Quantum Computing – Top 3 Microsoft Breakthroughs with Krysta Svore



Hello everyone and welcome to Microsoft Mechanics live today I'm joined by one of the foremost experts in quantum computing Krysta Svore and we're gonna attempt to demystify the core concepts around the topic and make it real for everyone and we're also gonna look at Microsoft's contribution in the field developing a full system approach comprising of three major breakthroughs first with the topological qubit a new way to be able to do Quantum computations also the scalable quantum hardware with cryogenic control and a complete software stack for programming and controlling quantum computers but first please give me a warm welcome for Krysta Svore for thanks for joining us today so Microsoft has been pioneering some really important breakthroughs in the space of quantum computing but before we go into the significance of all these innovations can you help explain what quantum computing is yeah of course so quantum computing is really a big paradigm shift it's a new and different form of computing that really goes exponentially beyond the capabilities of our best and most powerful supercomputers today and it's grounded in the principles of quantum mechanics which underlies the fundamental theory of nature at the smallest atomic and subatomic scales so with a quantum computer it's quantum mechanics that really governs how we store the information and then how we process and compute on the information so everybody here is probably a computer scientist so how does this compare to today's computing right so today computers of course are based on transistors and transistors are like a switch they support only two states like a light switch with two positions so either on or off and of course this train lates to the binary ones and zeros that we ultimately program in now as you need more information with a transistor based machine and more compute power you need to add more ones and zeros into your machine and therefore you're adding more and more transistors in and of course we're striving to fit as many transistors as we can as Satya mentioned in the keynote you know we have billions of transistors in the Xbox One X we're trying to fit as many as possible into a very small chip size and you know we're gonna hit some physics limits so it's a lot of transistors and now they're getting down to 10 nanometers an even smaller which is mind-boggling itself but the quantum chips are gonna be a lot smaller right well yeah so we're operating you know on a smaller scale in terms of the information but ultimately you know this quantum computer is gonna sit in a large dilution refrigerator but indeed the the information the way we hold and process information really operates differently so in the transistor model you have this on/off switch a quantum computer is you know somewhat more like a dimmer switch so you can take on both the 0 and the 1 state simultaneously and that's what we call a superposition state so there's a couple ways to think about this right so what would it look like I guess when we take about I think I think about binary now versus in the quantum world yeah so so in binary today if we we take 4 light switches let's say write a string of 4 bits and we take all the zeros and ones you work through all the possible string combinations you can get you know 16 values with this 2 to the 4th (16) possible states but that combination of the zeros and ones you know you can only operate on one of those strings at a given time now in a quantum state let's take for qubits instead of 4 bits now these four qubits in fact store all of those 16 states at once in 4 qubits so that means that the quantum computer can have a lot more information and process a lot more information simultaneously but campus-level then can to the compute that we might be able to do with maybe a supercomputer and parallel computing and transistors now yeah it's a great great question so you know you think of quantum computers as having this massive parallelism now classically we could try to achieve that with a parallel computer but now I need that many more resources that many more switches and I operate on all of the possible states simultaneously but I need four bits for every possible string in a quantum computer again I only need the four qubits to operate on all of those states at once in terms of in terms of the way that quantum is working now we're creating kind of lossless power grids right yeah so in terms of the applications of a quantum computer what we what we hope to achieve with the quantum computer is to solve truly revolutionary problems you know world-changing problems and you know clean energy is one thing that we would really strive to strive to overcome right and solve so there with a quantum computer we can look at materials and we can study materials that may actually super conduct at a high temperature and in turn that would allow us to create a lossless power grid and so forth and and these applications go beyond just materials we can look in chemistry look for better catalysts and then of course artificial intelligence and machine learning are our other other areas and a lot of these problems are really really hard to be able to solve in terms of Materials Science and all the things that we would do maybe with classical computer this this gets a lot easier and a lot more powerful with quantum but what are some of the ways that people might be able to harness this power then in the future right so so here in quantum computing we're dealing with these you know the very very small scale subatomic particles that are governed by the laws of quantum mechanics and so we need a totally different approach to computing into programming and we actually have to really learn how to program and develop these new applications and algorithms it's a totally new