems appear on in physical devices as well as we discuss later. So if you implement this in GPU, you can. Then you can get in order off to magnitude. Speed up. And you can also modify this to solve Max sad problems. Uh, quite efficiently. You are competitive with the best heuristic solvers. This is a weather problems. In 2016 Max set competition eso so this this is this is definitely this seems like a good approach, but there's off course interesting limitations, I would say interesting, because it kind of makes you think about what it means and how you can exploit this thes observations in understanding better on a low continues time complexity. If you monitored the discrete number the number of discrete steps. Don't buy the room, Dakota integrator. When you solve this on a digital machine, you're using some kind of integrator. Um and you're using the same approach. But now you measure the number off problems you haven't sold by given number of this kid, uh, steps taken by the integrator. You find out you have exponential, discrete time, complexity and, of course, thistles. A problem. And if you look closely, what happens even though the analog mathematical trajectory, that's the record here. If you monitor what happens in discrete time, uh, the integrator frustrates very little. So this is like, you know, third or for the disposition, but fluctuates like crazy. So it really is like the intervention frees us out. And this is because of the phenomenon of stiffness that are I'll talk a little bit a more about little bit layer eso. >>You know, it might look >>like an integration issue on digital machines that you could improve and could definitely improve. But actually issues bigger than that. It's It's deeper than that, because on a digital machine there is no time energy conversion. So the outside variables are efficiently representing a digital machine. So there's no exponential fluctuating current of wattage in your computer when you do this. Eso If it is not equal NP then the exponential time, complexity or exponential costs complexity has to hit you somewhere. And this is how um, but, you know, one would be tempted to think maybe this wouldn't be an issue in a analog device, and to some extent is true on our devices can be ordered to maintain faster, but they also suffer from their own problems because he not gonna be affect. That classes soldiers as well. So, indeed, if you look at other systems like Mirandizing machine measurement feedback, probably talk on the grass or selected networks. They're all hinge on some kind off our ability to control your variables in arbitrary, high precision and a certain networks you want toe read out across frequencies in case off CM's. You required identical and program because which is hard to keep, and they kind of fluctuate away from one another, shift away from one another. And if you control that, of course that you can control the performance. So actually one can ask if whether or not this is a universal bottleneck and it seems so aside, I will argue next. Um, we can recall a fundamental result by by showing harder in reaction Target from 1978. Who says that it's a purely computer science proof that if you are able toe, compute the addition multiplication division off riel variables with infinite precision, then you could solve any complete problems in polynomial time. It doesn't actually proposals all where he just chose mathematically that this would be the case. Now, of course, in Real warned, you have also precision. So the next question is, how does that affect the competition about problems? This is what you're after. Lots of precision means information also, or entropy production. Eso what you're really looking at the relationship between hardness and cost of computing off a problem. Uh, and according to Sean Hagar, there's this left branch which in principle could be polynomial time. But the question whether or not this is achievable that is not achievable, but something more cheerful. That's on the right hand side. There's always going to be some information loss, so mental degeneration that could keep you away from possibly from point normal time. So this is what we like to understand, and this information laws the source off. This is not just always I will argue, uh, in any physical system, but it's also off algorithm nature, so that is a questionable area or approach. But China gets results. Security theoretical. No, actual solar is proposed. So we can ask, you know, just theoretically get out off. Curiosity would in principle be such soldiers because it is not proposing a soldier with such properties. In principle, if if you want to look mathematically precisely what the solar does would have the right properties on, I argue. Yes, I don't have a mathematical proof, but I have some arguments that that would be the case. And this is the case for actually our city there solver that if you could calculate its trajectory in a loss this way, then it would be, uh, would solve epic complete problems in polynomial continuous time. Now, as a matter of fact, this a bit more difficult question, because time in all these can be re scared however you want. So what? Burns says that you actually have to measure the length of the trajectory, which is a new variant off the dynamical system or property dynamical system, not off its parameters ization. And we did that. So Suba Corral, my student did that first, improving on the stiffness off the problem off the integrations, using implicit solvers and some smart tricks such that you actually are closer to the actual trajectory and using the same approach. You know what fraction off problems you can solve? We did not give the length of the trajectory. You find that it is putting on nearly scaling the problem sites we have putting on your skin complexity. That means that our solar is both Polly length and, as it is, defined it also poorly time analog solver. But if you look at as a discreet algorithm, if you measure the discrete steps on a digital machine, it is an exponential solver. And the reason is because off all these stiffness, every integrator has tow truck it digitizing truncate the equations, and what it has to do is to keep the integration between the so called stability region for for that scheme, and you have to keep this product within a grimace of Jacoby in and the step size read in this region. If you use explicit methods. You want to stay within this region? Uh, but what happens that some off the Eigen values grow fast for Steve problems, and then you're you're forced to reduce that t so the product stays in this bonded domain, which means that now you have to you're forced to take smaller and smaller times, So you're you're freezing out the integration and what I will show you. That's the case. Now you can move to increase its soldiers, which is which is a tree. In this case, you have to make domain is actually on the outside. But what happens in this case is some of the Eigen values of the Jacobean, also, for six systems, start to move to zero. As they're moving to zero, they're going to enter this instability region, so your soul is going to try to keep it out, so it's going to increase the data T. But if you increase that to increase the truncation hours, so you get randomized, uh, in the large search space, so it's it's really not, uh, not going to work out. Now, one can sort off introduce a theory or language to discuss computational and are computational complexity, using the language from dynamical systems theory. But basically I I don't have time to go into this, but you have for heart problems. Security object the chaotic satellite Ouch! In the middle of the search space somewhere, and that dictates how the dynamics happens and variant properties off the dynamics. Of course, off that saddle is what the targets performance and many things, so a new, important measure that we find that it's also helpful in describing thesis. Another complexity is the so called called Makarov, or metric entropy and basically what this does in an intuitive A eyes, uh, to describe the rate at which the uncertainty containing the insignificant digits off a trajectory in the back, the flow towards the significant ones as you lose information because off arrows being, uh grown or are developed in tow. Larger errors in an exponential at an exponential rate because you have positively up north spawning. But this is an in variant property. It's the property of the set of all. This is not how you compute them, and it's really the interesting create off accuracy philosopher dynamical system. A zay said that you have in such a high dimensional that I'm consistent were positive and negatively upon of exponents. Aziz Many The total is the dimension of space and user dimension, the number off unstable manifold dimensions and as Saddam was stable, manifold direction. And there's an interesting and I think, important passion, equality, equality called the passion, equality that connect the information theoretic aspect the rate off information loss with the geometric rate of which trajectory separate minus kappa, which is the escape rate that I already talked about. Now one can actually prove a simple theorems like back off the envelope calculation. The idea here is that you know the rate at which the largest rated, which closely started trajectory separate from one another. So now you can say that, uh, that is fine, as long as my trajectory finds the solution before the projective separate too quickly. In that case, I can have the hope that if I start from some region off the face base, several close early started trajectories, they kind of go into the same solution orphaned and and that's that's That's this upper bound of this limit, and it is really showing that it has to be. It's an exponentially small number. What? It depends on the end dependence off the exponents right here, which combines information loss rate and the social time performance. So these, if this exponents here or that has a large independence or river linear independence, then you then you really have to start, uh, trajectories exponentially closer to one another in orderto end up in the same order. So this is sort off like the direction that you're going in tow, and this formulation is applicable toe all dynamical systems, uh, deterministic dynamical systems. And I think we can We can expand this further because, uh, there is, ah, way off getting the expression for the escaped rate in terms off n the number of variables from cycle expansions that I don't have time to talk about. What? It's kind of like a program that you can try toe pursuit, and this is it. So the conclusions I think of self explanatory I think there is a lot of future in in, uh, in an allo. Continue start computing. Um, they can be efficient by orders of magnitude and digital ones in solving empty heart problems because, first of all, many of the systems you like the phone line and bottleneck. There's parallelism involved, and and you can also have a large spectrum or continues time, time dynamical algorithms than discrete ones. And you know. But we also have to be mindful off. What are the possibility of what are the limits? And 11 open question is very important. Open question is, you know, what are these limits? Is there some kind off no go theory? And that tells you that you can never perform better than this limit or that limit? And I think that's that's the exciting part toe to derive thes thes this levian 10.

