Podcasts

Podcast with Jack Hidary, Sandbox at Alphabet

3
November
,
2021

My guest today is Jack Hidary, AI and Quantum director at Sandbox at Alphabet. Jack and I spoke about the Hybridization of quantum computing, abstraction layers in software, the new edition of his quantum computing book, and much more.

Listen to additional podcasts here

THE FULL TRANSCRIPT IS BELOW

Yuval Boger (Classiq): Hello, Jack, and thanks for joining me today.

Jack Hidary (Alphabet): Hi, Yuval, great to be here.

Yuval: So who are you, and what do you do?

Jack: My name's Jack Hidary, author of Quantum Computing and Applied Approach. I currently am director of AI and Quantum at Sandbox at Alphabet. Sandbox at Alphabet is a unit of Alphabet, and we focus on enterprise solutions at the crossroads of quantum physics and AI.

Yuval: Let's talk about your book, I think there's a second edition coming out pretty shortly after the first edition. So tell me what's in the book, and why did we need a second edition so quickly?

Jack: Well, it's really interesting, the first edition, why I even write the first book, I was, along with my colleagues, I was teaching Quantum Computing and related topics, both inside of Google and outside. We were training students at universities, even some high schools, and also inside Google, we have our own internal university. We don't offer degrees, but we have an internal, professional learning courses. And we realized that there was really no textbook that really served our purposes at that time. Of course, we all know Mike and Ike, which is a wonderful textbook, and we still use Mike and Ike today. But Mike and Ike, now, it's been about 18 years since the last edition, and there's very strong theoretical work in Mike and Ike, but there was a need to have something that was more practical, more hands-on, and also recognize that there were now multiple frameworks for coding these quantum computers and frameworks that were open source and so easily available.

And more and more, I realized that these initial quantum computers would be going online on various clouds, and so students would want to try them out. And so my thought was, if I could write a textbook that combined the core theory that you need and the framework that you need, along with practical, hands-on coding examples, how to actually get your hands dirty and code for these kinds of initial machines, or even just run it on a simulator, that would be a great benefit both to my students and also to the thousands of people I was sure wanted to join this industry. So Yuval, that was really the raison d'etre of moving and writing this book to begin with, and in the last basically year and six months, two years since the publication of the first edition, a lot has happened in this field.

And what was wonderful is, I got so much great feedback and input from the readership. We have many faculty using the textbook in their courses for PhD students, for Master's students, even advanced undergrads, if they get past their core courses for physics, they often now can jump into quantum computing in their junior or senior year. So really it's being used in many, many great universities. I heard from them, I heard from their students. I also heard from many companies, Yuval, where the companies are using it as corporate training in their own settings, and so I had really a great list made for me by the readers of what to add to the second edition. And that's where the second edition came.

Yuval: And I assume it doesn't just cover Cirq, right? Does it also talk about Qiskit, or Q#, or other frameworks?

Jack: Yeah, really, really good point. From the beginning, from the first edition, and now also in the second edition, I really wanted this to be a textbook for everyone. And so it covers all the major platforms, both in terms of the physical platforms, right? There are seven major ways of building a quantum computer today, and the book is agnostic on which way is best. It points out just the underlying technology, and it gives the reader a lot of the citations and bibliography on how to investigate even more deeply, how you build these different instantiations, be it a photonic quantum computer, superconducting qubit, trapped ion, there are so many different ideas now about how to build a quantum computer, and that's part of what we've seen in the last two years, a flourishing of diversity in terms of companies popping out right and left, spinning out of universities, all kinds of wonderful activity on the hardware instantiation side.

Then of course you have, on the framework and library side, a tremendous growth as well, and the book covers Qiskit and Microsoft's SDK, and it covers Cirq and covers Rigetti, and all the major frameworks that people will encounter out there and has both examples of code in the book. But what I'm also excited about is the development of the GitHub site for the book, which has even more code, has all the code from the book, but even other examples as well. And that's where I also keep that up-to-date and also problem sets if they're being used in a core setting.

Yuval: As you think about quantum computers today, we have a limited number of qubits, and limited capabilities. But let's fast forward and - say in two years - when people are ready for the third edition of your book, there may be something that approaches a several hundred qubits or a thousand qubits. Do you think that the software frameworks today truly allow people to write code for a thousand qubit machine?

Jack: I think the software frameworks are very flexible and do anticipate the scaling of quantum computers. I think what we'll need is more of the abstraction that we see in classical computing. So if we go back to classical computing in the fifties and sixties, which was a time when you really had to know which hardware implementation you were writing for, and then over the course of 20 years after that, the industry really moved towards abstraction. And to the point where you could write code and run it on quite a range of hardware, implementations, and chips. So we achieved that in the classical world where today, obviously I could write Python code or Java code, and I could run it on quite a range of chipsets, but in the heart, in the quantum world right now, we're still have to be cognizant to some extent of which hardware implementation we're writing for and which one it will run on.

There are still idiosyncrasies that we have to be aware of. I think that over the next five, 10 years we'll make great headway towards abstraction the same way we did in the classical world, and I think that there's going to be the ability of individuals to know less, have to know less about the hardware that they're writing for, and that it's running on. Most coders today in the classical world, do not know much about the internals of an AMD versus an Intel versus an Nvidia type chips. They really are pretty ignorant of a lot of those details, and I think there'll be many companies, and Classiq is one of them, that are making headway towards helping developers get to their goal, and their goal is to write a great quantum circuit that accomplishes their mission, not necessarily to know all the underlying details of the hardware.