paradigm so at Microsoft what we're doing is putting together you know a full-stack approach a comprehensive approach to enable scalable quantum computing and that includes everything from the hardware to the software to the application layer you know what are the revolutionary applications we're really gonna run on this and to do that we have to bring together a very diverse team so we've actually been working in the space for almost 20 years at Microsoft which is pretty incredible and I think one of the truly distinct things about our approach at Microsoft is how we've assembled this amazing research and engineering team in labs all around the world dedicated to advancing quantum computing and we've partnered with academia and universities and we've brought together a truly diverse expertise and and that's really required to build out this full stack so ultimately our goal is to realize a scalable and programmable quantum computer so that we can tackle the world's most challenging problems and then of course with that ultimately we envision this quantum computer sitting in Azure right so just as in as we have today in Azure you can access a GPU CPU FPGA right as your accelerator options we envision a quantum computer sitting in Azure as one of your possible accelerators that you can call from the apps that you're building in the cloud and and then ultimately you know achieve a solution you may otherwise not be able to achieve classically that's great so imagine picking out a VM instead of going for the D series VM or E or F-series VM you've actually got a quantum set of compute that you're working off of so really great to see that and and really imagine the scale that you can achieve through all the compute that we'll be able to do with quantum in the future right so yeah exactly I mean scaling is one of the key things here we want to be able to achieve a very scaled out quantum computer with with you know thousands and beyond in terms of the numbers of qubits and to do that you need hardware so one of the the key pieces that we're working on is building out the scalable qubit the qubit that's going to be robust enough to actually scale to these large numbers and qubits are inherently fragile so they have to stay in this cold environment because they they want it you know they otherwise they're disturbed by the thermal vibrations and by you know temperature the heat itself so the quantum world in the quantum world qubits are very susceptible to noise and the type of qubit were working on at Microsoft a topological qubit is extremely robust to noise which means we'll be able to scale with far fewer resources than other competing qubit approaches and there are actually different types of qubits right I mean what makes the topological qubit special and more robust yeah so this topological qubit it really is a very distinct approach and it's extremely robust to noise and and it's it's robust to noise because the information is actually stored non locally so it's stored in a more global fashion if you will and because of that it means it's less error-prone and with with being less error-prone you can scale up much more easily so the overhead to achieve a long computation or a computation with many many operations is far far less several orders of magnitude less then kind of the resources required if you were using another type of qubit so the type of qubit and the system we're building at Microsoft is really about reliably storing information and computing on this information and being able to do that at a large scale long algorithms lots of operations and part of this is in terms of being able to actually operate the compute and do all of that across different temperature spectrums what does that what does the compute environment actually look like in terms of leveraging that topological qubit right so so here's a great a great picture right at the at the bottom of this stack of this kind of this complete system we have the the quantum computer where the qubits sit and that's that's you know sitting it 0.01 degrees from absolute zero right it's very very cold minus 459 degrees Fahrenheit roughly and then above that sits the cryogenic computer this is a classical computer unlike any classical computer we've ever seen or developed and at Microsoft we're developing this computer this computer controls the quantum computer and it sits at roughly 20 millikelvin that's around minus 452 degrees Fahrenheit so not a whole lot warmer than the quantum chip and and then beyond that you know we sit in the outs in this room temperature world you know around 70 degrees and and then we're gonna program the quantum computer from from this temperature and so we we have to build out a software stack and all of this hardware a complete stack that enables us to scale this whole system to a thousand qubits and beyond and just to put it into context you know at a thousand qubits you're looking at around 10 to the 300 times the processing capacity of today's supercomputers assuming a petabyte petabytes supercomputer so you know this it's not like the next generation of supercomputers is going to get you that much closer this is this is you know far far beyond that any classical computing we can imagine and and and then we have this software that will allow you to actually harness that power and program quantum applications so part of the challenge that in terms of the software because we've been talking about all of the kind of subatomic particle they're really really small levels and you know the size of the qubits and all these things but you have to make the tools so there's something that people are familiar with right so what are we been doing in terms of the software stack to make it something that people can leverage that maybe in the future will be able to actually start building the right computation right so so we're