Hi, I'm Hideo Mabuchi from Stanford University. This is my presentation on coherent nonlinear dynamics, and combinatorial optimization. This is going to be a talk, to introduce an approach, we are taking to the analysis, of the performance of Coherent Ising Machines. So let me start with a brief introduction, to ising optimization. The ising model, represents a set of interacting magnetic moments or spins, with total energy given by the expression, shown at the bottom left of the slide. Here the cigna variables are meant to take binary values. The matrix element jij, represents the interaction, strength and sign, between any pair of spins ij, and hi represents a possible local magnetic field, acting on each thing. The ising ground state problem, is defined in an assignment of binary spin values, that achieves the lowest possible value of total energy. And an instance of the easing problem, is specified by given numerical values, for the matrix j and vector h, although the ising model originates in physics, we understand the ground state problem, to correspond to what would be called, quadratic binary optimization, in the field of operations research. And in fact, in terms of computational complexity theory, it can be established that the, ising ground state problem is NP complete. Qualitatively speaking, this makes the ising problem, a representative sort of hard optimization problem, for which it is expected, that the runtime required by any computational algorithm, to find exact solutions, should asyntonically scale, exponentially with the number of spins, and four worst case instances at each end. Of course, there's no reason to believe that, the problem instances that actually arise, in practical optimization scenarios, are going to be worst case instances. And it's also not generally the case, in practical optimization scenarios, that we demand absolute optimum solutions. Usually we're more interested in, just getting the best solution we can, within an affordable cost, where costs may be measured in terms of time, service fees and or energy required for computation. This focus is great interest on, so-called heuristic algorithms, for the ising problem and other NP complete problems, which generally get very good, but not guaranteed optimum solutions, and run much faster than algorithms, that are designed to find absolute Optima. To get some feeling for present day numbers, we can consider the famous traveling salesman problem, for which extensive compilations, of benchmarking data may be found online. A recent study found that, the best known TSP solver required median runtimes, across a library of problem instances, that scaled as a very steep route exponential, for an up to approximately 4,500. This gives some indication of the change, in runtime scaling for generic, as opposed to worst case problem instances. Some of the instances considered in this study, were taken from a public library of TSPs, derived from real world VLSI design data. This VLSI TSP library, includes instances within ranging from 131 to 744,710, instances from this library within between 6,880 and 13,584, were first solved just a few years ago, in 2017 requiring days of runtime, and a 48 core two gigahertz cluster, all instances with n greater than or equal to 14,233, remain unsolved exactly by any means. Approximate solutions however, have been found by heuristic methods, for all instances in the VLSI TSP library, with, for example, a solution within 0.014% of a known lower bound, having been discovered for an instance, with n equal 19,289, requiring approximately two days of runtime, on a single quarter at 2.4 gigahertz. Now, if we simple-minded the extrapolate, the route exponential scaling, from the study yet to n equal 4,500, we might expect that an exact solver, would require something more like a year of runtime, on the 48 core cluster, used for the n equals 13,580 for instance, which shows how much, a very small concession on the quality of the solution, makes it possible to tackle much larger instances, with much lower costs, at the extreme end, the largest TSP ever solved exactly has n equal 85,900. This is an instance derived from 1980s VLSI design, and this required 136 CPU years of computation, normalized to a single core, 2.4 gigahertz. But the 20 fold larger, so-called world TSP benchmark instance, with n equals 1,904,711, has been solved approximately, with an optimality gap bounded below 0.0474%. Coming back to the general practical concerns, of applied optimization. We may note that a recent meta study, analyze the performance of no fewer than, 37 heuristic algorithms for MaxCut, and quadratic binary optimization problems. And find the performance... Sorry, and found that a different heuristics, work best for different problem instances, selected from a large scale heterogeneous test bed, with some evidence, the cryptic structure, in terms of what types of problem instances, were best solved by any given heuristic. Indeed, there are reasons to believe, that these results for MaxCut, and quadratic binary optimization, reflect to general principle, of a performance complementarity, among heuristic optimization algorithms, and the practice of solving hard optimization problems. There thus arises the critical pre processing issue, of trying to guess, which of a number of available, good heuristic algorithms should be chosen, to tackle a given problem instance. Assuming that any one of them, would incur high cost to run, on a large problem of incidents, making an astute choice of heuristic, is a crucial part of maximizing overall performance. Unfortunately, we still have very little conceptual insight, about what makes a specific problem instance, good or bad for any given heuristic optimization algorithm. This is certainly pinpointed by researchers in the field, as a circumstance and must be addressed. So adding this all up, we see that a critical frontier, for cutting edge academic research involves, both the development of novel heuristic algorithms, that deliver better performance with lower costs, on classes of problem instances, that are underserved by existing approaches, as well as fundamental research, to provide deep conceptual insight, into what makes a given problem instance, easy or hard for such algorithms. In fact, these days, as we talk about the end of Moore's law, and speculate about a so-called second quantum revolution, it's natural to talk not only about novel algorithms, for conventional CPUs, but also about highly customized, special purpose hardware architectures, on which we may run entirely unconventional algorithms, for common tutorial optimizations, such as ising problem. So against that backdrop, I'd like to use my remaining time, to introduce our work on, analysis of coherent using machine architectures, and associated optimization algorithms. Ising machines in general, are a novel class of information processing architectures, for solving combinatorial optimization problems, by embedding them in the dynamics, of analog, physical, or a cyber-physical systems. In contrast to both more traditional engineering approaches, that build ising machines using conventional electronics, and more radical proposals, that would require large scale quantum entanglement the emerging paradigm of coherent ising machines, leverages coherent nominal dynamics, in photonic or optical electronic platforms, to enable near term construction, of large scale prototypes, that leverage posting as information dynamics. The general structure of current of current CIM systems, as shown in the figure on the right, the role of the easing spins, is played by a train of optical pulses, circulating around a fiber optical storage ring, that beam splitter inserted in the ring, is used to periodically sample, the amplitude of every optical pulse. And the measurement results, are continually read into an FPGA, which uses then to compute perturbations, to be applied to each pulse, by a synchronized optical injections. These perturbations are engineered to implement, the spin-spin coupling and local magnetic field terms, of the ising hamiltonian, corresponding to a linear part of the CIM dynamics. Asynchronously pumped parametric amplifier, denoted here as PPL and wave guide, adds a crucial nonlinear component, to the CIM dynamics as well. And the basic CIM algorithm, the pump power starts very low, and is gradually increased, at low pump powers, the amplitudes of the easing spin pulses, behave as continuous complex variables, whose real parts which can be positive or negative, by the role of soft or perhaps mean field spins. Once the pump power crosses the threshold, for perimetric self oscillation in the optical fiber ring, however, the amplitudes of the easing spin pulses, become effectively quantized into binary values, while the pump power is being ramped up, the FPGA subsystem continuously applies, its measurement based feedback implementation, of the using hamiltonian terms. The interplay of the linearized easing dynamics, implemented by the FPGA , and the thresholds quantization dynamics, provided by the sink pumped parametric amplifier, result in a final state, of the optical plus amplitudes, at the end of the pump ramp, that can be read as a binary strain, giving a proposed solution, of the ising ground state problem. This method of solving ising problems, seems quite different from a conventional algorithm, that runs entirely on a digital computer. As a crucial aspect, of the computation is performed physically, by the analog continuous coherent nonlinear dynamics, of the optical degrees of freedom, in our efforts to analyze CA and performance. We have therefore turn to dynamical systems theory. Namely a study of bifurcations, the evolution of critical points, and typologies of heteroclitic orbits, and basins of attraction. We conjecture that such analysis, can provide fundamental insight, into what makes certain optimization instances, hard or easy for coherent ising machines, and hope that our approach, can lead to both improvements of the course CIM algorithm, and the pre processing rubric, for rapidly assessing the CIM insuibility of the instances. To provide a bit of intuition about how this all works. It may help to consider the threshold dynamics, of just one or two optical parametric oscillators, in the CIM architecture just described. We can think of each of the pulse time slots, circulating around the fiber ring, as are presenting an independent OPO. We can think of a single OPO degree of freedom, as a single resonant optical mode, that experiences linear dissipation, due to coupling loss, and gain in a pump near crystal, as shown in the diagram on the upper left of the slide, as the pump power is increased from zero. As in the CIM algorithm, the non-linear gain is initially too low, to overcome linear dissipation. And the OPO field remains in a near vacuum state, at a critical threshold value, gain equals dissipation, and the OPO undergoes a sort of lasing transition. And the steady States of the OPO, above this threshold are essentially coherent States. There are actually two possible values, of the OPO coherent amplitude, and any given above threshold pump power, which are equal in magnitude, but opposite in phase, when the OPO cross this threshold, it basically chooses one of the two possible phases, randomly, resulting in the generation, of a single bit of information. If we consider two uncoupled OPOs, as shown in the upper right diagram, pumped at exactly the same power at all times, then as the pump power is increased through threshold, each OPO will independently choose a phase, and thus two random bits are generated, for any number of uncoupled OPOs, the threshold power per OPOs is unchanged, from the single OPO case. Now, however, consider a scenario, in which the two appeals are coupled to each other, by a mutual injection of their out coupled fields, as shown in the diagram on the lower right. One can imagine that, depending on the sign of the coupling parameter alpha, when one OPO is lasing, it will inject a perturbation into the other, that may interfere either constructively or destructively, with the field that it is trying to generate, via its own lasing process. As a result, when can easily show that for alpha positive, there's an effective ferromagnetic coupling, between the two OPO fields, and their collective oscillation threshold, is lowered from that of the independent OPO case, but only for the two collective oscillation modes, in which the two OPO phases are the same. For alpha negative, the collective oscillation threshold, is lowered only for the configurations, in which the OPO phases are opposite. So then looking at how alpha is related to the jij matrix, of the ising spin coupling hamilitonian, it follows the, we could use this simplistic to OPO CIM, to solve the ground state problem, of the ferromagnetic or antiferromagnetic angles, to ising model, simply by increasing the pump power, from zero and observing what phase relation occurs, as the two appeals first start to lase. Clearly we can imagine generalizing the story to larger, and, however, the story doesn't stay as clean and simple, for all larger problem instances. And to find a more complicated example, we only need to go to n equals four, for some choices of jij for n equals four, the story remains simple, like the n equals two case. The figure on the upper left of this slide, shows the energy of various critical points, for a non frustrated n equals for instance, in which the first bifurcated critical point, that is the one that, by forgets of the lowest pump value a, this first bifurcated critical point, flows asyntonically into the lowest energy using solution, and the figure on the upper right, however, the first bifurcated critical point, flows to a very good, but suboptimal minimum at large pump power. The global minimum is actually given, by a distinct critical point. The first appears at a higher pump power, and is not needed radically connected to the origin. The basic CIM algorithm, is this not able to find this global minimum, such non-ideal behavior, seems to become more common at margin end, for the n equals 20 instance show in the lower plots, where the lower right pod is just a zoom into, a region of the lower left block. It can be seen that the global minimum, corresponds to a critical point, that first appears that of pump parameter a around 0.16, at some distance from the adriatic trajectory of the origin. That's curious to note that, in both of these small and examples, however, the critical point corresponding to the global minimum, appears relatively close, to the adiabatic trajectory of the origin, as compared to the most of the other, local minimum that appear. We're currently working to characterise, the face portrait typology, between the global minimum, and the adiabatic trajectory of the origin, taking clues as to how the basic CIM algorithm, could be generalized, to search for non-adiabatic trajectories, that jumped to the global minimum, during the pump up, of course, n equals 20 is still too small, to be of interest for practical optimization applications. But the advantage of beginning, with the study of small instances, is that we're able to reliably to determine, their global minima, and to see how they relate to the idea, that trajectory of the origin, and the basic CIM algorithm. And the small land limit, We can also analyze, for the quantum mechanical models of CAM dynamics, but that's a topic for future talks. Existing large-scale prototypes, are pushing into the range of, n equals, 10 to the four, 10 to the five, 10 to the six. So our ultimate objective in theoretical analysis, really has to be, to try to say something about CAM dynamics, and regime of much larger in. Our initial approach to characterizing CAM behavior, in the large end regime, relies on the use of random matrix theory. And this connects to prior research on spin classes, SK models, and the tap equations, et cetera, at present we're focusing on, statistical characterization, of the CIM gradient descent landscape, including the evolution of critical points, And their value spectra, as the pump powers gradually increase. We're investigating, for example, whether there could be some way, to explain differences in the relative stability, of the global minimum versus other local minima. We're also working to understand the deleterious, or potentially beneficial effects, of non-ideologies such as asymmetry, in the implemented using couplings, looking one step ahead, we plan to move next into the direction, of considering more realistic classes of problem instances, such as quadratic binary optimization with constraints. So in closing I should acknowledge, people who did the hard work, on these things that I've shown. So my group, including graduate students, Edwin Ng, Daniel Wennberg, Ryatatsu Yanagimoto, and Atsushi Yamamura have been working, in close collaboration with, Surya Ganguli, Marty Fejer and Amir Safavi-Naeini. All of us within the department of applied physics, at Stanford university and also in collaboration with Yoshihisa Yamamoto, over at NTT-PHI research labs. And I should acknowledge funding support, from the NSF by the Coherent Ising Machines, expedition in computing, also from NTT-PHI research labs, army research office, and ExxonMobil. That's it. Thanks very much.

