When AI meets quantum mechanics

Bandwidth for Changelog is provided by Fastly. Learn more@fastly.com we move fast and fix things here at Changelog because of Rollbar. Check them out@rollbar.com and we're hosted on Linode Cloud Servers. Head to linode.com changelog this episode is brought to you by DigitalOcean, the simplest cloud platform out there and we're excited to share. They now offer dedicated virtual droplets and unlike standard droplets which use shared virtual CPU threads, their two performance plans general purp CPU optimized They have dedicated virtual CPU threads. This translates to higher performance and increased consistency during CPU intensive processes. So if you have build boxes, cicd, video encoding, machine learning, ad serving, game servers, databases, batch processing, data mining, application servers or active front end web servers that need to be full duty CPU all day every day, then check out Digital Ocean's dedicated virtual CPU droplets. Pricing is very competitive starting at 40 bucks a month. Learn more and get started for free with a 100 of credit at.co changelog again.co changelog. Welcome to Practical AI, a weekly podcast about making artificial intelligence practical, productive and accessible to everyone. This is where conversations around AI, machine learning and data science happen. Join the community and slack with us around various topics of the show@changelog.com community. Follow us on Twitter. We're @practicalai FM. And now onto the show.

Welcome to Practical AI. This is Daniel Whitenack. I'm a data scientist with SIL International and I'm joined here by Chris Benson, who is a Chief AI strategist at Lockheed Martin. How you doing, Chris?

Doing great, Daniel.

How's it going today?

It's going good. It's allergy and mowing season and you know, I finally got peer pressured enough into mowing my lawn last night. So, you know, feeling that a little bit, but, but all around good. Otherwise. Otherwise models are training and, and having fun. There you go. How about you?

I'm, I'm doing fine. Also, mowing the lawn. You just, just take a giant ziplock bag and jump into it and zip.

It up, you know, and go out.

There and push it around and you.

Know, try to try, try to avoid, try to avoid the pollen that way.

Yeah, exactly. So I'm really excited today. As our listeners know, my background is originally in computational physics and so I always love when we have guests that kind of overlap with that area. It kind of brings me back to my grad school days and today the topic that we're going to talk about is pretty exciting. So we're going to talk about Quantum computing and how that overlaps with machine learning and AI, how machine learning and AI are impacting quantum computing, and then some related things. So today we're joined by Marcus Edwards, who is a graduate student at the Institute for Quantum Computing at the University of Waterloo, and Dr. Shahini Ghosh, who's a professor at the Wilfrid Laurier University. Welcome. Thank you guys for joining us.

Thank you for having us.

Yeah, it would be great to hear just a little bit of a background from each of you, how you got into physics and quantum computing, how you got interested maybe in related things like AI and those sorts of things. So maybe Dr. Ghosh, if you want to start us out with that.

So I've been interested in physics and more generally science for a long time. And yes, I was one of those nerdy kids who loved Star Trek back when I was a kid.

There's nothing wrong with that.

Excellent. But other than the sci fi kind of astronauts, I was also inspired by real astronauts. When I was a kid in India. One of my heroes was Rakesh Sharma, who was the first Indian to go to space. So I always dreamed of following him into space. And that's not something that's happened as yet, but hopefully you never know, you have time left. Exactly. So in the meantime, I thought I'd do something that's almost as exciting, which is physics, of course. So that's what brought me into physics. And then when I was an undergrad, I was lucky because I got to do a summer research project on quantum physics. So that was my first real taste of this very bizarre world and I kind of liked it. And from there I went to grad school in the US at the University of New Mexico. And back then that was one of the first research groups in this brand new area called quantum Information science, which is basically the broad area that includes quantum computing, quantum communication, and everything else we hear about today. So I feel like I got in on the ground floor. It was exciting times, and I've seen the whole field grow and evolve to what it is today. So it's been a great, wonderful journey, kind of like a Star Trek exploration journey. So I think it's going well so.

Far, I gotta say. Any bio that can bring Star Trek into it as part of your bio, that works for me. Marcus, can you tell us a little bit about yourself?

I certainly can, yeah. Actually, my link to quantum physics and quantum computing really is Dr. Ghost. I attended Wilfrid Laurier University in my undergrad in a double major in computer science and physics. Yeah. So I Originally joined that program just because of the interest in sort of the fundamental problem of information science and computing and wanting to get into the physics of it in that sort of exploration. Quantum mechanics was sort of the most interesting facet and sort of felt like that's where people were asking the most fundamental and sort of groundbreaking questions. And so, yeah, so I started working with Dr. Ghost in my undergrad, doing a directed research study and getting involved that way. And yeah, it was actually Dr. Ghost who encouraged me to continue on with it. And that's why largely I'm at the Institute for Quantum Computing now doing my grad studies. And yeah, I guess I also am a bit of a technologist. So I'm currently a front end team lead at Delphix Capital Markets Inc. Where I'm doing full stack software development. And so having some of that more practical like technology experience and bringing that together with the quantum physics is really exciting to me. And that's sort of where quantum machine learning comes into it. And we'll go from there.