engineering this cryogenic system the quantum chip that cryogenic control and then the software stack and the key there is tools for developers like everyone in the audience today we want to give you tools to be able to start building quantum solutions and start understanding and developing a generation of quantum applications and so to do that we've actually designed a special language a domain-specific language that's really designed for scalable quantum computing so it has some awesome features it kind of looks and feels like languages you already know like F# and C# and Python but then in it also allows you access to quantum subroutines in quantum libraries and ways to actually call some of the special properties of quantum computing we've put this in Visual Studio integrated it right into Visual Studio so that you have also the kind of environment that you're used to and tools that developers love and within there it allows us to do quantum debugging so you can actually debug your quantum code which which is very distinct to our tool chain you can visualize the quantum state as it evolves through the program and you can see in this slide here that you also have syntax coloring making it more readable and understandable now you can also test your test your quantum program on a simulator and the simulator runs on your classical machine and allows you to actually simulate what your quantum application would do and locally you can do that on around 30 cubits and then we also have an azure based simulator that allows you to do that up to around 40 qubits and beyond and of course you know simulating those sizes they don't sound like a lot of qubits but realize that's you know for every additional qubit it's double the memory so we're looking at for 230 qubits it would require 10 to the 80 bytes which is more bytes than there are particles in our visible universe so simulating 40 qubits and 30 cubits is in fact a pretty nice number for the classical machine so just a little bit more powerful than say simulating a mobile phone can we see an example then of what this code actually does can actually what we walk through what the what the intent of this code that we're actually showing here would do right so this this this code here this is our new language and it's in Visual Studio and here what we're showing is an example of what we like to call the quantum "hello world" right everyone starts a new programming language or writing hello world let's do the same in quantum computing in our case hello world is really a an algorithm for quantum teleportation so this is one of the basic operations that quantum quantum mechanics gives us if you will and so this is the circuit diagram in in quantum computing you read a circuit diagram like music from left to right so think of it as the lines being the qubits and then the gates the operations if you're an engineer electrical engineer you might recognize this you read the operations like notes so left to right and the qubit information flows through and then evolves under these operations now the corresponding code is on the left and Alan Geller and his team back in Redmond who really designed this language as I mentioned earlier to you know draw from the best concepts of languages like C# F# and Python and then of course bring with it the new concepts of quantum computing and so here in the code we're really performing the operations from the circuit shown on the right and this allows us to send information so teleportation allows us to send information from one qubit to another qubit even if those qubits have been taken across the universe you will still get information from the person holding the one qubit to the other person holding this other qubit and so that's the code shown here it's really great the hello world equivalency in quantum computing is actually teleportation right we've moved on from the normal hello world out of applications that we might be building so quantum computing in terms of all the power that can do and all the things that you can do simultaneously the syntax looks familiar all the things that we'll be able to do in the future what can we do in terms of going beyond that simple hello world scenario that we've just shown here right so to go beyond hello world we have to start to bring in some of the key concepts behind quantum algorithms and in hello world it actually uses what's called entanglement and superposition to send that message so when you design a quantum algorithm the first thing you do is prepare a superposition so earlier we talked about the ability to have 0 & 1 simultaneously you prepare all the qubits in a state so that they're all in this superposition state then we typically build entanglement in the circuit in the quantum operation in the quantum algorithm by performing quantum operations and then through this ultimately we hope to yield quantum speedups from the quantum algorithm now entanglement that we use say in teleportation and all of our quantum algorithms entanglement is when you take two qubits so initially they're separate and then we perform a set of quantum operations on these two qubits and what it does is it intrinsically correlates the information between these qubits so Jeremy if I give you one qubit and then you go across the universe and I do something to the qubit I have it will automatically cause something a reaction a state change in your qubit even though we didn't talk and and we're separated by the universe and we're separated by the universe right so from the compute perspective I guess that means that we are able to perform operations and perform them on all the different entangled qubits at the same time right yeah so it really gives you this amazing correlation this amazing parallelism inside your quantum computer now with that power we can explore very complex problems that take you know