>> From Boston, Massachusetts, it's the Cube, covering LiveWorx 18, brought to you by PTC. (light electronic music) >> Welcome back to Boston, everybody. This is the Cube, the leader in live tech coverage. We're covering day one of the LiveWorx conference that's hosted by PTC. I'm Dave Vellante with my cohost Stu Miniman. Professor Linda A. Hill is here. She's the Wallace Brett Donham Professor of Business Administration at the Harvard Business School. Professor Hill, welcome to the Cube. Thanks so much for coming on. >> Thank you for having me. >> So, innovation, lot of misconceptions about innovation and where it stems from. People think of Steve Jobs, well, the innovation comes from a single leader and a visionary who gets us in a headlock and makes it all happen. That's not really how innovation occurs, is it? >> No, it is not, actually. Most innovation is the result of a collaboration amongst people of different expertise and different points of view, and in fact, unless you have that diversity and some conflict, you rarely see innovation. >> So this is a topic that you've researched, so this isn't just an idea that you had. You've got proof and documentation of this, so tell us a little more about the work that you do at Harvard. >> So really over 10 years ago, I began to look at the connection between leadership and innovation, because it turns out that like a lot of organizations, the academy is quite siloed, so the people studying innovation were very separate from the ones who studied leadership, and we look at the connection between the two. When you look at that, what you discover is that leading innovation is actually different from leading change. Leading change is about coming up with a vision, communicating that vision, and inspiring people to want to fulfill that vision. Leading innovation is not about that. It's really more about how do you create a space in which people will be willing and able to do the kind of collaborative work required for innovation to happen? >> Sometimes I get confused, maybe you can help me, between invention and innovation. How should we think about those two dimensions? >> Innovation and invention. The way I think about it is an innovation is something that's both an invention, i.e. new, plus useful. So it can be an innovation or it can be creative, but unless it's useful and addresses an opportunity or a challenge that an organization faces, for me, that's not an innovation. So you need both, and that is really the paradox. How do you unleash people's talents and passions so you get the innovation or the invention or the new, and then how do you actually combine that, or harness all of those different ideas so that you get something that is useful, that actually solves a problem that the collective needs solved? >> So there's an outcome that involves changing something, adoption, as part of that innovation. >> For instance, one of the things that we're doing a lot right now is we're working with organizations, incumbents, I guess you'd call them, that have put together these innovation labs to create digital assets. And the problem is that those digital assets get created, they're new, if you will, but unless the core business will adopt them and use them, they get implemented, they're not going to be useful. So we're trying to understand, how do you take what gets created in those innovation labs, those assets, if you will, and make sure that the organization takes them in and scales them so that you can actually solve a business problem? >> Professor Hill, a fascinating topic I love digging into here. Because you see so many times, startups are often people that get frustrated inside a large company. I've worked for some very large companies, so which have had labs, or research division, and even when you carve aside time for innovation, you do programs on that, there's the corporate antibodies that fight against that. Maybe talk a little bit about that dynamic. Can large companies truly innovate? >> Yes, large companies can truly innovate. We do see it happening, it is not easy by any means, and I think part of the dilemma for why we don't see more innovation is actually our mindset about what leadership is about and who can innovate. So if I could combine a couple of things you asked, invention, often when we talk to people about what is innovation, they think about technology, and they think about new, and if I'm not a technologist and I'm not creative, then I can't play the game. But what we see in organizations, big ones that can innovate, is they don't separate out the innovators from the executors. They tell everybody, guess what, your job no matter who you are, of course you need to deal with making sure we get done what we said we'd deliver, but if we're going to delight our customers or we're ever going to really get them to be sticky with us, you also need to think about not just what should you be doing, but what could you be doing. In the literature, in the research, that's called how do you close an opportunity gap and not just a performance gap? In the organizations we look at that are innovative, that can innovate time and again, they have a very democratic notion: everybody has a role to play. So our work, Collective Genius, is called Collective Genius because what we saw in Pixar was the touchstone for that work, is that they believe everybody has a slice of genius. They're not equally big or whatever, but everybody has a contribution to make, and you need to use yours to come up with what's new and useful. A lot of that will be incremental, but some of it will be breakthrough. So I think what we see with these innovation labs and the startups, if you will, is that often people do go to start them up, of course they eventually have to grow their business, so a part of what I find myself doing now is helping startups that have to scale, figure out how to maintain that culture, those capabilities, that allowed them to be successful in the first place, and that's tough one for startups, right? >> Yeah, I think Pixar's only about a 1,500 person company and they all have creativity in wat they do. I'm wondering if there's some basic training that's missing. I studied engineering and I didn't get design training in my undergraduate studies. It wasn't until I was out in the workforce that I learned about that. What kind of mindset and training do you have to do to make sure the people are open to this? >> One of the things that I did related to this is about five years ago, I told our dean of Harvard Business School that I needed to join the board of an organization called Arts Center. I don't know if you were aware of Arts Center in Pasadena. It's the number one school of industrial design in the U.S., and people don't know about it 'cause I always laugh at them. The man who designed the Apple store is a graduate there. The man who designed Tesla car and et cetera, so they're not so good at it, but one of the things that we've all come to understand is design thinking, lean startup, these are all tools that can help you be better at innovation, but unless you create an environment around that, people are going to be willing to use those tools and make the missteps, the failures that might come with it, know how to collaborate together, even when they're a large organization, I mean it's easier when you're smaller. But unless you know how to do all that, those tools, the lean startup or digital or design thinking or whatever, ' cause I'm working with a lot of the people who do that, and deep respect for them, nothing gets done. In the end, we are human, we all need to know first off that it's worthwhile to take the risk to get done whatever it is you want to get done, so what's the purpose of the work, how's it going to change the world? The second thing is we need to share a set of values about learning because we have to understand, as you well know, you cannot plan your way to an innovation, you have to act your way. And with the startup, you act as fast as you can, right, so somebody will give you enough money before you run out of money. Same similar process you have to do in a large company, an incumbent, but of course it's more complicated. The other thing that makes it more complicated is companies are global, and the other part of it that makes it more complicated that I'm seeing like in personalized medicine: you need to build an ecosystem of different kinds, of nanotechnologists, biotechnologists, different expertise to come together. All of this, frankly, you don't learn any of it in school. I remember learning that you can't teach anyone how to lead. You actually have to help people learn how to lead themselves and technologists will frequently say to me, i don't know why, you're a leadership professor? Well, this is a technical problem. We just haven't figured out the platform right, and once we get it right, all will be. No, once you get it right, humans are still going to resist change and not know how to necessarily learn together to get this done. >> I wonder if, are there any speacial leadership skills we need for digital transformation? Really kind of the overarching theme of the show here, help connect the dots for us. >> So the leading change piece is about having a vision, communicating it, and inspiring people. What it really does turn out when we look at exceptional leaders of innovation, and all of us would agree that they've done wonderful things time and again, not just once, they understand that is collective. They spend time building a culture and capabilities that really will support people collaborating together. The first one they build is, how do we know how to create a marketplace of ideas through debate and discourse? Yeah, you can brainstorm, but eventually, we have to abrade and have conflict. They know how to have healthy debates in which people are taught terms of skills, basic stuff, not just listening and inquiring, but how to actively advocate in a constructive way for your point of view, these leaders have to learn how to amplify difference, whereas many leaders learn how to minimize it. And as the founder of Pixar once said, you can never have too many cooks in the kitchen. Many people believe you can. It's like today, you need as much talent as you can get. Your job as a leader, what are the skills you need to get those top cooks to be able to cook a meal together, not to reduce the amount of diversity. You got to be prepared for the healthy fight. >> You've pointed this out in some of your talks is that you've got to have that debate. >> Yes, you have to. >> That friction, to create innovation, but at the same time it has to be productive. I know it can be toxic to an organization, maybe talk about that a little. >> I think one of the challenges is what skills do people need to learn? One is, how do you deal with conflict when people are very talented and passionate? I think many people avoid conflict or don't know how to engage that constructively, just truly don't, and they avoid it. I find that many times organizations aren't doing what they need to do because the leadrr is uncomfortable. The other thing, and I'm going to stereotype horribly here, but I'm an introvert, that book quiet is wonderful, but one of the challenges you have if you're more introverted or if you're more technical and you tend to look at things from a technical point of view, in some ways is that you often find the people with that kind of, that's what drives them, there's a right answer, there's a rational answer we need to get through or get to, as opposed to understanding that really innovative ideas are often the combination of ideas that look like they're in conflict initially, and by definition, you need to have the naive eye and the expert working together to come up with that innovative solution, so for someone who's a technologist to think they should listen to someone who's naive about a technical problem, just the very basic mindset you have about who's going to have the idea. So that's a tricky one, it's a mindset, it's not even just a skill level, it's more, who do you think actually is valuable? Where is that slice that you need at this moment going to come from? It may not be from that expert, it may be from the one who had no point of view. I heard a story that I was collecting my data, and apparently, Steve Jobs went to see Ed Land. We're here in Boston over Polaroid, which is one of our most innovative companies, right, in the history. And he said, what do I need to learn from you? And what Land said to him is, whenever my scientist and technologist get stuck, I have some of the art students or the humanities students come in and spend time in the lab. They will ask the stupid question because they don't know it's stupid. The expert's not going to ask the stupid question, particularly the tech expert, not going to ask it. They will ask the question that gets the first principles. I think, but I wouldn't want to be held to this, the person who was telling me the story, that's partly how they came up with the instant camera. Some naive person said, why do I have to wait? Why can't I have it now? And of course, silly so-and-so, you don't know it takes this, that, and the others. Then someone else thought, why does she have to wait? I think it was really a she who asked the question, the person telling me this, and they came up with a different way. Who said it has to be done in a darkroom in that way? I think that there's certain things about our mindset independently of our skill, that get in the way of our actually hearing all the different voices we need to hear to get that abrasion going in the right way. >> Listening to those Columbo questions, you say, can sometime lead