One thing, in terms of the second edition, just to go back to that and tie it to this, I added several sections, new sections to the second edition that were not in the first edition. In the second edition, I also expanded on all these software frameworks across the board and provided more detail and some more examples, but also included a new section on quantum error correction, as an example, because I think that's your point about the quantum computers getting better and better and numbers of qubits, we're going to start hopefully getting better at quantum error correction in terms of actually being able to realize full tolerant qubits. We won't have many of those full tolerant qubits in the next few years, but over the next five to 10 years, we're going to see, I think, real significant progress towards full tolerant error-corrected computing.

So that's a new section in this edition that was not really fleshed out in the first edition. I also spend much more time in updated quantum machine learning. In this case, giving an example from TensorFlow Quantum, but also on the website, there's other examples from PennyLane, from other frameworks, and I think quantum machine learning gets more exciting as the quantum computers get to greater stages of development, because then we can really think about hybridized computing. We can think about cloud computing in a hybridized manner where I have my CPU, my GPU, or TPU, and then I have my QPU, and having that hybridized environment is a really powerful paradigm for computing that I think in the next five to 10 years will be a great resource for developers and for companies

Yuval: Let's dive a little bit deeper into quantum computing on the cloud, because I think that there are sort of two approaches. One is to provide capacity, saying "Okay, I'm a cloud provider, and here's a quantum computer, and you can submit a job, and you can get a result and so on." But the other approach is to say, "Here's an API," just like there's a Maps API or an Alexa voice recognition API. Could there be just an “optimization as a service” using quantum API that I don't care about what the underlying hardware is, but I'm doing the optimization on the quantum computer? Which way do you see it going on capacity or on the API side?

Jack: I mean, Yuval, I would really want to even see more abstraction, right? So I'm looking for a day for what I call smart code, which is code that will automatically recognize the different parts of its own code base and recognize which kind of processor that particular part of the code base should run on. And so right now, we have to sub-routine out to a quantum computer and specifically run it, and so if I were running Shor's algorithm as an example, then actually a lot of it runs on a classical computer. And then there's a sub-routine that I would sub-routine out on the quantum computer, and then I would return out classical information back to the classic computer, but it'd be nice for the code itself to start to recognize which kind of processor is it optimal to run on, be at CPU or GPU or QPU.

So I think we're going to hit increasing levels of abstraction and also abstraction in terms of which type of quantum computer I might want to run on as well, just like which type of classical processor I might want to run on. And so I think that the cloudification, as it were, of quantum computing is a very healthy thing. The fact that quantum computers are cloud native, they're born on the cloud, is a very big advantage to this big trend in computing. Previous cycles of computing over the last 60, 70 years obviously did not start on the cloud, and so people had to buy these large machines, install these large machines ,have “care and feeding” of these large machines, and they really had a lot of overhead in dealing with that. And of course, the moment it was delivered, it was obsolete.

Whereas now on the cloud, I'm really excited to see all the different cloud providers adopt this so readily, because I think it would really drive not only adoption, but also innovation cycles could go much faster, because we don't have to wait till the 'install base' has to upgrade all its hardware. By having it on the cloud, every minute the users can get access to better and better technology, and be it clouds that are hosting their own technology, but now also we're seeing the fact that many of the cloud providers are hosting other people's technology.

And that I think is an exciting level of ecosystem that we didn't see two, three years ago. At that point, of course, IBM was hosting its own quantum computers as an example, but now we see cloud providers hosting multiple quantum computers from multiple providers, and I think it's going to get even more exciting as there are more companies that are scaling in the space. So I really believe that the fact that they're cloud native is a big part of how quantum computing will be adopted in the near future.

Yuval: When you were describing your book, you mentioned that one of the motivations is to train people. And obviously there seems to be a big shortage of people who are versed in quantum, who can create quantum circuits. All of a sudden there's a growth in companies who are hiring for that. Do you see abstraction as a solution to that as well? Because if, supposedly, if code is more abstracted then I don't need to understand quantum physics, I don't need to understand which gate is what, just like I don't need to understand assembly language when I write JavaScript.

Jack: Yeah, thank God. So again, back to the fifties. At that time, people had to know assembly and even machine code at that point. But yeah, I agree with you. The shortage is great already, the imbalance between supply and demand for quantum engineers and quantum researchers. And it's getting worse and worse every single day, because wonderful companies are getting funded by VCs and others and they're hiring, and capital is not the issue. Capital is not the issue. Companies who have credible teams, incredible roadmaps are raising money, and it's wonderful to see institutional investors and VCs and governments all investing in these nascent, quantum companies. It's really a very strong trend I've seen just in the last two years since the first edition. It's really ramped up dramatically, and the problem when I spoke to founders, many founders of the quantum companies call me, and the number one issue I hear is, "I can't get the talent. Where do I get the talent from?"

And so, I've personally been trying to help as many universities as possible to ramp up their programs, to double down on quantum information sciences, on QIS, but to your point, we now have to expand, and we can expand beyond the physics departments and electric engineering departments. Really, if we think about CS and think about a coder, who maybe is already in industry for five to 10 years, coding away at either a large, one of the big tech companies, or maybe a start-up, now, really they could start to join in on the quantum revolution, because many companies are making it easier and easier to join in. And I do think we continue to need very strong educational tools. That's again, the motivation for me for continuing to invest in this book and to put time into it, both in the physical format and also online on the site, because we really need on-ramps to this highway.