Awesome. Yeah, that's super exciting. And I'm definitely, I'm really excited to hear about your passion for kind of merging that practical side of software engineering with the quantum physics. I've really appreciated that in our previous conversations. So maybe if one of you could just, I mean, we're all the time on practical AI. We're talking about a lot of times like GPUs and other ways to accelerate computing. And a lot of our listeners might not be, they might have heard of quantum computing, but not really understand how it fits into the wider scheme of, you know, is it a way to accelerate, you know, regular computers? Is it, is it something different? So if one of you could just kind of describe in general what quantum computing is and how it fits into that scheme of accelerating computing, that would be awesome.

Sure. Dr. Ghost, why don't you take a shot at this first and I'll add anything.

So quantum computing, yes, does offer the promise of super fast speed up for certain types of problems. But this is not just yet another faster computer. So it's not just about how you read in the news all the time, oh, now we have yet another faster processor from intel or AMD or whatever. That's not what we're talking about. We're talking about an entirely different technology. So it's kind of the difference between, for example, transportation by horse and carriage versus transformation transportation by cars. It's not like you can just build better and better horses and make a car. You can't. So in that Sense, it's a completely different technology because it's harnessing different laws of physics than what we use to build current computers. And the laws that we're talking about are the laws of physics that govern the behavior of individual particles like electrons and photons and so on. And those tend to be rather peculiar laws. So one of the things probably a lot of people have heard of is this idea of quantum uncertainty, or they may have heard about the idea of superposition, where a particular particle can have two different properties at the same time. So in the language of computing, that translates into a bit or some piece of information. A quantum bit doesn't have to be just 0 or 1, but have a superposition of 0 and 1, which means it has a probability of being 0 and a probability of 1. So that may seem like that would lead to more uncertainty in computing, which is true, sure. But if you're smart about it, then you can actually harness this uncertainty to do actually, better computing and build new types of applications. So one of the very first such applications was to realize that uncertainty can lead to information security in the sense of encryption and hiding information. From there, we explored new types of algorithms for other kinds of applications, such as encryption. And not just encryption, but cryptography and mathematical tasks, such as factoring large numbers is another big example. Doing searches more efficiently. And all of these come from realizing that all these strange quantum properties essentially give us new math to work with. And when we have more laws of more rules to work with, then we can combine the rules in more clever ways. It's like taking chess, for example, the rules of chess, and then saying, hey, what if we could play 3D chess like in Star Trek? And then you can make a lot more moves and you can play a much more interesting game. So that's really broadly what quantum computing is all about.

Yeah. Thank you so much. So if I'm understanding right, I mean, there's kind of a basic set of operations and hardware that have powered, even if they're faster, computers over time have powered classical, normal sort of computers that people think of over time that are really built around maybe things like transistors or other things that have a certain state, like 1 or 0. So am I right in saying that in a quantum computer, there's not necessarily the idea of a transistor, but something that has, you know, maybe not just one or zero, but a certain number of states. And because you have more possibilities, there's sort of fundamentally new things that you can do that are a different kind of space of operations. Than what was enabled on the other hardware, Is that right?

That's exactly right. So quantum processor would involve gates that are not just flipping off a bit from zero to one or just multiplying or. And gates, which we are very familiar with in regular classical hardware processors. For quantum processors, we are allowed to build even more gates that we couldn't do before because as you correctly said, there are many more different types of potential manipulations you can do because you're not just restricted to two things, 0 and 1.

Yes. Can I just say that I think it's awesome that quantum information science is a field that lets you sort of go back to and design at the level of the comparative transistor. Like who's going to let you redesign the transistor in any other field, Right?

Yeah, it's kind of like going back to a golden age almost.

It is. Especially if you're really interested in sort of that technology focused research. Right. So yeah, So I love that comparison to the transistor and designing at that level and thinking about things in different ways. I also really liked that comparison with 3D chess actually, because, yeah, there are models of quantum information science that are being developed and some that have been developed quite to quite a far extent and experimentally tested, experimentally demonstrated. And these models are enabling us to do computation despite not fully understanding the underlying physics of what's going on necessarily. I don't necessarily think anyone truly understands quantum quantum mechanics, but sort of, which is kind of fun how it fits with the 3D chess analogy. We can model quantum mechanical interactions and many body physics using high dimensional vector spaces and tensor mathematics, et cetera, which ends up leading us to the fact that actually quantum mechanics and quantum computing has a lot of analogs that fit well with machine learning and other fields that deal with high dimensional mathematics.

So I'm kind of wondering, we're always hearing in the news about quantum computer, quantum computing, quantum computers, but I don't know that that in my own mind I understand what the current state of kind of practical quantum computers, you know, you know, just like I might, you know, work on a traditional computer, a classical computer. Where, where are they at this point? Is that, is this something that we're expecting to be available anytime soon or are people going to have access to them? And if not, what is the roadmap to get there?