to an outcome that is radically different. There's a lot of conversation in our industry, the technology industry, about, we call it the cordially shock clock, the companies are on a cordially reporting mechanism or requirement from the SEC. A lot of complaints about that, but at the same time, it feels like at least in the tech business, that U.S. companies tend to be more innovative. But again, you hear a lot of complaints about, well, they can't think for the long term. Can you help us square that circle? >> It's funny, so one thing is you rarely ever get innovation without constraint. If you actually talk to people who are trying to innovate, there needs to be the boundaries around it in which they're doing the constraint. To be completely free rarely leads to, it is the constraint. Now we did do a study of boards to try to understand when is a board facilitating innovation and when is a board interfering with it? We interviewed CEOs and lead directors of a number of companies and wrote an article about that last year, and what we did find is many boards actually are seen as being inhibitors. They don't help management make the right decision. Then of course the board would say now management's the one that's too conservative, but this question about how the board, with guidance, and all of these issues have come up when you're looking at research analysts and who you play to, and I've been on corporate boards. One thing is that the CEO needs to know that the board is actually going to be supportive of his or her choices relative to how you communicate why you're making the choices you're making. So there is pressure, and I think it's real. We can't tell CEOs, no, you don't need to care about it, 'cause guess what, they do get in trouble if they don't. On the other hand, if they don't know how to make the argument for investing in terms of helping the company grow, so in the long run, innovation is not innovation for innovation's sake, it's to meet customer needs so you can grow, so you need to have a narrative that makes sense and be able to talk with people, the different stakeholders, about why you're making certain choices. I must say that I think that many times companies may be making the right choice for the long haul, and get punished in the short run, for sure that happens, but I also think that there are those companies that get a way with a lot of investment in the long haul, partly because they do, over time, deliver, and there is evidence that they're making the right choices or have built a culture where people think what they're saying might actually happen or be delivered. What's happening right now because of the convergence of industries, is I think a lot of CEOS, it's a frightening time, it is difficult to sustain success these days, because what you have to do is innovate at low cost. Going back to some other piece about boards, one of the things we've found is so many board members define innovation as being technology. Technology has a very important enabling role to play in otherwise, but they have such a narrow definition of it in a way that again, they create a culture to let the people in the innovation lab innovate, but not one where everybody understands that all of us, together, need to innovate in ways that will also prepare us to execute better. They don't see the whole culture transformation, digital transformation often requires cultural transformation for you to be able to get this stuff done, and that's what takes a long time. Takes a long time to get rid of your legacy systems and put in these new, or get that balance right, but what takes even longer is getting the culture to be receptive to using that new data capability they have and working in different ways and collaborating when they've been very siloed and they're paid to be very siloed. I think that unless you show, as a CEO, that you are actually putting all of those building blocks in place, and that's what you're about, you understand it's a transformation at that level, you're just talking to the analysts about, we're going to do x, and there's no evidence about your culture or anything else going on, how you're going to lead to attract and retain the kind of talent you need, no one's buying that, I think that that's the problem. There's not a whole story that they're telling about how this goes together and they're going to move forward on it. >> To your other point, is there data to suggest, can you quantify the relationship between diversity and innovation? >> There are some data about that, I don't have it. I find it's very funny, as you can see, I'm an African-American woman. My work is on leadership globalization and innovation. I do a lot of work on how you deliver global strategies. I often find when I'm working with senior teams, they'll ask me, would you help us with our inclusion effort? And I think it's partly because of who I am and diversity comes up in our work, and if you actually build the environments for talking about, they tend to be more inclusive about diversity of thought. Not demographic diversity, those can be separate as we well know because we know Silicon Valley is not a place where you see a lot of demographic diversity, but you might see diversity of thought. I haven't asked, it's interesting, I have had some invitations by governments, too. Japan, which has womenomics, which is a part of their policy If they need to get more women in the economy, frankly, otherwise they can't grow as an economy. It turns out that the innovation story is the business case that many businesses or business people find one that they can buy into, doesn't feel like you're doing it 'cause it's the right thing, or not that you shouldn't do the right thing, but helping them understand how you really, really make sure that the minority voice is heard, and I mean minority of thought, independent of demographic, but if you create an environment as a leader where you actually run your team so that people do feel they can speak up, as you all know. It's so often, I'll talk to people afterwards and they'll say, I didn't say what I really thought about those ideas because I didn't want to be punished or I didn't want to step in that person's territory. People are making decisions based on varying complete information everyone knows. What often happens is it gets escalated up. We had this one senior team complaining, everything is so slow here, a very big bank, not the one I'm on the board of, another very big bank we're working with. Everything's so slow, people won't do anything. So when we actually ask people, what's happening? Why aren't you making decisions? First off, decisions making rights are very fuzzy in this organization, except for at the very top, so what they say is all decisions, actually, they're made on the 34th floor. We escalate 'cause if you make a decision, they're going to turn it over anyway, so we've backed off, or we don't say what we think 'cause I don't want them to say what they think about my ideas 'cause we actually have very separate business units here. >> We might get shot. >> You might get shot. That's the reality that many people live in, so we're not surprised to see that not very many organizations can innovate time and again when we think about the reality of what our contexts are. The good news for us is that in part, millennials won't tolerate some of these environments in the same way, which is going to be a good thing. I think they're marvelous to work with, I'm not one of them obviously, but I think a lot of what they're requesting, the transparency, the understanding the connections between what they do and are they having impact, the desire to be developed and be learning, and wanting to be an organization they're not ashamed of but in fact they're very proud to be a part of what's happening there, I think that that requires businesses and leaders to behave differently. One of the businesses we studied, if the millennial wants to know who's on the front line, he or she is making a difference. They had to do finance differently to be able to show, to draw the cause and effect between what that person was doing every day and how it impacted the client's work. That ended up being a really interesting task. Or a supply chain leader, who really needed them to think very differently about supply chain so they could innovate. What he ended up doing is, instead of thinking about our customers being the pharmaceutical company, the CBS or the big hospital chain or whatever it is, think about the end customer. What would we have to do with supply chain to ensure that that end patient took his or her pill on time and got better? And when they shifted the whole meaning of the work to that individual patient in his or her home, he was able, over time, to get the whole supply chain group organization to understand, we're not doing what we need to do if we're really going to reduce diabetes in the world because the biggest problem we have is not when they go and get their medication, it's whether they actually use it properly when they're there. So when you switched it to that being the purpose of the work, the mindset that everyone had to have, that's what we're delivering on. Everyone said, oh, this is completely appropriate, we needed digital, we need different kind of data to know what's going on there. >> Don't get me started on human health. Professor Hill, for an introvert, you're quite a storyteller, and we appreciate you sharing your examples and your knowledge. Thanks so much for coming on the Cube. It was great to meet you. >> Been my pleasure, glad to know you, thank you. >> Keep it right there, everybody, Stu and I will be back right after this short break. You're watching the Cube from LiveWorx in Boston. We'll be right back. (light electronic music)

>> Live from San Francisco, it's theCUBE. Covering DockerCon '18. Brought to you by Docker and its ecosystem partners. >> Welcome back to theCUBE. We are live in San Francisco at DockerCon 2018. I'm Lisa Martin with John Troyer on a stunning day here in San Francisco. This event draws between 5,000 and 6,000 people in only its fifth year. They did a very good job during the general session this morning, John, of having some great female leaders on stage and we're very pleased to welcome another female leader to theCUBE for the first time. Alex Qin, you are the Director of Technology at Gakko. Welcome to theCUBE. >> Thank you, thank you. It's great to be here. >> So, you're speaking here at DockerCon 2018, I want to get to that in a second, but tell us a little bit about Gakko. What do you guys do? >> Um, yeah. So we're a global education design studio based in Tokyo and New York and what we do is we put on experimental education programs and build experimental education technology that aim to reclaim the magic of learning. So, we put on summer camps, we have coding classes, music classes and we build software for early learners. >> And by early learners what age group are you talking about? >> So ages three to five. What we build is beautiful story and art driven apps for kids ages three to five to be able to spend time more thoughtfully on tablets 'cause nowadays kids are always on tablets no matter what we do and so what we want to do is create a world that they can be in, in which parents feel like, this is a good place for my child to spend time. They're learning, it's artful, it's thoughtfully built. >> Great, well Alex you are also the founder of The Code Collective. >> The Code Cooperative, yes. >> The Code Cooperative, I'm sorry. How did you get started with that and can you tell us a little bit about that as well? >> Yes, so The Code Cooperative is my passion project and I started it in 2016, the day after the presidential elections actually, and it's an organization that teaches formerly incarcerated individuals computer literacy and coding, so that they can build websites and technical solutions to the problems they've identified in the criminal justice system. >> Some examples of that might be? >> A story I love to tell is from the pilot class. I had one student who was a 65 year old man and he'd been in prison for over 20 years and so at 65, he took our class and he learned HTML, CSS and JavaScript and built a website that aims to educate visitors about the legacy of slavery and Jim Crow in the criminal justice system today. Just like an interactive quiz. Yeah, that was really cool. It was called The Criminal Injustice System. >> Nice, nice. >> What were some of the drivers that really led you to go, you know what? We've got a huge opportunity here to take some of these people who have had made some different choices and really, sort of, rehabilitate them in a way that's gonna enable tech for good. What were some of those things that you just went, we've got to do this? >> That's a good question. Well, I read the book, The New Jim Crow, which you may have heard of. It's an incredible book that really details a lot of the problems that exist today within the U.S. criminal justice system and I thought to myself, I want to learn more about the justice system and contribute positively to justice system reform, but I don't know anything about it. So what I should do is work with people who have been through the system, learn from them and empower them to highlight the issues that they see within the justice system and that's something that I think is really important. When it comes to building technology, right now the gatekeepers of tech are kind of a homogenous group and we tend to build tech solutions for the entire world, but actually the people who are best equipped to solve problems are those who have experienced them and so that's why I decided to start The Code