And it's very intimidating to get into quantum computing for most people. Even coders, who are experienced classical coders, it's a bit intimidating for them to suddenly take on. And the Python code itself is very straightforward. It's not so much the code. It's all the framework, one needs to understand, what are we doing? What is exactly as a qubit? And people, by the way, are curious, even though more and more they won't have to know what exactly the physical form of a qubit is, they're curious to know, and that's why I included that chapter, Yuval, because my students kept asking, "Okay, I understand the abstraction of a qubit, and you can have multiple qubits, and now we can have superposition, we could put them in state of superposition. We can entangle two qubits with each other, but what is a qubit? How do you physically make this?"

People are curious. So I think it's true that they won't have to know all the innards of each of these quantum computers, but I still think that most coders I train are still very curious as to how you build them. So I think some knowledge of that is helpful, but more and more, I think we need to move towards a wider aperture in terms of who we're training to bring into this field, and I think that the coding community, many of them have really been left out so far from this revolution. I really hope that we can bring them in to this revolution.

I also want to note on the point of view of diversity, we know that physics has a very big challenge in terms of lack of diversity as does CS still. CS is making some headway now, but certainly physics has a challenge, and so as we build this industry, as we build this ecosystem, I think it's very important to keep that in mind. I've been reaching out and working with a number of universities with very strong, diverse populations of students, and really hoping that as we build this ecosystem, we can have a very diverse stem pipeline in this area.

Yuval: You are in touch with many companies and probably many industries. Where do you see quantum going into production first? What kind of industries, what kind of applications would you guess would go first into production, beyond proof of concepts?

Jack: Well, it's a great question. And let's talk about quantum computing, but then Yuval, I suggest we move into other quantum technologies as well. When people think quantum these days, or they read about it online, on a website, or in a paper in a newspaper, often quantum computing is the focus, but it turns out there are lots of other parts of the quantum technology than just the building and use of quantum computers. And I find that people are not focusing enough on the other areas of quantum technology itself, but starting with quantum computing, I think there's several initial applications. Right now, of course, the quantum computers are very early. Right now, we don't have enough qubits to make one error corrected qubit, one logical qubit, as many of the listeners on this podcast may know, we talk in terms of both physical qubits, and then we use perhaps say a thousand physical qubits to then represent one logical qubit in a typical error correction scheme. And so the fact that we haven't reached a thousand physical qubits in one machine means that we don't have achieved one logical qubit.

I think that the industry is making good headway towards that, I'm very encouraged that there's diversity of thinking in terms of the companies trying different pathways to get to error corrected quantum computing. But as we get beyond the current NISQ era, the noisy intermediate scale quantum era that we're in right now, that’s the term coined by John Preskill, I think we'll start getting into the next level of computing where we'll have some, not only additional qubits, but high quality qubits, right? That's one thing that I hope listeners will be keeping in mind that it's not just a number of qubits, I know that horse race is very exciting, but it's also the quality of those qubits, the fidelity, and the ability to get to logical qubits.

As that happens, I think there'll be a number of interesting things to do. Of course, quantum chemistry comes to mind, there's many applications of quantum chemistry that we'd all love to run on scaled, quantum computers, and applications there would be in Pharma. Right now, the time period of molecule to medicine, of compound to clinic is still about 13 years in the United States, about a billion and a half to $2 billion to get from that molecule into an actual FDA approved medicine, and that's just simply too long. If we can use a more scaled quantum computers of the future to cut that time down, to do the work in silico that today we have to do on the bench, that would be a tremendous impact on the world, and the same thing for the chemicals industry, for the Dows, DuPonts, BASF, and others, there's tremendous opportunity for those companies.

There's a big trend, Yuval, in the chemicals industry now, to go towards greener chemistry. By greener, I mean less toxic, and so many companies are experimenting with new kinds of inputs. Instead of having petroleum inputs into a plastic, you can have a bio input into a plastic to make sure it degrades, it biodegrades after some period of time, and we don't have the calamity that we currently have in the oceans and the waterways of the world. You also have material science, think about batteries, think about every car company that I know has now announced publicly a very definite roadmap of the next 15, 20 years to get to full and or partial electrification. That means a lot more batteries in vehicles, and the battery chemistry today, is simply not as well suited for this kind of scale-up as it could be, so we really need to rethink that.

So there are a lot of applications in those verticals that we can think about. There's also different applications outside of molecular chemistry. We can think about optimization tools for finance, for risk analysis. I think there's a lot of exciting work being done on the theoretical side at this point with how quantum computers can affect financial services. So I think those are some of the initial areas that I think about, and certainly I think people with the vast array of resources being thrown at the scaling of quantum computers, I have optimism that we'll start to see more error corrected qubits coming soon to us. So that's what I would say on quantum computing. If we can turn now, Yuval, to the other areas of quantum technology outside of quantum computing, I think there are some very exciting areas as well.

Think about quantum sensing. Quantum sensing in a way is the flip side of quantum computing. Why is it so hard to have more qubits in a quantum computer? Why is there a “touchdown zone party” every time people add 20 more qubits to their machine? The reason is they're very fragile, right? It's very easy to get a qubit to decohere if it's impinged by a magnetic field or other kinds of external stimuli. But that's the very reason why quantum sensors are really exciting, and quantum sensors are not new. They've been around now in academia for decades. In fact, the Squid is a quantum sensor that goes back many decades, but the Squid had several problems and several issues, which is that it is a superconducting, and therefore has to be cryogenic. And by being cryogenic, obviously you're limited in terms of the kind of form factors and implementational ideas you can have. But newer type of quantum sensors in the last 20 years have emerged, which are room temperature or near room temperature, and don't have some of the issues that Squids have.