Sure. So I suspect that you may know this, but in 2010, Lockheed Martin actually became the first customer of one of the first companies providing commercially available quantum computing devices. Now this is the company I'M talking about is D Wave. And they don't provide quantum computers per se, not universal quantum computing, but computational devices that make use of quantum physics for sure. And you know, that was really exciting because Lockheed Martin was actually able to demonstrate one of the first practical uses of what's called a quantum annealing machine, which is what D wave provides, debugging a chunk of 30 year old code from an F16 aircraft. It was just a cool story and I thought it was a cool connection. So that's one example. There are these sort of almost application specific quantum devices that are actually now available, though there aren't too many in the world. And we also see other companies like IBM, Google, intel, all working on their own quantum computing projects. Microsoft actually has one too. And these are all sort of at varying levels and focusing on different technologies because there are many different formulations and approaches to implementing quantum computing. But one of the most notable ones, I think just because of how far they've come and how well they're doing with marketing and getting researchers on board, is IBM. IBM announced this year their system called IBM Quantum One. And what it is is a 20 qubit quantum computer. And it is that sort of universal quantum computer which application specific devices like D waves are not. And that's exciting because it's sort of commercially available. It's also available to research. Researchers like myself use it for just asking interesting questions about physics and seeing what actually happens and if it matches our expectations. It's kind of like a lab that you can access through the cloud at school. And yeah, now these computers, though you may suspect aren't changing the world yet. These are sort of, at this point sort of toy machines. And they're really expensive toys. They're several million dollars of sort of parts and work going into each one. But they're still at the point where even if they have a large number of qubits, like 20, that's not really enough to get us to the point where we're doing any sort of large scale optimization problems or really enhancing machine learning yet largely due to just the challenges in engineering that come along with it. Once you're trying to maintain a large quantum system, that becomes very difficult.

So kind of as a follow up to that, I'm going to ask both of you for an answer. If you'll put on your super prediction hat and magically look into your crystal ball. Do you think there's a point in, in the future here where quantum computers become as ubiquitous as our classical computers are? Or will they? Do you Think they're always going to be specialized and if so, just, you know, pulling a number out of the air. How long do you think we are from that?

Okay, I'll go first and jump in. Although of course it's very dangerous to ever make predictions about technology because we never ever get it right. That's the only prediction I can make for certain that I will be wrong.

Absolutely.

But that being said, currently I don't see any evidence that we are going to have desktops or laptops that are quantum computer based, because for one thing, we don't need them. So it's a bit of an overkill to use a quantum computer to do emails, for example. So you're never ever going to need that. That's not what quantum computing technology is all about.

So I think, I don't know, I.

Think I might need one to run all my Chrome tabs.

Quantum Chrome, yeah.

I'd say in the near future there's going to be definitely specific applications, as I said, perhaps either encryption or in, you know, things like what we call quantum simulation. So using a quantum computer to try to simulate other quantum systems, like small molecules perhaps that could be used for drug development and things like this. Those will happen in the small scale. In the relatively near term, maybe we're talking about the next decade or so, but I think in the longer term, and even now, I think what we're doing now is envisioning more of a hybrid system where there will be perhaps a front end which is familiar to users today, which looks no different than what you're using now. And then in the back end, maybe there'll be some layers of either quantum communication happening or some kind of quantum processing happening that you may not even see. And as Mark has described, our first access right now to any kind of quantum device, which is the IBM device, is through the cloud. So that might become the way forward where most of this will be cloud based, where we don't have it in our homes all the time necessarily, but it'll be there in the background. So I guess I'll leave it at that.

Yeah, not many people have like a TPU pod from Google. It's sitting in their closet to run their neural nets either. So it seems like a similar model.

I'm just bumming. Apparently I'm not going to have a MacBook Pro quantum version anytime soon here.

Yes, well, it's unlikely. I will say though, that one thing that really excites me is bringing this sort of cooperation closer together between classical and quantum computing and sort of optimizing that as much as possible. While it's certainly less likely that we'll ever have a quantum processing unit or something alongside our CPU in our laptops, I don't think it's impossible. My prediction would be that, of course, the first access that sort of becomes, quote, unquote, ubiquitous, if it gets that far to quantum computing, would be through the cloud. And so that would be, yeah, similar to what we have now that IBM is providing. Other companies are starting to look at that as well. There's Rigetti, Google and Xanadu actually all have their own sort of software portals now that they intend to continue to market and develop and optimize for quantum computing and enabling that communication. Now, of course, communicating with a quantum process is interesting because it becomes a bottleneck when you sort of wrap a quantum process in classical processes on either end. However, having a completely quantum computer to me would be probably the most useless computer that exists. Just because as sort of humans, we expect to have interpretable information coming out at us and going into the computer from us. And so I think that we will always have that sort of classical interface at least. And where quantum will actually be useful is in, as Dr. Goh said, the back end. Whether that's a chip or a remote database somewhere that's been implemented using quantum physics or a remote processor doing particular pieces of hybrid algorithms. And some of the most exciting hybrid algorithms are in machine learning. And there's work being done here in Waterloo, at IQC and at the Perimeter Institute on creating these hybrid algorithms and optimizing them for cooperation between CPUs and QPU's. That's really exciting stuff. Foreign.

This episode is brought to you by Discover Bot. Learn everything there is to know about bots at Discover Bot. Practical AI. Discover Bot was built by Amazon Registry Services as an online community for bot creators and makers of all skill levels to learn from one another, to share stories. And they regularly publish guides and resources to answer questions like how to set up payments to your bot, how to stop shopping cart abandonment, what KPIs are worth measuring, how to write an engaging chatbot dialogue. You can even register bot domains there. Learn more and explore this huge library of bot resources at Discover Bot. PracticalAI. Again, Discover Bot PracticalAI.