Cooperative. >> Nice. Alex, you're talking here, you've got an interesting titled session, I'll make sure I get it right, Shaving My Head Made Me a Better Programmer. If I can connect that to the rest of the DockerCon, maybe, I mean, Docker has been very good at their whole history about developer experience, making things easier for people and I think sometimes people don't realize not only when you make things easier, you actually can bring in new audiences. Kids, prisoners, right, are able to use today's technology where 30, 40 years ago they wouldn't have had access to it because it's easier, it's more powerful, it's more ubiquitous. But sometimes we get stuck in old tropes and so I'd love for you to kind of talk a little about your talk and kind of, what you're going to be talking about here at the show. >> Sure, yes. So, my talk is called Shaving My Head Made Me a Better Programmer and it's a little bit of a misleading title, but basically it's the story of my journey though the tech industry as a minority woman. So I studied computer science and I've been a software engineer for my entire career and yet, I've encountered a lot of challenges because of my gender, because of how I present to the world and when I shaved my head, a lot of those challenges kind of disappeared because I wasn't perceived as feminine anymore and so when I realized that tech isn't the meritocracy that I thought it was, I kind of started on this new quest to make tech as diverse and inclusive as possible so that people from all backgrounds, all genders can learn to code and write code happily and safely and it's just the story of how that happened and the lessons I've learned and some tips on how to make organizations more inclusive because that's the bulk of my work now. >> So you were a C.S. major in New York? >> Yes. >> So were you always interested in STEM as a kid or was it something that you got into when you were in college? What was that sort of age that you found it really exciting and said, no matter what, even if there's very few women here, I love this, I want to do this? >> That's a great question. So I am originally from France, actually. And when I was growing up there was really little computer science education in schools, but I really wanted to be an astronaut when I got to college so I joined the engineering program at my school and I'd never coded at that point, but one of the requirements was an intro to programming class in Python. So I took it and I fell in love with it immediately and I was like, I'm majoring in computer science, this is so cool, this is the coolest thing I've ever done and as I entered the computer science world I realized, oh, there's not that many women here and actually, I'm treated very differently. So, I fell in love with it and then because I love it so much I just kind of powered through. >> Your passion is very palpable, so at any point did you feel, sort of, out of place? Going, I'm one of the only females here, or did you say, I don't care, I like this. >> Yeah, it's both. I mean, you feel out of place when there's very few people who look like you in the room. Even if you don't want to feel out of place, even if you try to pretend that's not the case, you can't help but feel that and when I was starting out and throughout my career, people didn't necessarily want to work with me, didn't believe I was a good programmer, even though I was at the top of all my classes and so even though I tried to make the most out of my experience, I couldn't really escape the stigma attached to my gender in this field. >> Alex, we're at an interesting part of our culture now, I suppose, especially online. On one hand, social media has elevated a lot of folks' voices that would not have been heard otherwise because of gatekeepers. On the other hand, we have our current online discourse, which is kind of, not very pleasant sometimes. So I am interested both kind of how you're navigating that online and then maybe as a followup, then as you work with companies, how you're working with them and what you're telling them, but in terms of online, I love Twitter and yet it frustrates me. Facebook as well, et cetera. How do you navigate that online yourself? >> That's a great question. Honestly, I have been kind of retreating from social media. I haven't really experienced too many negative interactions on social media because I'm not really a big presence there. I did kind of have a really bad experience once during a Grace Hopper conference. I tweeted something during the Male Allied panel of like, 2015, or something and that got picked up by some GamerGate writers and then a lot of people started tweeting negative things at me, but that's kind of the extent of my negative experiences online. I do think that, as you say, social media has allowed for uplifting of voices that were previously unheard, has allowed for activism to organize. There's so many positive things that come from social media and also it has a really nefarious affect on people and I think that something needs to change in terms of how these companies build their software. It needs to be safer for all people and also needs to be built more ethically. Less trying to manipulate our psyches. >> That's, I think, super important. Luckily at least that's a conversation now, right Lisa? That at least Facebook, I think eventually as a society we'll, I hope, we'll get through this and figure this out, but I don't feel like we're particularly literate with social at this point. But I did want to ask about your work with companies. You said you do talk with some companies about diversity and things like that, is there any either signs that folks are getting it right or things that you start off with as you're working, if someone asks, how do we become a more diverse workforce? >> Yeah, that's a good question. I can't really point to any companies that, I say, are doing amazing. There are some companies where I know folks are very happy. Slack is one of them, thoughtbot is another one of them. I'll say Gakko, but a few tips I generally give organizations is that you need to work to understand the problem. Why is there a lack of diversity in tech? Why is your team not diverse? Then you need to measure your data. You can't make a positive change if you don't know how much you're changing, right? So gather diversity data on your team, not just in terms of who's there, but who's in a leadership role. Who gets promoted? Who gets fired? Who's a manager? And then you need to commit. That's, I think, the place where a lot of people struggle is there's a lot of candidates who fit this, kind of, homogenous image of what a programmer is and so it can be easy sometimes to be like, well we need to hire someone right now so let's just hire this person. But in order to actually make a change you need to commit and you need to say I'm not going to compromise on the goals that we've set. >> You're absolutely right, that commitment word is exactly what's needed to drive that accountability to hold organizations up to that. I was just at VMware a couple of weeks ago in Palo Alto at the Women Transforming Technology event and we had a whole day of all talking with females in tech, which I always loved to do and theCUBE is very passionate about supporting that. The cultural change is imperative. We talk about digital transformation at every event and there's the CIO that says, hey we have to change the culture here to transform digitally, but also to start moving those numbers from, what, less than 25% of tech roles are held by women. The culture has to change. It seems like you're in a position, potentially, to actually influence the culture at these companies that you talk to about opening their eyes to commit. Does that excite you from within? >> Yes, I do talk to a lot of organizations about this, but I think the work that I do that might actually tip the scale is, basically, the education programs that I run in New York. All of my classrooms reflect the diversity of New York, both in terms of student and teacher bodies. So all of my students learn in an environment that is extremely diverse. They learn from teachers who look like them and I wish I learned to code in that way. Another important thing we teach our students is how to code as an ethical endeavor. So we teach our students to measure the ethical ramifications of their decisions when they build software so that hopefully the technologists of tomorrow, the CTO's of tomorrow they build code in a way that is best for humanity. They build code with empathy. >> Goin' back to your day job. You're working with kids. We talked about getting through social media, cultural change. Its going to depend on the next generation. So Alex, are the kids alright? Are they gonna save us? >> The kids are pretty alright. I mean, so my classroom is basically coding meets social entrepreneurship so all of our kids build an app that solves a problem they've identified in their communities and these kids are just coming up with the most beautiful solutions, like, more brilliant than any adult that I've met. I feel good about the future. >> Well, it's key to get those different perspectives and when you were saying, they're having the opportunity to code and create apps that are relevant to them that's where you can really ignite that passion. >> Exactly, that's so important >> It is important because when you're passionate about something, and we saw that on stage today with a lot of the Docker folks and Microsoft and McKesson, when you're passionate about something and really making a change, you can feel it. So it's good to hear that we're going in the right direction. Also, we're in this age, you talked about ethics, where it's essential. Because technology, we see a lot of examples where tech is not used for good and there's world leaders getting some of the leaders of tech companies together saying, I'm challenging you, make tech for good because we're seeing too much of the negative right now. How does that influence, whether it's the breaches at Equifax, or, there was a breach recently at MyHeritage, the DNA testing companies, to Cambridge Analytica. How do you see the kids, the young kids respond to that, going, that's a really poor use of tech. Are they aware of that? >> I think some kids are and in our classroom we spend some time talking about, we have discussions about, ethics of software. So that's something that's very important to us. But largely, most classrooms in the United States, no, I mean computer science education is not a standard in most classrooms in the US. In New York state, only 1% of high schoolers actually have access to any kind of computer science education and so most kids, they might hear tid bits from the T.V. or social media or something, but they're not necessarily informed enough to make one, good decisions as consumers and two, good decisions as potential technologists. So that's something that we are trying to spread and I hope other folks are also trying to work on. >> Another thing that I think is shocking is when we were at the Women Transforming Technology event just a few weeks ago at VMware in Palo Alto, they just announced with Stanford, Stanford is investing 15 million dollars into their gender research. VMware and Stanford wanting to look at what are the barriers for women in tech and minorities in tech and starting to dissolve some of those barriers. One of the things they actually had in their press release announcing this big 15 million dollar investment from VMware and Stanford is a Mckinsey report that said 20%, sorry, enterprise organizations that have females in management positions, probably executive management positions, didn't specify positions, are 20% more profitable. You just think, the numbers are saying when you have more thought diversity, you're actually going to be a more profitable organization, but I think to your point earlier, Alex, there has to be a commitment and there has to be a group within an organization that stands accountable. >> Absolutely. >> So we are thankful for you. (Alex laughs) for donating some of your time today to tell us what you're doing, it's good to hear the next generation, John, I think they got our backs. >> Alright, that's good. >> And Alex, have a great time with your very provocative session this afternoon. >> Thank you. >> We thank you so much for your time and it's really cool to hear how you're using your passion for tech for good. >> Thank you so much, it was great to be here. >> We want to thank you for watching theCUBE. I'm Lisa Martin with John Troyer. From San Francisco at DockerCon 2018. Stick around, John and I will be right back with out next guest. (upbeat music)