So I think the era of quantum sensing is upon us, and I think that will be a very exciting era for us. And in a way we'll be more implementational in a more near term way as opposed to quantum computing, because all you need is a handful of qubits in a sense to really get going with quantum sensors. You don't need a huge critical mass, because really you have the inverse situation here, you actually want the world to have stimuli that impinge on this sensor. So I think that's one area that's quite exciting.

I think there's going to be a lot of exciting work done also in cybersecurity as we start to get ramped up in quantum computers, and we get to the point where our cybersecurity architectures have to be rethought in terms of the current implementations and protocols. A lot of innovation is already happening in this area and will continue to happen. And again, I think that what I would like people to see also, coming full circle back to quantum computers, is a hybridized paradigm and viewpoint of the future. I hope that as more and more people join the quantum revolution, they really see quantum as part of a larger landscape with classical computing, GPUs and TPUs are really gaining tremendous inroads in terms of their application space, and I think we need to combine all these in a hybridized solution.

Yuval: So my next to last question is: let me give you a magic wand, and let's say that you can control what Classiq and hardware and software vendors in the quantum industry do for the next 18 months. What would you have us focus on?

Jack: Well, on the hardware side, what I would focus on is a few things. First, just to pick up on the education piece, I think one area that every company, every company has an opportunity, and if I can say responsibility to work on, is the education piece. I think there's, even a small company that just raised 10 to $20 million, they don't have a lot of resources, but there's always something to do with a local university to bring more people into this field and to bring a more diverse population into this field. So one thing I would just, for the listeners out there who are at the quantum companies, or even also at classical companies as well, it's something that all of us can join in on. And I welcome, certainly please reach me via the GitHub site for the book, my email is there or other ways to reach me via this podcast, via Yuval.

And I want to continue to partner with others to see how we can expand the educational aspects. I think leaving it just to traditional programs at universities alone, that was not going to be enough. We really need to expand having companies do corporate training inside their companies, but also offer training to those around them, and really everyone doubling down on that. So I would just, that's the first thing before any technology issue, I hope that both traditional companies and quantum companies can join in on. And that's something that we all can participate and collaborate, there's a lot of room for collaboration there.

The second thing is I hope that we can, for the cloud companies out there, I think the cloud companies already are off to a great start, they're providing more and more tools, they're providing explanations on their websites, they're providing kind of sandbox areas on their websites to play around with the machines. And so I think that's really good. I hope they will continue to have a free option for academic, for students, and not just for traditional students who are enrolled in a degree program, but also I think let's recognize more and more the non-traditional student, right? Somebody might be 35 years old, might be 40 years old, whatever age they are. They're now a student in this new field, let them also come in at low to no cost to try these technologies out for the quantum computing companies and for the cloud companies.

Third, abstraction. Let's continue to drive towards abstraction, a combined hybridized abstraction, so that we can have abstraction for our code, both for the classical processes we're writing to, but also for the quantum processes we're writing to.

And also, I hope to see one day, distributed quantum computing, DQC. I think that's an exciting part of the future. It's quite a few years till DQC can be realized on a wide scale basis, but we know the value of distributed computing in the classical world, and I believe strongly that a DQC future is an exciting one where we incorporate over quantum coherent networks, a future that we can link up different quantum devices. Some might be computing devices, some might be sensors, but a quantum coherent network that ties in these different quantum devices around the world.

Of course, we have to get to quantum coherent links just to scale some of the architectures of being considered today. And so many of the quantum computing companies that we know of want to use quantum coherent links just right in their own labs right now, to link up blocks of qubits so they can scale their own architecture. So I think there's a lot of excitement there, and I hope that as an industry, we can really work on a future that has a quantum coherent network. CalTech is doing great work there, as is MIT, as are a number of universities in Europe. There's some initial initiatives across these universities that are funded by the various governments, but I think over time, I think DQC will be a great, wonderful part of our computing future.

Yuval: Excellent. That sounds like a lot of work for 18 months, but we'll do our best. So Jack, what's the best way to get in touch with you and learn more about your work?

Jack: Sure. The website, if people just go on their favorite search engine or be it Google, be in otherwise they could just type in GitHub, and my last name Hidary, and quantum, and you'll find the full GitHub link. And I welcome to you to my website, and my email is on there and please pop an email to me. It's just jack@hidary.com. And also you could find the problem sets for the book. We don't provide the solutions on the website, because the faculty asked us to hold those back. But faculty who are reaching us can get the solution manual, but also non-traditional faculty. Again, there's a lot of people who reached out to us, Yuval, who said, "Hey, I'm teaching an informal course in my corporation, and I'd like to be considered a faculty." And absolutely we do that. So you don't have to be part of any formal university to do that. And I hope that non-traditional learning, continuous learning for adults will be a big part of how people use this textbook.

Yuval: That's great, Jack, thank you so much for joining me today.

Jack: Thanks, Yuval. Great to be with you.


My guest today is Jack Hidary, AI and Quantum director at Sandbox at Alphabet. Jack and I spoke about the Hybridization of quantum computing, abstraction layers in software, the new edition of his quantum computing book, and much more.