So I really appreciate the, the explanation that both of you gave on the front of what quantum computing is and kind of the current state of it. Maybe some things that are coming in the future. One thing that I. One question that I have in my mind and I played a little bit around with Systems like Rigetti and others. But I'm wondering from your perspective, could you just describe what does it look like to kind of program a quantum computer? Like do I just pull up a jupyter notebook and use Python to say pandas read from quantum computer something. What's the interface currently and how do you envision, what type of different thinking and different sorts of practical operations are you dealing with when you're programming a quantum computer?

Now this is a very, very fun question and one that's very exciting because again, when it comes to quantum information science, you're free to sort of reimagine and rethink every layer of the stack. So for example, what sort of language do we use to program a quantum computer at the low level? There are a few sort of different takes at this, but it's all in the very early stages and almost all open source so that anyone who is interested can actually start learning how people are thinking about implementing assembly languages for quantum computers. IBM has open qasm, they call it, it's quantum assembly language. It's open sourced. So for example, I have the source on my laptop here on my Mac and I'm able to play with that a lot. Now, sort of an interesting question is do we want our language, even our low level language, our quantum assembly language, to be opinionated and sort of limit the scope of what a quantum physicist can describe? Because quantum physics is very complex and the gate model is what's typically used to describe quantum computing processes. But it's certainly not the only model for quantum physics and even quantum computing, there are plenty of different types of devices. Even so, you might say we want to use a continuous variable quantum computer or an adiabatic quantum computer. And if we want to do that, because for some reason there's some appeal to practically implement one of those in the future. Maybe IBM's QASM is no longer relevant in that case because it's was designed to express its particular type of quantum computing. So yeah, so there's sort of a bit of a emerging open source community around quantum computing, which is very interesting. It's largely centered around IBM and what they're doing, but there's alternatives like what Xanadu is offering, which is around a different type of quantum computing called continuous variable quantum computing. And I've been a contributor to Xanadu's Strawberry Fields library, which is a Python library, for about five months and it's trying to define basically the tensorflow of quantum computing, which is a Python library that, yes, allows you to Define these sort of gate level operations and express quantum computing processes in that way, but also provide these more abstracted tools, sort of like you'd expect from a package like tensorflow. So if you want to implement a machine learning process that uses quantum physics in the background, you may use their package Strawberry Fields or Penny Lane to actually do that in a more sort of expressive way. And it's just really fun seeing all this stuff sort of getting built and imagined for the first time. And it's an open area of research, honestly, to define these sort of languages well, and work is going on in that area at IQC and in companies like Xanadu, and it's really cool.

I appreciate that, Markus. That's a great explanation. Dr. Ghosh, how do you see you have the existing kind of AI and machine learning community that's out there and you have this quantity, these quantum communities that are in development. And you know, how, how are those communities similar and different? How do they look at each other and how, how might they go about collaborating to where, you know, this idea that Marcus mentioned about kind of the, the tensorflow of quantum computing, how does that come into being where it's actually utilized in the community?

Yeah, that's a really exciting question to explore right now. And so this whole area of quantum machine learning, as we call it now, is rapidly growing. And as Marcus mentioned, Waterloo is a big hub for this kind of work, as is Toronto, of course, in AI in general. And so I think both quantum can benefit from AI and machine learning, and machine learning can benefit from quantum too. So what I mean is that there's, you know, if you look at the ma, the level of the mathematical structures of quantum theory and the mathematical structures of machine learning, there's a lot in common. So one thing that people have been exploring is given that we don't actually have existing large scale quantum computers today, we still have to rely on our current regular computers to be modeling and analyzing quantum systems. And what happens is that it's a very challenging problem because every time you add one more quantum bit to your system to try to model it, you double the computational space that you need to simulate. So that's a huge challenge and which prevents us from, in fact, for example, even modeling molecules of a few hundred atoms, for example, is where we get stuck. Even the world's best supercomputers can't handle it. However, if you look at the best ways to map that kind of information into classical computers, turns out the sort of mathematical frameworks like, you know, the tensor network Structure and so on that you also use for machine. And the neural network, the tensor network from quantum maps onto the neural network structure that's being used right now for machine learning. So there's a lot of work in trying to explore how can we efficiently explore quantum physics using the same kind of structures and approaches that are being used in machine learning. And there have been some initial successes for doing things like, for example, looking at the magnetic properties of different materials and so on. So those are very exciting because it means that there are actual benefits to be had from using the mathematics of machine learning to also analyze quantum theory. But it can also go the other way because we can also think about what happens when we do have working quantum computers at a scale large enough to do something interesting. The question then becomes, can we take some of the machine learning algorithms that are existing today and build quantum versions of those algorithms that are much more efficient? So what I mean by that, for example, and this is not a machine learning example, but one of the first math problems that was shown to be much, much better if you run on a quantum computer, is this idea of factoring a large number. And this is very useful, of course, because this is, in fact, what would enable us to hack into current encryption like rsa. So what we know is that there is a quantum version of factoring that can run much, much faster if we had a quantum computer. So then the question is, are there quantum versions of current machine learning algorithms that would learn much faster once we have a quantum computer? And there's a lot of work on that end. And if that, and as we build more of those and have them ready to go, and at the same time develop a good software and language that we can use to program future quantum computers, then we might be able to do all these exciting new kinds of problems that we can't do today. And that's what we call quantum advantage. So that's where I see that field.