>> Live from Orlando, Florida, it's theCUBE! Covering Cisco Live 2018. Brought to you by Cisco, NetApp, and theCUBE's ecosystem partners. (bubbly music) >> Hello everyone and welcome back to the live coverage, here with theCUBE, here in Orlando, Florida, for Cisco Live 2018. I'm John Furrier, my co-host Stu Miniman, for the next three days of wall-to-wall live coverage. We have the co-founders of the Network Collective here Eyvonne Sharp and Jordan Martin, thanks for joining us today, Network Collective. Sounds great, sounds like it's a collection of networks, so what's goin' on, what do you guys do? First let's talk about what you guys do, obviously you guys do a lot of podcasting, a lot of diggin' into the tech, what is Network Collective? >> Network Collective is a video podcast that Jordan and I started. We really felt like there was a need to build community around network engineers, and that really a lot of network engineers are very isolated in their job, there's only a couple people where they work, they know what they know, and they don't have a lot of peers. And so we see Network Collective as a way to bring network engineers together to learn about their craft, and also share with one another in a community that's more than a once-a-year conference like Cisco Live. >> That's awesome, I love the video podcasting, more than ever now the need for kind of peer review, conversations around learning because the world's shifting. In the keynote today the CEO of Cisco talked about the old way, and the new networks that are coming. We've been talking about no perimeters for years, but now security threats are real, gotta keep that strain solid, keep managing that, but also bring in a new kind of a cloud-hybrid, multi-cloud world, requires real skill adoption, new things. What're you guys seeing, what's your thoughts, and what's some of the things that you guys are exploring on your video podcast around these trends? >> You wanna take that Jordan? >> Sure. So, I think the rate of change of networking is faster than it's been in a very long time, so we've, we've had to, we've kinda not had a whole lotta churn in the things we've had to learn, I mean it's been complex and difficult, and there's been challenges in getting up to speed, but, with the transition to more developer-focused, and developer-centric model of deploying equipment, it is--and the integration of cloud into what is essentially our infrastructure. It's changing so much, that it's good to get together and have those conversations because it's very difficult to navigate this by yourself, it's, you know, it's a lot to learn. >> You know, I wanna push back little bit on that 'cause, you know, I've been in networking my whole career. When I used to, I used to speak at Interop, and I'd put down, you know, here's the rate of change, and here are the decades. And it's like, you know, okay 10 gig, here's where the standards are, here's where the first pieces are, it's gonna take years for us to deploy this. I don't disagree that change is happening overall faster, but how are people keeping up with it? Are the enterprises that the networking people work for allowing them to roll out some of these changes a little faster? So, give us a little insight as to what you're hearing from the community? >> I think, I mean technology, I mean, we've got Moore's law, right? I mean technology has always been changing rapidly, I think the thing that is different is the way network engineers need to interact with their environment. Five to 10 years ago, you could still operate in an environment where you still did a lot of static routing, for example. Now, with the cloud, with workloads moving around, there is no way to run a multi-cloud enterprise network without some serious dynamic routing chops, whether that's BGP, or EIGRP, or OSPF, or all of the above, and a lot of network engineers are still catching up with some of those technologies, they're used to being able to do things the way that they've always done them. And I think there needs to be a mind shift where we start thinking about things dynamically, in that, you know, an IP address may not live in the same geography, it may move from on-premises to the cloud to another cloud, and we have to be able to build networks that are resilient enough, and flexible enough, to be able to support that kind of mobility. >> Yeah, I love that Eyvonne. Right, you talked about the multi-cloud world. Jordan, a follow up question I have for you, how does the networking person look at things when there's a lot of the networking that are really outside of their control when you talk about really, the cloud world today. >> Sure, and before we jump there, I wanna say, the change that we're talking about though, is a bit different than what's happened. So, what we've seen traditionally would be speeds and feeds, but what's changing is the way we operate networks, and that hasn't changed a lot. Now, as for, you know, how do you view it when you don't-- well that's a challenge that everyone is facing. We see networking getting further and closer to the host. And, when we see networking inside of VMware, I mean this has been something that's been around for, you know, a while now, but, we're just getting comfortable with the idea of hypervisor, and now we've got, you know, we've got containers, and we've got networking in third party services that we don't necessarily have access to, we don't have full control over, and it's a completely different nomenclature we have to relearn all the terms because of course, no one reused the stuff we were familiar with, because this all started from a developer mindset. It all makes sense where it came from, but now we're catching up. And so it's, the challenge is not only understanding what needs to be done in all these different environments, but also understanding, just the terminology, and what is means. What is a VPC? Well VPC means something completely different to a networker that has never touched Amazon, than it does to somebody who has worked at Amazon completely there's overlapping terms and confusion around that and it's just a matter of, I think you need some broader coordination. There's been discussion about something like a full stack engineer, I think that's a pretty rare thing, I don't know how, how likely it is that you're gonna be expert level in all different disciplines, but you do need, you do need cross-team collaboration more than you have traditionally. We've had these silos, those no longer work in a multi-cloud world, it just doesn't, just doesn't work anymore. >> One of the things that came out with the keynote was, the networks next act was the main theme, as they talked about this new way, I mean, they use secure, intelligent platform, you know, for digital business, you know, level one marketing there, more complex than a few years ago and then the onslaught of new things coming, AI, augmented reality, machine learning, and I'd put blockchain in there, I thought they would put blockchain in the keynote to hype it up a bit, but, then they introduced the multi-cloud concept at that point. So in the keynote, multi-cloud didn't come up until the next act came up, so obviously that is a key part of what we're seeing, we saw Google clouds CEO Diane Greene come on. How are network engineers looking at the multi-cloud? 'Cause, I mean, how are they, toe in the water, are they puttin' the tow in the water? (chuckles) What is multi-cloud to them? Because, I mean, we talk about Kubernetes all the time, from an app standpoint, but, networks have been locked down for many, many years, you talked about some of the chops they need, what are those next chops for a network engineer when it comes to taking the road to multi-cloud? >> Sure, I mean I think if you're going to do any kind of multi-cloud interconnect, you've gotta know VGP. But at the same time, you need to understand some of those fundamental concepts that, the reason developers are pushing to the cloud, is not cost, although I've heard that a lot, that you know this cloud thing can't be cheaper, but it's really about enabling the business to move faster, and so we need to start thinking that way as network engineers more too. We have I think historically, our mentality, we've even trained our network engineers to go slow, to be very deliberate, to plan out your changes, to have these really complex change windows, and we need to start thinking differently, we need to think about how to make modular changes, and to be able to allow our workloads to move and shift in ways that don't provide a lot of risk, and I think that's a new way of thinking for networking engineers. >> Yeah. >> Well, we're sitting here in the DevNet zone, and that was one of the highlights in the keynote, talking that there are over 500 thousand developers now registered on this platform that they've built here. Bring us inside a little bit, you know, is it, what was it, John, DevNet sec? There's all of these acronyms as to, you know, how developers-- >> Yeah, NetApp was their big thing. >> how the network and the operations go together. What are you seeing, what's working, what's some of the challenges? >> I think this is a shift of necessity. As we see more problems solved in the network, we're adding complexity at that layer that hasn't been seen before, before it was routing, we just had to get traffic from one place to another, then we added security, so okay we have security, but we, we create these choke points in security where we can send all the traffic through this place and just like, we can use filtering, or some sort of identification there. Well then we start moving to cloud and we talk about dynamic workloads, and we talk about things that could just shift anywhere in the world, well now our choke point is gone, and so now we have to manage all the pieces, all the solutions, all the things we're putting into the network, but we've gotta manage it in a distributed way. And so that's where I think the automation's, why it's such a big push right now is because, we have to do it that way, there's no way to manually put these features in the network and be able to manage them at any type of scale without automating that process, and that's why, I think, we see the growth of DevNet, I mean, if you've been here the past few years, it's gone from a little thing to a much, much bigger thing, there's a lot of people looking at automation specifically, that 500 thousand number is, rather large. Really impressive that there's that many people looking at networks from a programmatic way. But in the meantime, I think that there's also, a bit of a divide here, 'cause I think that there's, a lot of people are looking this way, but I think there's, we talk about this on the show pretty often, there's really two types of networks. There's the networks at companies where, it really is, they see their network as a competitive advantage, and those places are definitely looking at automation, and they're looking at multi-cloud. But we also see another trend in networking, and that is to, I want some simple, push button, just put it out, get packets A to B, and I don't wanna mess with it, I don't want expensive engineers on staff, I don't want-- So I feel like the industry's almost coming to a divide. That we're gonna have two different types of networks, we're gonna have the network for the place that the just want packets going A to B, and they really don't want much, and the other side of that divide is gonna be very complex networks that have to be managed with automation. >> Talk about that other divide, it's between, I mean, I love that conversation, because, that almost kinda comes into like the notion of networks as a service. Because if you wanna have less expensive people there, but yet have the reliability, how do those companies grow and maintain the robust resiliency of these networks, and have the high performance, take advantage of the goodness, well what does it matter? I mean, how are they, how is the demographic of the network evolving, 'cause, either they're stunted for growth, or they have an enabler. How do you view that, how do you take that apart? >> I think we have to, we have to look at our business needs, and evaluate the technologies that we use appropriate to that. There are times for complexity, I think we've pushed, as Jordan very eloquently described, a lot of complexity down into the network, and we're working, I think, now, as the entire industry to maybe back some of that out. But one of the things that I hear a lot when we talk about automation and things like DevNet and developers is, I believe a lot of network engineers are afraid their jobs are going away, but if you look at what's going on, we have more connected devices than we've ever had before, and that's not gonna stop, and all of those connected devices need networks. And so really what's happened is we've reached a complexity inflection point, which means we have to have better tools, and I think that's really what we're talking about is, is how do we, instead of doing everything manually, how do we look at the network as a system, and manage it as a system, with tools to manage it that way. >> Your point about that jobs going away, I love that comment because, that's a sunk fallacy because, there's so much other stuff happening, talk about security, so the basic question, I mean first of all, guys your job's not goin' away! (laughs) Check! It's only a, well, kind of, you don't stay current, so it's all the learning issues, the progression for learning. But really it's the role of the network engineers and the people running the networks, I mean, I remember back in, the old way, the network guys were the top dogs, they were kickin' butt, takin' names, they ran the show, a lot was riding on the network. But as we go into this new dynamic environment, what are the roles of the network? Is it security? I mean, what are some of the things that people are pivoting to, or laddering up to from a roles standpoint that you see, in terms of a progression of new discovery, new skills. Is there a path, have you seen any patterns, for the growth of the person? >> I really think network engineers need to at least understand what the cloud is and why it exists. And they need to understand more about the applications and what they mean to the business. I think we have created a divide sometimes where, you know, my job is just to get packets from point A to B, and I don't really need to understand what we do as an organization, and I think that those days are going to be behind us, we need to understand, you know, what applications are critical, why do we need to build the systems the way that we need to build them and use that information from the business? So