Listen to additional podcasts here

THE FULL TRANSCRIPT IS BELOW

Yuval Boger (Classiq): Hello, Jack, and thanks for joining me today.

Jack Hidary (Alphabet): Hi, Yuval, great to be here.

Yuval: So who are you, and what do you do?

Jack: My name's Jack Hidary, author of Quantum Computing and Applied Approach. I currently am director of AI and Quantum at Sandbox at Alphabet. Sandbox at Alphabet is a unit of Alphabet, and we focus on enterprise solutions at the crossroads of quantum physics and AI.

Yuval: Let's talk about your book, I think there's a second edition coming out pretty shortly after the first edition. So tell me what's in the book, and why did we need a second edition so quickly?

Jack: Well, it's really interesting, the first edition, why I even write the first book, I was, along with my colleagues, I was teaching Quantum Computing and related topics, both inside of Google and outside. We were training students at universities, even some high schools, and also inside Google, we have our own internal university. We don't offer degrees, but we have an internal, professional learning courses. And we realized that there was really no textbook that really served our purposes at that time. Of course, we all know Mike and Ike, which is a wonderful textbook, and we still use Mike and Ike today. But Mike and Ike, now, it's been about 18 years since the last edition, and there's very strong theoretical work in Mike and Ike, but there was a need to have something that was more practical, more hands-on, and also recognize that there were now multiple frameworks for coding these quantum computers and frameworks that were open source and so easily available.

And more and more, I realized that these initial quantum computers would be going online on various clouds, and so students would want to try them out. And so my thought was, if I could write a textbook that combined the core theory that you need and the framework that you need, along with practical, hands-on coding examples, how to actually get your hands dirty and code for these kinds of initial machines, or even just run it on a simulator, that would be a great benefit both to my students and also to the thousands of people I was sure wanted to join this industry. So Yuval, that was really the raison d'etre of moving and writing this book to begin with, and in the last basically year and six months, two years since the publication of the first edition, a lot has happened in this field.

And what was wonderful is, I got so much great feedback and input from the readership. We have many faculty using the textbook in their courses for PhD students, for Master's students, even advanced undergrads, if they get past their core courses for physics, they often now can jump into quantum computing in their junior or senior year. So really it's being used in many, many great universities. I heard from them, I heard from their students. I also heard from many companies, Yuval, where the companies are using it as corporate training in their own settings, and so I had really a great list made for me by the readers of what to add to the second edition. And that's where the second edition came.

Yuval: And I assume it doesn't just cover Cirq, right? Does it also talk about Qiskit, or Q#, or other frameworks?

Jack: Yeah, really, really good point. From the beginning, from the first edition, and now also in the second edition, I really wanted this to be a textbook for everyone. And so it covers all the major platforms, both in terms of the physical platforms, right? There are seven major ways of building a quantum computer today, and the book is agnostic on which way is best. It points out just the underlying technology, and it gives the reader a lot of the citations and bibliography on how to investigate even more deeply, how you build these different instantiations, be it a photonic quantum computer, superconducting qubit, trapped ion, there are so many different ideas now about how to build a quantum computer, and that's part of what we've seen in the last two years, a flourishing of diversity in terms of companies popping out right and left, spinning out of universities, all kinds of wonderful activity on the hardware instantiation side.

Then of course you have, on the framework and library side, a tremendous growth as well, and the book covers Qiskit and Microsoft's SDK, and it covers Cirq and covers Rigetti, and all the major frameworks that people will encounter out there and has both examples of code in the book. But what I'm also excited about is the development of the GitHub site for the book, which has even more code, has all the code from the book, but even other examples as well. And that's where I also keep that up-to-date and also problem sets if they're being used in a core setting.

Yuval: As you think about quantum computers today, we have a limited number of qubits, and limited capabilities. But let's fast forward and - say in two years - when people are ready for the third edition of your book, there may be something that approaches a several hundred qubits or a thousand qubits. Do you think that the software frameworks today truly allow people to write code for a thousand qubit machine?

Jack: I think the software frameworks are very flexible and do anticipate the scaling of quantum computers. I think what we'll need is more of the abstraction that we see in classical computing. So if we go back to classical computing in the fifties and sixties, which was a time when you really had to know which hardware implementation you were writing for, and then over the course of 20 years after that, the industry really moved towards abstraction. And to the point where you could write code and run it on quite a range of hardware, implementations, and chips. So we achieved that in the classical world where today, obviously I could write Python code or Java code, and I could run it on quite a range of chipsets, but in the heart, in the quantum world right now, we're still have to be cognizant to some extent of which hardware implementation we're writing for and which one it will run on.

There are still idiosyncrasies that we have to be aware of. I think that over the next five, 10 years we'll make great headway towards abstraction the same way we did in the classical world, and I think that there's going to be the ability of individuals to know less, have to know less about the hardware that they're writing for, and that it's running on. Most coders today in the classical world, do not know much about the internals of an AMD versus an Intel versus an Nvidia type chips. They really are pretty ignorant of a lot of those details, and I think there'll be many companies, and Classiq is one of them, that are making headway towards helping developers get to their goal, and their goal is to write a great quantum circuit that accomplishes their mission, not necessarily to know all the underlying details of the hardware.