So I guess I'm curious, from a very practical standpoint, you know, within, with deep learning now being dominated by the linear algebra and derivatives, you know, that we're always taking as we're, as we're training models, is, is that going to be superseded by different quantum techniques? Or, you know, would you. In other words, would the current math and quantum be somehow working together, or are you essentially going to, to replace the current mathematics with a quantum variant of that to get, you know, better performance? Where you want to go on, on.

That quantum mathematics is essentially linear algebra, but in a particular space which we call Hilbert space, which essentially means that you have complex numbers and vectors and there are certain properties of the space. So it's basically a. Yeah, as I said, a larger set of mathematical rules that we can use. So as I. And again, if you have more rules to a game, then you can play, you can figure out different ways to win the game. So that's really what I mean by using quantum math to do. I mean, the task remains the same, but the way you solve the task changes because you're allowed to use these different rules.

Gotcha.

Yes. And one of the interesting things that we get to do as researchers is actually recast problems that are currently being solved by AI and solved by classical algorithms into sort of the Hilbert space and into quantum problems. So if we can do this recasting of the problem and come up with then a quantum solution, it's often very interesting to see what kind of advantage that quantum solution provides over the classical one. And so I think we already see some companies actually providing services, not too many, but there's one qbit, for example, which does actually provide sort of a consulting service. And. And as a part of that, recasts problems into quantum problems and attempts to find solutions to that. So I imagine if quantum computing explodes and everything's great, that more people would start to do that type of work. And one thing to mention too though, is that not only is it going to be advantageous to recast AI problems into quantum AI problems, but classical AI also is helping us to understand sort of what's fundamentally different and interesting about quantum physics. Because if we can sort of go backwards and take a quantum problem and turn it into an AI problem and it's sort of solved and we can do it today, then maybe that's not as interesting a problem to physicists anymore. Right. And that, that does happen. And actually one interesting phenomena is that as sort of quantum physicists come up with more interesting things about quantum computing and what could happen. Classical sort of computational scientists and data scientists and stuff are coming up with sort of analogs or solutions and getting a little bit better in order to keep up with quantum in some ways. And so I think there will always be a collaboration between the two and we'll have classical and quantum computing sort of helping each other out as we go forward.

Yeah, it's interesting to kind of, I don't know, come full circle on this. So when I was in grad school, I was working on computational physics and Dr. Ghost, as you said, a lot of these kind of just brute force techniques will get you to maybe modeling a hundred or a few hundred atoms or molecules. And I remember at the time when I was in school, it was kind of the first time I had seen kind of at the end of my grad school, people started to apply machine learning techniques to figure out the energy functionals and things that we were trying to figure out just from scratch by writing good equations and kind of instantly outperformed everything that we were doing. And it was kind of a shock to, to all of us, but it just kind of illustrates, as you were saying, Markus, the power that you can achieve in certain cases by reimagining a problem as an AI or as a machine learning problem. I'm curious. So I was kind of just browsing around as you were talking on the Xanadu website and a couple others, and I see certain phrases like machine learning toolbox for quantum computing powered by TensorFlow and other things. So is that more on the side of say, using TensorFlow, using AI to kind of learn more about quantum computing? Or are those things more on the side of kind of doing, creating a quantum computing module for TensorFlow?

Ah, yes, okay. So yeah, one thing that's being done by Xanadu and that I'm playing around with too in my research is using tools that exist and are inherently classical, of course, like TensorFlow and TPUs and GPUs and you know, looking at all the different classical technology that exists and seeing what can be done to sort of emulate, simulate or predict the outcomes of quantum computations. And so yes, so what Xanadu is doing is actually using TensorFlow as a back end. So when you are using PennyLane or Strawberry Fields to sort of express a quantum experiment and you choose to use the TensorFlow backend, what it's doing is using TensorFlow to actually simulate that quantum process.

So it's kind of helping you learn about how the experiment might go. Is that a way to put it or.

Sure, yeah. I mean, it's implementing the mathematics of it, of the model. So if you have TensorFlow as your backend, then it's possible that you could actually express gate level quantum computations and get them sort of compiled down to, well, not compiled, but sort of interpreted and changed into tensorflow code, which is Python. That's like creating matrices, doing matrix multiplication, defining a tensor network and then that can get sent off to Google, who then implements that on a TPU and gives you your results through the Google Cloud, or I think they're calling it Anthos now or whatever. But you know, yeah, so it's using it as a, as a simulator backend to sort of help physicists learn and inform themselves and also demonstrate theories like there is certainly power in the classical infrastructure that exists. What's really cool is how the machine learning infrastructure is getting used now to sort of simulate many body physics. And TensorFlow is just one example.

This episode is brought to you by strong DM. Strong DM makes it easy for DevOps to enforce the controls infosec teams require manage access to any database server in any environment. And in this segment we're talking to Jim Mordko, VP of Engineering at Hearst. He's sharing how they're using strong DM within their team of 90 plus engineers. It now takes them just 60 seconds to offboard a team member from a data source.

We have an engineering team of somewhere in the area of 80 or 90 engineers. Because we've got so many services and many databases and so many developers, we need a reasonable way to manage access to them. It was, it was a somewhat painful and labor intensive process. Our DevOps team would literally have to manage every one of the permissions for everybody who wanted access. So StrongDM has been a real godsend in that area for us. Requests for access to specific databases were pretty much manual. Now we've adopted StrongDM. It's something that you don't even know is there. Once it's installed, it just works. It's very simple. We've set up a multitude of data sources so that, that when somebody's onboarded, we just give them access to StrongDM. It's pretty simple. Our DevOps team, they have a very minimal effort required to enable each data source to be connected to StrongDM. And then installing the client software is very, very simple and straightforward. You can use whatever client you want to to talk to the database. So there's really no training necessary.

All right. If your team can benefit from nearly instant onboarding and offboarding, that's fully SoC2 compliant. Head to StrongDM.com to learn more and request a free demo. Again, StrongDM.com.

So earlier I know that you had mentioned the quantum emulation project that you were working on. Could you describe that a bit more?

Sure. So this is one of my research projects with Dr. Ghost through the IQC. So the quantum emulation project is sort of the umbrella for all of our research into quantum emulation. And this is sort of the research that has led me to get involved with Xanadu, contributing to their library for Python. It's also sort of encapsulates my Thinking about how can hardware potentially emulate quantum physics? And. Yeah, so the project itself is multifaceted.

As you explain that, could you also kind of define what quantum emulation would be?

Oh, sure, yeah. So this is a. Yeah, interesting word choices. Emulation versus simulation. Right. So simulation and emulation, I'm thinking of.

Like the Nintendo games I play on my Raspberry PI.

Right? Yeah, okay. It's not that, but related. So when I say emulation, what I'm referring to is a physical system that more closely behaves like a quantum physical system rather than a software simulation. So it's very kind of a nitpicky sort of difference between simulation and emulation. But when I'm talking about an emulator, it would be something implemented in hardware that's designed sort of from physics to behave more like quantum physics. Whereas if I was talking about a simulation, I would probably just be talking about something I wrote in Python.

Okay, and the pieces of hardware in this sort of emulation, what are those. Are those pieces of classical hardware that are kind of bolted together to. Along with certain software elements to do this emulation? Or what are. What are those pieces? Are those, like, you know, nodes in the cloud or what are. What are we talking about there?

Yeah, okay, so when I'm talking about a quantum emulator, there's sort of actually a bunch of research around emulation of quantum computing. There have been papers about doing this with FPGAs, there's been a paper about how rough and hard it is to do this with analog computing elements like op amps and such. But, yeah, so I'm sort of looking at the whole. The whole spread of options, taking a look at what can be done with FPGAs, what can be done with analog, how can this be orchestrated efficiently to work with sort of the machine learning tools that exist, which are mostly accessible through the cloud? And. Yeah, so one of the sort of questions that I want to answer with the research is what can be done to take this to the next level? We have stuff like what Xanadu is producing, which is awesome expressive Python. They're also working on a specific type of backend hardware, which is continuous variable quantum computing. We have IBM and we have all these cloud tools as well. Can we bring them together and what would be missing to make this a viable system, and is there something I can do to add that secret sauce or whatever is missing?

Dr. Ghosh, what does success look like from your perspective? Where should the project be heading and what kinds of things are you hoping to see come out of the project.

You mean Marcus's project or.

Yeah, I'm sorry, I was thinking that both of you were participating in that. So I can turn it to either one of y', all, whichever one would like to like to take a stab at it.

Well, so what Markus is working on is part of a broader research program that's in my team, which is why I wanted to clarify that you're talking about our research in general. So, you know, research success is basically exploring something that nobody has ever done before and figuring out whether or not you actually build something new that is useful or not. You learn something. That's the definition of research. So in that sense, you know, this is a great area to be in because nobody knows anything about anything. So that's really how we approach it, as in, let's. It's almost like playing. Right here's this kind of unexplored territory, kind of like a new planet in Star Trek.

There you go.

You did make it.

And so we, you know, we beamed out of this planet and we start exploring. And part of that is to say, well, given our current tools, which is that we have some access to kind of toy quantum computers as of now, and we have this huge, powerful new tool set offered through, you know, AI and machine learning, and we have our current hardware, which are, you know, supercomputers and, you know, whatever is the latest processor today, Given what we have today, what's the best we can do and what can we learn about what can and cannot be done? So, for example, in Marcus project, which, if it comes down to it, is going to be limited by the fact that we don't have a real quantum computer. So we're going to try to, as he said, emulate it using what hardware we can build now, for example. And the reason to do that is not because we expect to somehow replace an actual quantum computer, but it is to explore what is the actual power of a full quantum computer. Right. Where is that transition happening? What is that special fuel that we will not be able to emulate? So in this sense, almost success is not succeeding at a certain task, as in here is where we would really need a quantum computer. And so that's what we should focus on. So that's kind of how we approach this project and research in general. I don't know if that is very clear, but research, by definition is just a lot of going down blind alleys and failing a lot and then finding some unexpected discovery and then taking it from there.

Yeah, I have maybe a Practical question from my perspective, as kind of being previously in academia and also now viewing, like you were saying, Dr. Ghost, the kind of the powerful tools that are available right now in TensorFlow and AI. And I'm just thinking like back to when I was in grad school, I think a lot of that in some ways would be overwhelming for me to take in, in addition to quantum physics and all of the other things. So I was wondering, from Marcus, maybe your perspective, or Dr. Ghost, from your perspective as a team and a research group in general, how have you found the process of looking at the problem set that's in front of you, deciding to use AI and TensorFlow and those sorts of things, and figuring out how to apply TensorFlow to your research problem? Do you have any tips for those out there that are maybe doing some sort of research, whether that's an R and D in industry or in academia or elsewhere, and they see the power of what maybe they could achieve with AI, but it seems overwhelming for them. Do you have any tips as far as them getting into this and starting to apply these sorts of techniques in their research?

Absolutely, yeah. I think one of the best things I did was just getting involved in the open source stuff that's out there and the slacks that are available. There's a great Slack that's hosted by Xanadu where people are discussing this stuff. And yeah, I think it's awesome that these communities are starting to emerge because there's really smart people thinking about really interesting things that you can have discussions with whether it's in person or not. And the other thing is just to read a ton. I mean, Dr. Ghost will have more to say too, but sort of my strategies have been get involved and ask questions of people who are experts and read a lot. So I read a lot of papers and that's sort of what I do when I'm done working for the day.

So just to build on that broadly, research seems to be not so well defined when you start. And I think a lot of, especially for students, that's hard to get into. But there's actually a method to all of what we do. And as Marcus alluded to it, we have to get up to speed, so we have to do the background sort of legwork and try to understand where the field is currently. So pick a particular topic and then try to read up on it to whatever level you're comfortable with. And as you read, read something and try to learn a field, you will automatically find that there are some pieces you don't understand and Sometimes that's about you just being confused. So then you have to go and read up a little more on that. But sometimes it's because that just happens to be an open question in the field and that's how you can identify new questions that you can go and do research in. And so that's kind of a technique that, you know, you have to sort of just like everything else, you have to, you know, practice it and get used to it, and that then it becomes more natural to the point where you don't even know you're doing it. But every time you look at a new area, you sort of scan the field. You do like almost a survey, and then you identify parts that have been unexplored. And then you try to think, okay, this has not been explored. Let me go and find out who else is working on it. Maybe I can work with them. And if nobody else is working on it, then either it's really hard or you found something interesting to work on and should go ahead and try to, you know, find an approach to attack that problem if that's something you're interested in. So that's broadly the method.

Yeah, appreciate that. And I think that there's. I don't know that over time I've learned that the people that I feel like are able to adapt very quickly and really advance quickly are those that are willing to just ask a lot of questions and be willing to not be prideful and say, like, oh, I feel like I should know this, but really, if you don't know, be willing to ask, be willing to research. I think I've learned over time that people's spheres of knowledge, individual spheres of knowledge, are much smaller than I originally envisioned, that no one has all of the pieces of information to do a lot of these sorts of problems. And so it involves a lot of being willing to, to discuss and ask questions. So appreciate that perspective. Maybe to kind of bring us to a little bit of a close here regarding quantum computing in general. And I know that Marcus, you've mentioned quite a few things and open source things to start with. If people want to start and get exposed to quantum computing, maybe they're software engineers, but they're really interested in quantum computing, computing, maybe even contributing to open source projects where, where might be a good place for them to start to learn kind of the, the basics of quantum computing and then maybe start building something.

Sure. So there are a few places to go. There's something called the Quantum Open Source foundation, and they sort of have a collection of great projects. But in terms of getting involved sort of from the ground up, I think the best place for software developers to start is with IBM, just because their focus is really on sort of introducing quantum computing to the world and to especially from a software perspective. So they have really great tutorials. They have a fantastic Python library called Qiskit which enables you to sort of learn through tutorials and through expressive programming how quantum computing might work. And as I mentioned, they have lots of open source stuff too. So I think that's a great starting point. And once you've sort of mastered what you're doing there, I think then you can branch off to more sort of exotic flavors of quantum computing like Continuous Variable, which is Xanadu's. And they have fantastic documentation by the way. But I think sort of the world opens up once you've learned it in one place. And IBM is really great at sort of onboarding, I think.

Awesome.

Well, thank you both for taking time out of your, out of your busy schedules and out of your attempts to, you know, kind of explore this whole new world of computing. Really appreciate you taking time to chat with us. It's been a great conversation and just we'll definitely put links in our show notes to all of those things that we've talked about and really excited for, for our listeners to explore those and for me to explore them myself as well. So thank you so much for taking time.

Thank you.

Awesome. Thank you very much. Thank you.

All right, thank you for tuning into this episode of Practically I. If you enjoyed the show, do us a favor, go on itunes, give us a rating, go in your podcast app and favorite it. If you are on Twitter or social network, share a link with a friend, whatever you gotta do. Share the show with a friend if you enjoyed it. And bandwidth for Changelog is provided by fastly. Learn more@fastly.com and we catch our airs before our users do here at Changelog because of rollbar. Check them out@robot.com changelog and we're hosted on Linode Cloud Servers. Head to linode.com changelog check them out. Support this show this episode is hosted by Daniel Whitenack and Chris Benson. The music is by Breakmaster Cylinder and you can find more shows just like this@changelow.com when you go there, pop in your email address. Get our weekly email keeping you up to date with the news and podcasts for developers in your inbox every single week. Thanks for tuning in. We'll see you next week.