I think for network engineers, I think cloud security, understanding applications, and learning the business and being able to talk that language is what's gonna be most valuable to them in their career in the future. >> Yeah, we've heard the term many times, I'm a plumber! Well, I mean, implying that moving packets from A to B. It gets interesting with containers. Policy-based stuff has been known concept in networking, QOS, these are things that are well known, but when we start lookin' at the trends up the stack, we're seeing that kinda thing goin' on, service meshes for instance, they talk about services from a policy standpoint, up the stack. That's always been the challenge for the Ciscos over the past 20 years is, how to move up the stack, should they move up the stack, but I think now seems to be a good time. Your reaction guys, to that notion of moving up the stack while maintaining the purity and the goodness of good networking. (laughing) >> I think that's the big challenge right now, right? The more we mesh it all together and we don't, we don't really define the layers that we've traditionally used, the more challenging it is to have experts in that domain, because the domain just grows so incredibly large. And so there's gotta be a balance here, and I think we're trying to find that, I don't know that we've hit that yet, you know where, where we understand where networking fits into all these pieces, how far into the host, or how far into the application does networking go, we've seen certain applications not using the host TCP/IP stack, right, just to find some sort of performance benefit and it, to me that seems like we're pushing really far into this idea of, you know, well if we don't have standards and define places where these things exist, it's gonna be very much the wild, wild west for a while, until we figure out where everything's going to be. And so I think it just presents challenges and opportunities I don't know that we have the answer about how far it goes yet. >> Well let me ask you a question, a good point by the way, we agree, it's evolving, it's a moving train as they say. But as, people that might be watching that might be a Cisco customer or someone deploying a lot of Cisco networks and products in his portfolio, what's your advice to them, what're you hearing that's a good first three steps to take today? Obviously the show's goin' on here, multi-cloud is in center of the focus, this new network age is here for the CEO. What are some things that people can do now that are safe and good first steps to continue on the journey to whatever this evolves into. >> Well I think as you're building your network you need to think about modularization, you need to think about how to build it in small, manageable pieces, and, even if you're not ready to take the automation step today, you need to think about what that's gonna look like in the future, so, if you really want to automate your network you have to have consistency, consistent policy, consistent configuration across your environment, and it's never too late to start that, or too early to start that, right? And so you can think about, if I wanted to take these 10 sites and I wanted to manage them as one, how would I build it? And you can use that kind of mental framework to help guide the decisions you make, even if you're not ready to jump into a full scale automation from soup to nuts. And also I think, it's important to start playing around with automation technology, there are all kinds of tools to do that, and you can start in an are that's either dev, or QA, that's not gonna be production impacting, but you really need to get your, wrap your hands around some of the tools that exist to automate, and start playing with those. >> Stay where you're comfortable, get in, learn, get hands on. Jordan, your thoughts? >> Yeah, so, I was just over here like nodding my head furiously, 'cause everything she said, I 100% agree with. >> Ditto. (laughing) >> Yes, ditto, exactly. The only thing I would add is that we think about automation a lot in the method of config push. Right, the idea of configuring a device in an automated way, but that's not the only avenue for automation. Start by pulling information from your devices, it is really, really low risk way to start looking at your network programmatically, is to be able to go out to all of your switches, all of your routers, all of your networking devices and pull the same information and correlate that data and get yourself some information that's with a broader view. Does nothing to effect the change or state of your network, but you are now starting to look at your network that way. And I will reiterate Eyvonne's point, you cannot automate a network if it's not repeatable. If every design, every topology, every location is a special snowflake, you will never be able to automate anything because you're gonna have a hundred unique automation scripts to run a hundred unique sites. >> You'd be chasin' your tail big time. >> You'd be chasin' your tail, and so it is critical, if you're not in that state now, what you need to do is start looking how to modularize, and make repeatable config blocks in your network. >> Well guys, thanks for comin' on, Eyvonne, Jordan, thanks for comin' on, appreciate you taking the time. Final question for ya, I know it's day one, we got two more days of live coverage here, but, if you can kinda project, and in your minds eye see the development of the show, what's bubbling up as the most important story that's gonna come out of Cisco Live if you had to look at some early indications from the keynotes and some of the conversations in the hallway, what do you think is the biggest story this year for Cisco Live? >> I think for me personally, I wanna understand what Cisco's cloud strategy looks like, to know where they're going with the cloud and how they're going to help stitch together all the different services that we have. The clouds are becoming their own monolith, they each do things their own way, and still the network is what is stitching all those services together to provide access. And so I think it's important to understand that strategy, and where Cisco's goin'. >> Jordan, your thoughts? >> My, what I'm really looking for from the show this year is how Cisco is gonna make orchestration approachable. We've seen this process of automation where only the hardcore programmers could do it, then we got some tools. And these tools, as we watch as more of Cisco's product platforms start to integrate with each other I think the key piece for enterprise shops that don't have that type of resource on staff is what tools are they gonna give them to make this orchestration, between the way in and the enterprise campus, and into the cloud and in the data center, how do we tie all that together and make that to like a nice, seamless way to operate your network? >> Hey, what a great opportunity to have another podcast called under the hood, see what's goin' on, lot of chops needed, thanks for comin' on, give a quick plug for the address for your podcast, where do we find it, what's the site, Network Collective, obviously you guys are doing great things. Share the coordinates. >> Sure, you can find us at thenetworkcollective.com we usually use the hashtag: #NetworkCollective I'm on Twitter @SharpNetwork. Jordan, you wanna tell people where to find you? >> Sure, @bcjordo on Twitter, and obviously if you wanna interact with Network Collective, @NetCollectivePC on Twitter as well. >> Alright, thanks so much for the commentary, great to have a little shared, little podcast here, live on theCUBE, here in Orlando, I'm John Furrier, with Stu Miniman for our coverage at Cisco Live 2018, stay with us for more, we've got two more days of this, got day one just gettin' started, be right back after this short break. (bubbly music)

>> Narrator: It's theCUBE covering Sapphire Now 2017 brought to you by SAP Cloud Platform and HANA Enterprise Cloud. >> Welcome back everyone to our special Sapphire Now 2017 coverage in our Palo Alto Studios. We have folks on the ground in Orlando. It's the third day of Sapphire Now and we're bringing our friends and experts inside our new 4500 square foot studio where we're starting to get our action going and covering events anywhere they are from here. If we can't get there we'll do it from here in Palo Alto. Our next guest is Satyen Sangani, CEO of Alation. A hot start-up funded by Custom Adventures, Catalyst Data Collective, and I think Andreessen Horowitz is also an investor? >> Satyen: That's right. >> Satyen, welcome to the cube conversation here. >> Thank you for having me. >> So we are doing this special coverage, and I wanted to bring you in and discuss Sapphire Now as it relates to the context of the biggest wave hitting the industry, with waves are ones cloud. We've known that for a while. People surfing that one, then the data wave is coming fast, and I think this is a completely different animal in the sense of it's going to look different, but be just as big. Your business is in the data business. You help companies figure this out. Give us the update on, first take a minute talk about Alation, for the folks who aren't following you, what do you guys do, and then let's talk about data. >> Yeah. So for those of you that don't know about what Alation is, it's basically a data catalog. You know, if you think about all of the databases that exist in the enterprise, stuff on Prem, stuff in the cloud, all the BI tools like Tableau and MicroStrategy, and Business Objects. When you've got a lot of data that sits inside the enterprise today and a wide variety of legacy and modern tools, and what Alation does is, it creates a catalog, crawling all of those systems like Google crawls the web and effectively looks at all the logs inside of those systems, to understand how the data is interrelated and we create this data social graph, and it kind of looks >> John: It's a metadata catalog? >> We call you know, we don't use the word metadata because metadata is the word that people use when you know that's that's Johnny back in the corner office, Right? And people don't want to talk about metadata if you're a business person you think about metadata you're like, I don't, not my thing. >> So you guys are democratizing what data means to an organization? That's right. >> We just like to talk about context. We basically say, look in the same way that information, or in the same way when you're eating your food, you need, you know organic labeling to understand whether or not that's good or bad, we have on some level a provenance problem, a trust problem inside of data in the enterprise, and you need a layer of you know trust, and understanding in context. >> So you guys are a SAS, or you guys are a SAS solution, or are you a software subscription? >> We are both. Most of this is actually on Prem because most of the people that have the problem that Alation solves are very big complicated institutions, or institutions with a lot of data, or a lot of people trying to analyze it, but we do also have a SAS offering, and actually that's how we intersect with SAP Altiscale, and so we have a cloud base that's offering that we work with. >> Tell me about your relation SAP because you kind of backdoored in through an acquisition, quickly note that we'll get into the conversation. >> Yeah that's right, So Altiscale to big intersections, big data, and then they do big data in the cloud SAP acquired them last year and what we do is we provide a front-end capability for people to access that data in the cloud, so that as analysts want to analyze that data, as data governance folks want to manage that data, we provide them with a single catalog to do that. >> So talk about the dynamics in the industry because SAP clearly the big news there is the Leonardo, they're trying to create this framework, we just announced an alpha because everyone's got these names of dead creative geniuses, (Satyen laughs) We just ingest our Nostradamus products, Since they have Leonardo and, >> That's right. >> SAP's got Einstein, and IBM's got Watson, and Informatica has got Claire, so who thought maybe we just get our own version, but anyway, everyone's got some sort of like bot, or like AI program. >> Yep. >> I mean I get that, but the reality is, the trend is, they're trying to create a tool chest of platform re-platforming around tooling >> Satyen: Yeah. >> To make things easier. >> Satyen: Yeah. >> You have a lot of work in this area, through relation, trying to make things easier. >> Satyen: Yeah. >> And also they get the cloud, On-premise, HANA Enterprise Cloud, SAV cloud platform, meaning developers. So the convergence between developers, cloud, and data are happening. What's your take on that strategy? You think SAP's got a good move by going multi cloud, or should they, should be taking a different approach? >> Well I think they have to, I mean I think the economics in cloud, and the unmanageability, you know really human economics, and being able to have more and more being managed by third-party providers that are, you know, effectively like AWS, and how they skill, in the capability to manage at scale, and you just really can't compete if you're SAP, and you can't compete if your customers are buying, and assembling the toolkits On-premise, so they've got to go there, and I think every IT provider has to >> John: Got to go to the cloud you mean? >> They've got to go to the cloud, I think there's no question about it, you know I think that's at this point, a foregone conclusion in the world of enterprise IT. >> John: Yeah it's pretty obvious, I mean hybrid cloud is happening, that's really a gateway to multi-cloud, the submission is