One thing, in terms of the second edition, just to go back to that and tie it to this, I added several sections, new sections to the second edition that were not in the first edition. In the second edition, I also expanded on all these software frameworks across the board and provided more detail and some more examples, but also included a new section on quantum error correction, as an example, because I think that's your point about the quantum computers getting better and better and numbers of qubits, we're going to start hopefully getting better at quantum error correction in terms of actually being able to realize full tolerant qubits. We won't have many of those full tolerant qubits in the next few years, but over the next five to 10 years, we're going to see, I think, real significant progress towards full tolerant error-corrected computing.

So that's a new section in this edition that was not really fleshed out in the first edition. I also spend much more time in updated quantum machine learning. In this case, giving an example from TensorFlow Quantum, but also on the website, there's other examples from PennyLane, from other frameworks, and I think quantum machine learning gets more exciting as the quantum computers get to greater stages of development, because then we can really think about hybridized computing. We can think about cloud computing in a hybridized manner where I have my CPU, my GPU, or TPU, and then I have my QPU, and having that hybridized environment is a really powerful paradigm for computing that I think in the next five to 10 years will be a great resource for developers and for companies

Yuval: Let's dive a little bit deeper into quantum computing on the cloud, because I think that there are sort of two approaches. One is to provide capacity, saying "Okay, I'm a cloud provider, and here's a quantum computer, and you can submit a job, and you can get a result and so on." But the other approach is to say, "Here's an API," just like there's a Maps API or an Alexa voice recognition API. Could there be just an “optimization as a service” using quantum API that I don't care about what the underlying hardware is, but I'm doing the optimization on the quantum computer? Which way do you see it going on capacity or on the API side?

Jack: I mean, Yuval, I would really want to even see more abstraction, right? So I'm looking for a day for what I call smart code, which is code that will automatically recognize the different parts of its own code base and recognize which kind of processor that particular part of the code base should run on. And so right now, we have to sub-routine out to a quantum computer and specifically run it, and so if I were running Shor's algorithm as an example, then actually a lot of it runs on a classical computer. And then there's a sub-routine that I would sub-routine out on the quantum computer, and then I would return out classical information back to the classic computer, but it'd be nice for the code itself to start to recognize which kind of processor is it optimal to run on, be at CPU or GPU or QPU.

So I think we're going to hit increasing levels of abstraction and also abstraction in terms of which type of quantum computer I might want to run on as well, just like which type of classical processor I might want to run on. And so I think that the cloudification, as it were, of quantum computing is a very healthy thing. The fact that quantum computers are cloud native, they're born on the cloud, is a very big advantage to this big trend in computing. Previous cycles of computing over the last 60, 70 years obviously did not start on the cloud, and so people had to buy these large machines, install these large machines ,have “care and feeding” of these large machines, and they really had a lot of overhead in dealing with that. And of course, the moment it was delivered, it was obsolete.

Whereas now on the cloud, I'm really excited to see all the different cloud providers adopt this so readily, because I think it would really drive not only adoption, but also innovation cycles could go much faster, because we don't have to wait till the 'install base' has to upgrade all its hardware. By having it on the cloud, every minute the users can get access to better and better technology, and be it clouds that are hosting their own technology, but now also we're seeing the fact that many of the cloud providers are hosting other people's technology.

And that I think is an exciting level of ecosystem that we didn't see two, three years ago. At that point, of course, IBM was hosting its own quantum computers as an example, but now we see cloud providers hosting multiple quantum computers from multiple providers, and I think it's going to get even more exciting as there are more companies that are scaling in the space. So I really believe that the fact that they're cloud native is a big part of how quantum computing will be adopted in the near future.

Yuval: When you were describing your book, you mentioned that one of the motivations is to train people. And obviously there seems to be a big shortage of people who are versed in quantum, who can create quantum circuits. All of a sudden there's a growth in companies who are hiring for that. Do you see abstraction as a solution to that as well? Because if, supposedly, if code is more abstracted then I don't need to understand quantum physics, I don't need to understand which gate is what, just like I don't need to understand assembly language when I write JavaScript.

Jack: Yeah, thank God. So again, back to the fifties. At that time, people had to know assembly and even machine code at that point. But yeah, I agree with you. The shortage is great already, the imbalance between supply and demand for quantum engineers and quantum researchers. And it's getting worse and worse every single day, because wonderful companies are getting funded by VCs and others and they're hiring, and capital is not the issue. Capital is not the issue. Companies who have credible teams, incredible roadmaps are raising money, and it's wonderful to see institutional investors and VCs and governments all investing in these nascent, quantum companies. It's really a very strong trend I've seen just in the last two years since the first edition. It's really ramped up dramatically, and the problem when I spoke to founders, many founders of the quantum companies call me, and the number one issue I hear is, "I can't get the talent. Where do I get the talent from?"

And so, I've personally been trying to help as many universities as possible to ramp up their programs, to double down on quantum information sciences, on QIS, but to your point, we now have to expand, and we can expand beyond the physics departments and electric engineering departments. Really, if we think about CS and think about a coder, who maybe is already in industry for five to 10 years, coding away at either a large, one of the big tech companies, or maybe a start-up, now, really they could start to join in on the quantum revolution, because many companies are making it easier and easier to join in. And I do think we continue to need very strong educational tools. That's again, the motivation for me for continuing to invest in this book and to put time into it, both in the physical format and also online on the site, because we really need on-ramps to this highway.