lifetime of the universe or billions of years classically and we can get solutions in only a matter of days or hours on the quantum computer but there's another key ingredient that you need to achieve those solutions and that's called interference quantum interference and it's like you know when you when you touch a tool right and you create ripples and these ripples send waves so remember in quantum mechanics we have this wave particle duality and and so think of just the waves the waves ripple across the pool and they start to interfere with each other now some of these ripples grow right the amplitude of those waves gets larger and some actually disappears some of those ripples and that's called constructive interference when it grows and destructive interference when it when it disappears when when those waves collide and you don't see them anymore and what we do is we really program the quantum computer with our algorithm design to cause the waves that we you know the solutions are actually encoded in these waves and we hope to increase the height of the waves for the ones that encode the true solution and then cause you know destructively interfere the waves that don't encode the solution and then ultimately we have to measure and we measure the quantum computer and what this does is gives me out one of these states namely one of these waves with a probability that's proportional related to the height of that wave so a quantum algorithm is actually a probabilistic algorithm you may have to run it a couple of times to get your solution with high high fidelity but you know ultimately you're programming this this tidepool almost right these ripples in your quantum algorithm and it's it's a true art to be able to do this well but when you do you really get some powerful speed ups and in a way it's it's a bit like signal processing right you know how you want to filter out some signals and then amplify others it totally looks like it so you actually want you're using basically this is a filtering mechanism again at that qubit kind of subatomic level as we're creating these waves so you can look for in this case the constructive interference when you can so I know a lot of this sounds very very futuristic to a lot of people but how far are we away from seeing true quantum computing well I think we're closer than you than you think you know we're we're already engineering these systems right and we're already developing the software as you as you as you can see today right we already have code that's written and we're releasing these tools to the developer community now quantum computers will truly be revolutionary right they're gonna allow us to solve the world's most challenging problems which is super exciting so do you see quantum computing replacing some of the transistor based computers that were used to no well so quantum computers won't replace all of the computers we have today right they're really they're really meant to be used for a certain class of problems problems in material science chemistry AI machine learning so you won't see them replacing all of your all of your classical computers it's not going to replace you know your computers do email or word right but what a quantum computer is really powerful at is modeling nature right so with a quantum computer we can start to get a better understanding of processes in nature and knowing just how powerful nature is I think it's going to be an absolutely amazing future right where we we can learn some pretty awe-inspiring solutions from the quantum computer by really learning more about nature and what it's doing around us and and then we can use these things to help optimize models and machine learning artificial intelligence health you know materials solving big challenges like global warming and clean energy it's pretty exciting and I think the sky's the limit we're all learning how to use this quantum computer to help solve these problems and problems we haven't even thought of very cool some of these problems might take a billion years to be able to solve with current compute but how can people that are watching today explore more of the technology for themselves and really learn more about quantum computer well we're really hoping to build a community around the development of quantum applications right this is this is new to all of us in terms of what we'll be able to do with a quantum computer and we need we need help you know we need people's help and insight into what what should we use these for so I encourage developers and engineers to really join our Microsoft quantum community that we have released today at Microsoft.com/quantum and there you can sign up and gain access to the language our new programming language that's embedded in Visual Studio you can get access to the simulator our debugging tools and then also a whole selection of samples and tutorials to get started including the quantum hello world example but also more detailed examples around quantum chemistry and how you use quantum computers in that space and so on so there's there's lots to get involved in and we're really excited to have a development community grow around quantum computing and the software tools themselves will be available by the end of the year for free but you can sign up now to get access and insight and you know insider information so sign up today so really great to see Thank You Krysta for helping demystify some of the mind-bending topics are on quantum computing you know you've seen entanglement you've seen superposition all these core topics are on there hopefully everybody here got a good idea of what what we're doing and also we've just scratched the surface so I'm really excited to see what's going to come next that's about all the time that we have for this episode of Microsoft Mechanics Live! and we'll see you next time [Applause]