when I build Norton, a guest in latency multi-cloud issues there, but the reality is not every workloads gone there yet, a lot of analytics going on in the cloud. >> Satyen: Yeah. >> DevTest, okay check the box on DevTest >> Satyen: That's right. >> Analytics is all a ballgame right now, in terms of state of the art, your thoughts on the trends in how companies are using the cloud for analytics, and things that are challenges and opportunities. >> Yeah, I think there's, I think the analytics story in the cloud is a little bit earlier. I think that the transaction processing and the new applications, and the new architectures, and new integrations, certainly if you're going to build a new project, you're going to do that in the cloud, but I think the analytics in a stack, first of all there's like data gravity, right, you know there's a lot of gravity to that data, and moving it all into the cloud, and so if you're transaction processing, your behavioral apps are in the cloud, then it makes sense to keep the data in an AWS, or in the cloud. Conversely you know if it's not, then you're not going to take a whole bunch of data that sits on Prem and move it whole hog all the way to the cloud just because, right, that's super expensive, >> Yeah. >> You've got legacy. >> A lot of risks too and a lot of governance and a lot of compliance stuff as well. >> That's exactly right I mean if you're trying to comply with Basel II or GDPR, and you know you want to manage all that privacy information. How are you going to do that if you're going to move your data at the same time >> John: Yeah. >> And so it's a tough >> John: Great point. >> It's a tough move, I think from our perspective, and I think this is really important, you know we sort of say look, in a world where data is going to be on Prem, on the cloud, you know in BI tools, in databases and no SQL databases, on Hadoop, you're going to have data everywhere, and in that world where data is going to be in multiple locations and multiple technologies you got to figure out a way to manage. >> Yeah. I mean data sprawls all over the place, it's a big problem, oh and this oh and by the way that's a good thing, store it to your storage is getting cheaper and cheaper, data legs are popping out, but you have data links, for all you have data everywhere. >> Satyen: That's right. >> How are you looking at that problem as a start-up, and how a customer's dealing with that, and what is this a real issue, or is this still too early to talk about data sprawl? >> It's a real issue, I mean it, we liken it to the advent of the Internet in the time of traditional media, right, so you had you had traditional media, there were single sort of authoritative sources we all watched it may be CNN may be CBS we had the nightly news we had Newsweek, we got our information, also the Internet comes along, and anybody can blog about anything, right and so the cost of creating information is now this much lower anybody can create any reality anybody can store data anywhere, right, and so now you've got a world where, with tableau, with Hadoop, with redshift, you can build any stack you want to at any cost, and so now what do you do? Because everybody's creating their own thing, every Dev is doing their own thing, everybody's got new databases, new applications, you know software is eating the world right? >> And data it is eating software. >> And data is eating software, and so now you've got this problem where you're like look I got all this stuff, and I don't know I don't know what's fake news, what's real, what's alternative fact, what doesn't make any sense, and so you've got a signal and noise problem, and I think in that world you got to figure out how to get to truth, right, >> John: Yeah. And what's the answer to that in your mind, not that you have the answer, if you did, we'd be solving it better. >> Yeah. >> But I mean directionally where's the vector going in your mind? I try to talk to Paul Martino about this at bullpen capital he's a total analytics geek he doesn't think this big data can solve that yet but they started to see some science around trying to solve these problems with data. What's your vision on this? >> Satyen: Yeah you know so I believe that every I think that every developer is going to start building applications based on data I think that every business person is going to have an analytical role in their job because if they're not dealing with the world on the certainty, and they're not using all the evidence, at their disposable, they're not making the best decisions and obviously they're going to be more and more analysts and so you know at some level everybody is an analyst >> I wrote a post in 2008, my old blog was hosted on WordPress, before I started SilicionANGLE, data is the new developer kid. >> That's right. >> And I saw that early, and it was still not as clear to this now as obvious as least to us because we're in the middle, in this industry, but it's now part of the software fabric, it's like a library, like as developer you'd call a library of code software to come in and be part of your program >> Yeah >> Building blocks approach, Lego blocks, but now data as Lego blocks completely changes the game on things if you think of it that way. Where are we on that notion of you really using data as a development component, I mean it seems to be early, I don't, haven't seen any proof points, that says, well that company's actually using the data programmatically with software. >> Satyen: Yeah. well I mean look I think there's features in almost every software application whether it's you know 27% of the people clicked on this button into this particular thing, I mean that's a data based application right and so I think there is this notion that we talked a lot about, which is data literacy, right, and so that's kind of a weird thing, so what does that exactly mean? Well data is just information like a news article is information, and you got to decide whether it's good or it's bad, and whether you can come to a conclusion, or whether you can't, just as if you're using an API from a third-party developer you need documentation, you need context about that data, and people have to be intelligent about how they use it. >> And literacies also makes it, makes it addressable. >> That's right. >> If you have knowledge about data, at some point it's named and addressed at some point in a network. >> Satyen: Yeah. >> Especially Jada in motion, I mean data legs I get, data at rest, we start getting into data in motion, real-time data, every piece of data counts. Right? >> That's exactly right. And so now you've got to teach people about how to use this stuff you've got to give them the right data you got to make that discoverable you got to make that information usable you've got to get people to know who the experts are about the data, so they can ask questions, you know these are tougher problems, especially as you get more and more systems. >> All right, as a start up, you're a growing start-up, you guys are, are lean and mean, doing well. You have to go compete in this war. It's a lot of, you know a lot of big whales in there, I mean you got Oracle, SAP, IBM, they're all trying to transform, everybody is transforming all the incumbent winners, potential buyers of your company, or potentially you displacing this, as a young CEO, they you know eat their lunch, you have to go compete in a big game. How are you guys looking at that compass, I see your focus so I know a little bit about your plan, but take us through the mindset of a start-up CEO, that has to go into this world, you guys have to be good, I mean this is a big wave, see it's a big wave. >> Yeah. Nobody buys from a start-up unless you get, and a start-up could be even a company, less than a 100-200 people, I mean nobody's buying from a company unless there's a 10x return to value relative to the next best option, and so in that world how do you build 10x value? Well one you've got to have great technology, and then that's the start point, but the other thing is you've got to have deep focus on your customers, right, and so I think from our perspective, we build focus by just saying, look nobody understands data in your company, and by and large you've got to make money by understanding this data, as you do the digital transformation stuff, a big part of that is differentiating and making better products and optimizing based upon understanding your data because that helps you and your business make better decisions, >> John: Yeah. >> And so what we're going to do is help you understand that data better and faster than any other company can do. >> You really got to pick your shots, but what you're saying, if I hear you saying is as a start-up you got to hit the beachhead segment you want to own. >> Satyen: That's right. >> And own it. >> Satyen: That's exactly. >> No other decision, just get it, and then maybe get to a bigger scope later, and sequence around, and grow it that way. >> Satyen: You can't solve 10 problems >> Can't be groping for a beachhead if you don't know what you want, you're never going to get it. >> That's right. You can't solve 10 problems unless you solve one, right, and so you know I think we're at a phase where we've proven that we can scalably solved one, we've got customers like, you know Pfizer and Intuit and Citrix and Tesco and Tesla and eBay and Munich Reinsurance and so these are all you know amazing brands that are traditionally difficult to sell into, but you know I think from our perspective it's really about focus and just helping customers that are making that digital analytical transformation. Do it faster, and do it by enabling their people. >> But a lot going on this week for events, we had Informatica world this week, we got V-mon. We had Google I/O. We had Sapphire. It's a variety of other events going on, but I want to ask you kind of a more of a entrepreneurial industry question, which is, if we're going through the so-called digital transformation, that means a new modern era an old one movie transformed, yet I go to every event, and everyone's number one at something, that's like I was just at Informatica, they're number one in six squadrons. Michael Dell we're number in four every character, Mark Hurr at the press meeting said they're number one in all categories, Ross Perot think quote about you could be number one depends on how you slice the market, seems to be in play, my point is I kind of get a little bit, you know weirded out by that, but that is okay, you know I guess theCUBE's number one in overall live videos produced at an enterprise event, you know I, so we're number one at something, but my point is. >> Satyen: You really are. >> My point is, in a new transformation, what is the new scoreboard going to look like because a lot of things that you're talking about is horizontally integrated, there's new use cases developing, a new environment is coming online, so if someone wanted to actually try to keep score of who number one is and who's winning, besides customer wins, because that's clearly the one that you can point to and say hey they're winning customers, customer growth is good, outside of customer growth, what do you think will be the key requirements to get some sort of metric on who's really doing well these are the others, I mean we're not yet there with >> Yeah it's a tough problem, I mean you know used to be the world was that nobody gets fired for choosing choosing IBM. >> John: Yeah. >> Right, and I think that that brand credibility worked in a world where you could be conservative right, in this world I think, that looking for those measures, it is going to be really tough, and I think on some level that quest for looking for what is number one, or who is the best is actually the sort of fool's errand, and if that's what you're looking for, if you're looking for, you know what's the best answer for me based upon social signal, you know it's kind of like you know I'm going to go do the what the popular kids do in high school, I mean that could lead to you know a path, but it doesn't lead to the one that's going to actually get you satisfaction, and so on some level I think that customers, like you are the best signal, you know, always, >> John: Yeah, I mean it's hard, it's a rhetorical question, we ask it because, you know, we're trying to see not mystical with the path of fact called the fashion, what's fashionable. >> Satyen: Yeah. >> That's different. I mean talk about like really a cure metro, in the old days market share is one, actually IDC used a track who had market shares, and they would say based upon the number of shipments products, this is the market share winner, right? yeah that's pretty clean, I mean that's fairly clean, so just what it would be now? Number of instances, I mean it's so hard to figure out anyway, I digress. >> No, I think that's right, I mean I think I think it's really tough, that I think customers stories that, sort of map to your case. >> Yeah. It all comes back down to customer wins, how many customers you have was the >> Yeah and how much value they are getting out of your stuff. >> Yeah. That 10x value, and I think that's the multiplier minimum, if not more and with clouds and the scale is happening, you agree? >> Satyen: Yeah. >> It's going to get better. Okay thanks for coming on theCUBE. We have Satyen Sangani. CEO, co-founder of Alation, great start-up. Follow them on Twitter, these guys got some really good focus, learning about your data, because once you understand the data hygiene, you start think about ethics, and all the cool stuff happening with data. Thanks so much for coming on CUBE. More coverage, but Sapphire after the short break. (techno music)