And it's very intimidating to get into quantum computing for most people. Even coders, who are experienced classical coders, it's a bit intimidating for them to suddenly take on. And the Python code itself is very straightforward. It's not so much the code. It's all the framework, one needs to understand, what are we doing? What is exactly as a qubit? And people, by the way, are curious, even though more and more they won't have to know what exactly the physical form of a qubit is, they're curious to know, and that's why I included that chapter, Yuval, because my students kept asking, "Okay, I understand the abstraction of a qubit, and you can have multiple qubits, and now we can have superposition, we could put them in state of superposition. We can entangle two qubits with each other, but what is a qubit? How do you physically make this?"

People are curious. So I think it's true that they won't have to know all the innards of each of these quantum computers, but I still think that most coders I train are still very curious as to how you build them. So I think some knowledge of that is helpful, but more and more, I think we need to move towards a wider aperture in terms of who we're training to bring into this field, and I think that the coding community, many of them have really been left out so far from this revolution. I really hope that we can bring them in to this revolution.

I also want to note on the point of view of diversity, we know that physics has a very big challenge in terms of lack of diversity as does CS still. CS is making some headway now, but certainly physics has a challenge, and so as we build this industry, as we build this ecosystem, I think it's very important to keep that in mind. I've been reaching out and working with a number of universities with very strong, diverse populations of students, and really hoping that as we build this ecosystem, we can have a very diverse stem pipeline in this area.

Yuval: You are in touch with many companies and probably many industries. Where do you see quantum going into production first? What kind of industries, what kind of applications would you guess would go first into production, beyond proof of concepts?

Jack: Well, it's a great question. And let's talk about quantum computing, but then Yuval, I suggest we move into other quantum technologies as well. When people think quantum these days, or they read about it online, on a website, or in a paper in a newspaper, often quantum computing is the focus, but it turns out there are lots of other parts of the quantum technology than just the building and use of quantum computers. And I find that people are not focusing enough on the other areas of quantum technology itself, but starting with quantum computing, I think there's several initial applications. Right now, of course, the quantum computers are very early. Right now, we don't have enough qubits to make one error corrected qubit, one logical qubit, as many of the listeners on this podcast may know, we talk in terms of both physical qubits, and then we use perhaps say a thousand physical qubits to then represent one logical qubit in a typical error correction scheme. And so the fact that we haven't reached a thousand physical qubits in one machine means that we don't have achieved one logical qubit.

I think that the industry is making good headway towards that, I'm very encouraged that there's diversity of thinking in terms of the companies trying different pathways to get to error corrected quantum computing. But as we get beyond the current NISQ era, the noisy intermediate scale quantum era that we're in right now, that’s the term coined by John Preskill, I think we'll start getting into the next level of computing where we'll have some, not only additional qubits, but high quality qubits, right? That's one thing that I hope listeners will be keeping in mind that it's not just a number of qubits, I know that horse race is very exciting, but it's also the quality of those qubits, the fidelity, and the ability to get to logical qubits.

As that happens, I think there'll be a number of interesting things to do. Of course, quantum chemistry comes to mind, there's many applications of quantum chemistry that we'd all love to run on scaled, quantum computers, and applications there would be in Pharma. Right now, the time period of molecule to medicine, of compound to clinic is still about 13 years in the United States, about a billion and a half to $2 billion to get from that molecule into an actual FDA approved medicine, and that's just simply too long. If we can use a more scaled quantum computers of the future to cut that time down, to do the work in silico that today we have to do on the bench, that would be a tremendous impact on the world, and the same thing for the chemicals industry, for the Dows, DuPonts, BASF, and others, there's tremendous opportunity for those companies.

There's a big trend, Yuval, in the chemicals industry now, to go towards greener chemistry. By greener, I mean less toxic, and so many companies are experimenting with new kinds of inputs. Instead of having petroleum inputs into a plastic, you can have a bio input into a plastic to make sure it degrades, it biodegrades after some period of time, and we don't have the calamity that we currently have in the oceans and the waterways of the world. You also have material science, think about batteries, think about every car company that I know has now announced publicly a very definite roadmap of the next 15, 20 years to get to full and or partial electrification. That means a lot more batteries in vehicles, and the battery chemistry today, is simply not as well suited for this kind of scale-up as it could be, so we really need to rethink that.

So there are a lot of applications in those verticals that we can think about. There's also different applications outside of molecular chemistry. We can think about optimization tools for finance, for risk analysis. I think there's a lot of exciting work being done on the theoretical side at this point with how quantum computers can affect financial services. So I think those are some of the initial areas that I think about, and certainly I think people with the vast array of resources being thrown at the scaling of quantum computers, I have optimism that we'll start to see more error corrected qubits coming soon to us. So that's what I would say on quantum computing. If we can turn now, Yuval, to the other areas of quantum technology outside of quantum computing, I think there are some very exciting areas as well.

Think about quantum sensing. Quantum sensing in a way is the flip side of quantum computing. Why is it so hard to have more qubits in a quantum computer? Why is there a “touchdown zone party” every time people add 20 more qubits to their machine? The reason is they're very fragile, right? It's very easy to get a qubit to decohere if it's impinged by a magnetic field or other kinds of external stimuli. But that's the very reason why quantum sensors are really exciting, and quantum sensors are not new. They've been around now in academia for decades. In fact, the Squid is a quantum sensor that goes back many decades, but the Squid had several problems and several issues, which is that it is a superconducting, and therefore has to be cryogenic. And by being cryogenic, obviously you're limited in terms of the kind of form factors and implementational ideas you can have. But newer type of quantum sensors in the last 20 years have emerged, which are room temperature or near room temperature, and don't have some of the issues that Squids have.

So I think the era of quantum sensing is upon us, and I think that will be a very exciting era for us. And in a way we'll be more implementational in a more near term way as opposed to quantum computing, because all you need is a handful of qubits in a sense to really get going with quantum sensors. You don't need a huge critical mass, because really you have the inverse situation here, you actually want the world to have stimuli that impinge on this sensor. So I think that's one area that's quite exciting.

I think there's going to be a lot of exciting work done also in cybersecurity as we start to get ramped up in quantum computers, and we get to the point where our cybersecurity architectures have to be rethought in terms of the current implementations and protocols. A lot of innovation is already happening in this area and will continue to happen. And again, I think that what I would like people to see also, coming full circle back to quantum computers, is a hybridized paradigm and viewpoint of the future. I hope that as more and more people join the quantum revolution, they really see quantum as part of a larger landscape with classical computing, GPUs and TPUs are really gaining tremendous inroads in terms of their application space, and I think we need to combine all these in a hybridized solution.

Yuval: So my next to last question is: let me give you a magic wand, and let's say that you can control what Classiq and hardware and software vendors in the quantum industry do for the next 18 months. What would you have us focus on?

Jack: Well, on the hardware side, what I would focus on is a few things. First, just to pick up on the education piece, I think one area that every company, every company has an opportunity, and if I can say responsibility to work on, is the education piece. I think there's, even a small company that just raised 10 to $20 million, they don't have a lot of resources, but there's always something to do with a local university to bring more people into this field and to bring a more diverse population into this field. So one thing I would just, for the listeners out there who are at the quantum companies, or even also at classical companies as well, it's something that all of us can join in on. And I welcome, certainly please reach me via the GitHub site for the book, my email is there or other ways to reach me via this podcast, via Yuval.

And I want to continue to partner with others to see how we can expand the educational aspects. I think leaving it just to traditional programs at universities alone, that was not going to be enough. We really need to expand having companies do corporate training inside their companies, but also offer training to those around them, and really everyone doubling down on that. So I would just, that's the first thing before any technology issue, I hope that both traditional companies and quantum companies can join in on. And that's something that we all can participate and collaborate, there's a lot of room for collaboration there.

The second thing is I hope that we can, for the cloud companies out there, I think the cloud companies already are off to a great start, they're providing more and more tools, they're providing explanations on their websites, they're providing kind of sandbox areas on their websites to play around with the machines. And so I think that's really good. I hope they will continue to have a free option for academic, for students, and not just for traditional students who are enrolled in a degree program, but also I think let's recognize more and more the non-traditional student, right? Somebody might be 35 years old, might be 40 years old, whatever age they are. They're now a student in this new field, let them also come in at low to no cost to try these technologies out for the quantum computing companies and for the cloud companies.

Third, abstraction. Let's continue to drive towards abstraction, a combined hybridized abstraction, so that we can have abstraction for our code, both for the classical processes we're writing to, but also for the quantum processes we're writing to.

And also, I hope to see one day, distributed quantum computing, DQC. I think that's an exciting part of the future. It's quite a few years till DQC can be realized on a wide scale basis, but we know the value of distributed computing in the classical world, and I believe strongly that a DQC future is an exciting one where we incorporate over quantum coherent networks, a future that we can link up different quantum devices. Some might be computing devices, some might be sensors, but a quantum coherent network that ties in these different quantum devices around the world.

Of course, we have to get to quantum coherent links just to scale some of the architectures of being considered today. And so many of the quantum computing companies that we know of want to use quantum coherent links just right in their own labs right now, to link up blocks of qubits so they can scale their own architecture. So I think there's a lot of excitement there, and I hope that as an industry, we can really work on a future that has a quantum coherent network. CalTech is doing great work there, as is MIT, as are a number of universities in Europe. There's some initial initiatives across these universities that are funded by the various governments, but I think over time, I think DQC will be a great, wonderful part of our computing future.

Yuval: Excellent. That sounds like a lot of work for 18 months, but we'll do our best. So Jack, what's the best way to get in touch with you and learn more about your work?

Jack: Sure. The website, if people just go on their favorite search engine or be it Google, be in otherwise they could just type in GitHub, and my last name Hidary, and quantum, and you'll find the full GitHub link. And I welcome to you to my website, and my email is on there and please pop an email to me. It's just jack@hidary.com. And also you could find the problem sets for the book. We don't provide the solutions on the website, because the faculty asked us to hold those back. But faculty who are reaching us can get the solution manual, but also non-traditional faculty. Again, there's a lot of people who reached out to us, Yuval, who said, "Hey, I'm teaching an informal course in my corporation, and I'd like to be considered a faculty." And absolutely we do that. So you don't have to be part of any formal university to do that. And I hope that non-traditional learning, continuous learning for adults will be a big part of how people use this textbook.

Yuval: That's great, Jack, thank you so much for joining me today.

Jack: Thanks, Yuval. Great to be with you.


About "The Qubit Guy's Podcast"

Hosted by The Qubit Guy (Yuval Boger, our Chief Marketing Officer), the podcast hosts thought leaders in quantum computing to discuss business and technical questions that impact the quantum computing ecosystem. Our guests provide interesting insights about quantum computer software and algorithm, quantum computer hardware, key applications for quantum computing, market studies of the quantum industry and more.

If you would like to suggest a guest for the podcast, please contact us.

See Also

No items found.

Start Creating Quantum Software Without Limits

contact us