Foreign.

Because you've stuck in to the end of the show. Here's a preview of Brain Science, our upcoming podcast, coming out very soon. The easiest way to subscribe is to subscribe to our master feed@the changelog.com master. Get all of our podcasts in one single feed, plus some extras that only hit the master feed, including Brain Science. Brain Science is a podcast for the curious. We're exploring the inner workings of the human brain so we can understand things like behavior change, habit formation, mental health, and this thing we call the human condition. It's hosted by myself, Adam Stacowiak, and Muriel Reese, a doctor in clinical psychology. It's brain science applied not just how does the brain work, but how do we apply what we know about the brain to better our lives? Here we go. I think that's the most interesting thing I find with this subject, is I've lived most of my life. Well, I've lived all my life with a brain, for one, but I've lived most of my life not even knowing or thinking about how I. How it operates. And so my curiosity comes from, okay, now that I'm. I'm aware that the brain is the most important organ in my body. Without it, nothing else exists. In terms of, like, being able to operate. It's the primary source of all things that makes our body our body, I begin to think, okay, well, now how does it actually work, you know, so that I can understand different things about my life, my personality, why I love, why I hate, why I like, why I dislike, you know, all these different things. Habits drive, you know, willpower. All these different things play into that, and I begin to think, like, okay, how can I know more about my brain? And when you mention these worn paths in these grooves, that means, like, whenever I'm mulling over a thought or having anxiety, the thing that I'm mulling over or having anxiety about becomes more and more true or more and more real as my neurons fire together. Based on what you said here with the power of thoughts, is that if I keep thinking that way, it becomes more true to me than maybe somebody else because I've worn the path. Is that accurate to say?

You're spot on. If I'm to draw an analogy, it would really be that our thoughts are the lens through which we see our world and make sense of it, which is how people can have such varied perspectives. The thoughts we have are really that powerful if you can imagine them creating the fabric of so much of your world. And like I mentioned earlier about, you know, sort of filing things according to our feelings because we're more apt to remember things according to feelings. And so we want to be aware of the sort of circular nature of my thoughts and my feelings. And that, like, how I feel creates certain thoughts, and certain thoughts create certain feelings. And so if I want to feel different, I really need to do different and I need to think different. Because all of this is energy, right? I mean, you ever walk into a room or an interaction with an individual and it just sort of feels off, and you're thinking, like, what did I miss? Or like, I couldn't put my finger on it, but something was off because there's, there's feelings and you catch vibes because emotions are energy and thoughts, just like the. The neurons that fire together, that's, you know, electricity in our brain.

It's a. Certainly a deep subject which I'm just barely familiar with. But basically our brain is, you know, everything is electricity. As you're saying, you know, this. It's. It's our brain being able to somehow miraculously be able to process this electricity into thoughts, memories, recall, autobiographical understanding, time, you know, past, present, future, all this different stuff. And somehow these electrical charges throughout our entire body at the cellular level as well as the brain level, power us. And like you're saying, it's. It truly is literally energy.

It is. I mean, so one of the things that is important to know when it comes to neurons is they abide by the all or nothing rule. And what I mean by that is that they either fire or they don't. So sort of like that thing at the carnival with the hammer and you smack the weighted plate and it either goes to the top to ding or it doesn't. That's how neurons are. They get to a sort of threshold of excitement, and then they fire or they don't. And so being able to be aware of the thoughts, I think, for example, if I'm. If I am having a really rough day, like, maybe I pay attention to what I'm thinking about. Like, am I thinking about a loss that I went through? Am I thinking about a really challenging problem that I don't know how to overcome? Or maybe it's a relationship that I wanted to go differently.

And.

And then I don't feel very good, and maybe my stomach starts to hurt, and then I maybe start to get a headache. You know, all of our thoughts contribute to our own internal systems, and they.

Shape who we are too. You know, like. Like I said earlier, if I'm mulling over something or if I'm anxious about something, I just can't stop thinking about to some degree can even reshape my personality because I think that's what we might call moods. If I'm in a bad mood, it might be because I have an experience going on in my brain or my thoughts that I can't seem to shake away that's bringing me into a negative state. You know, my perspective in that scenario is that I can't get these bad thoughts out of my brain or I can't stop being anxious or having anxiety about something and therefore I yell at my wife. Or I'm not so nice to my son because my mood is changed by my thought patterns.

Yeah, you're spot on. And this is what I think. It's so important that we can understand that it's really possible to change these because if you can recognize like, oh, I'm not really upset with my wife or my son, I'm feeling bad, then again, I can put my lid on and go, what other options do I have available to me? Like, maybe I need to go work out. Maybe I need to go write down some of what is in my mind so that I can change some of those thoughts. Or maybe I need to do some meditation or talk to a friend. This is how we cope with some of those negative thoughts and negative feelings as well.

That's a preview of Brain Science. If you love where we're going with this, send us an email to get on the list to be notified the very moment this show gets released. Email us@eners changelow.com in the subject line. Put in all caps Brain Science with a couple bangs if you're really excited. You can also subscribe to our Master feed to get all of our shows in one single feed at the changelog.com master or search in your podcast app for changelove master. You'll find it. Subscribe get all of our shows and even those that only hit the master feed again. Changelow.com master.

Sa.