18 Comments

  1. Deep Learning Partnership said:

    Great.

    June 26, 2019
    Reply
  2. Kitten Gamers said:

    Great for people beginning to study Quantum computing and people teaching themselves Quantum Computing(such as me).

    June 26, 2019
    Reply
  3. NOVEL VIEWS said:

    What if something happens to q2 and q1 at the same time?

    June 26, 2019
    Reply
  4. Dick Stephenson said:

    Is the information transfer between entangled particles instantaneous?

    June 26, 2019
    Reply
  5. crashonthehumble said:

    Thank you for this upload

    June 26, 2019
    Reply
  6. Dariusz Kajtoch said:

    How do you know the topological qubit is so robust against noise, if it wasn't obtained experimentally? It is pure theory.

    June 26, 2019
    Reply
  7. NuNu Ding-Lam said:

    wonder can we write assembler language or low level c language

    June 26, 2019
    Reply
  8. BetterSynthesis said:

    what are the computational limits of the topological qubit? is it limited to annealing and optimization problems like the dwave or does it open the full theoretical spectrum of q computing?

    June 26, 2019
    Reply
  9. Naimul Haq said:

    The binary states of 0 and 1 can be divided into many qubits, storing and computing many problems simultaneously. Our brain employs QC resulting in simulation of intelligence and consciousness.

    June 26, 2019
    Reply
  10. Walter Padelford said:

    Thanks!!

    June 26, 2019
    Reply
  11. Thiago Benazzi Maia said:

    Cool, looking for more videos -)

    June 26, 2019
    Reply
  12. iLikeit_ said:

    10^300…. dammit man. Amazing.

    June 26, 2019
    Reply
  13. American Citizen said:

    All guys in the audience?

    June 26, 2019
    Reply
  14. ELFATIH SHORA said:

    amazing …so the gen Sequencing will be solved in fraction of second ohhhhhh , this will revolutionize health care and this what I was thinking about long long time ago , well done

    June 26, 2019
    Reply
  15. Kempton Lam said:

    Big thanks to Dr. Krysta Svore (and Jeremy) for a very informative session. I've added this video to my crash course list of "Quantum Computing" videos that I've been listening/watching in loops! I'm currently scheduled to attend some lectures, so I'm seriously brushing up on my QC stuff to make myself look less like an idiot than I normally am. πŸ™‚
    P.S. As a video that built as an intro about Quantum Computing plus MSFT's contributions to the field, may I be frank and make a suggestion? Part of me thinks like a computer scientist (instead of an MBA) and feels Krysta could/should have mentioned those working QCs. Yes, I mean those D-Wave 2000-Qubits Quantum Computers just for the sake of completeness. Correct me if I'm wrong, when MSFT/IBM, etc etc are years away from working QCs, not mentioning working Quantum Computers actually exist NOW, to me, kinda turn this otherwise excellent informative video into merely a MSFT marketing video. I expect more from MSFT. To me, science, computer science and facts must come before marketing.

    June 26, 2019
    Reply
  16. Larry Castro said:

    All this lingo about quantum computing is beyond me but fascinating to hear all about it.

    June 26, 2019
    Reply
  17. jarnMod said:

    After session quiz.
    Name the 3 breakthroughs explained in this session.

    Look, my point is, this is a good session but I think it does not take into account how people consume info on the medium it was advertised on. In the end, I learned more about quantum mechanics than Microsoft contributions on it. May be you should just provide a list at the end. You know, remind the viewer about what is great, and what is special.

    June 26, 2019
    Reply
  18. Ingram Leedy said:

    Great simple explanation of how quantum computers work vs classical computers.

    June 26, 2019
    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *