What's Next: The Future of Quantum Computing

hello i'm dario gill and i'm speaking with you from the yorktown heights research laboratory it serves as the headquarters of ibm research and today is a great pleasure for me to share with you a perspective of the quantum computing era that is emerging and the implications that it will have for science for discovery and for business at large it is clear that the world is being challenged with ever more complex problems to solve the coronavirus and kovite 19 is a powerful example of the challenges we confront in humanity as we try to understand some of the deep functioning of nature and the implications it has for all of us in our societies here in the context of dealing with a structure that is no more than 100 nanometers inside we're struggling to figure out ways how to deactivate the functioning of the virus now we know that computation can help us if we look at the inner structure of the virus we see these spiking proteins and these proteins can be modeled and understood and therefore we could use them to then explore different kinds of compounds that could deactivate how they function here supercomputers can come to help us we can use these most powerful computers to model this protein structure and we can use another powerful force which is the powerful force of joining the resources and the capability across institutions in moments of crisis it is particularly important that we all come together and that's why we were very proud to coalesce and create in partnership with the federal government and other institutions in the tech sector as well as academia the kovit 19 high performance computing consortia where 30 plus members have aggregated over 400 petaflops of computing power and over a hundred thousand nodes to go and pursue a broad portfolio of projects that can help us understand the pandemic understand the evolution of the virus and accelerate the pace at which we can develop antivirals and ultimately vaccines what this tells us is that when confronting with exponential problems we have enormous computational demands in fact exponential problems require exponential computation and today we're fighting the pandemic within the context of the computational paradigm we have the world of bits but there's another computational paradigm that is emerging and is the world of cubits so moving forward bits and qubits will be foundational computational paradigms with which we can tackle complex problems in the world so let's try to understand the difference between them and where the power of qubits rise if we look at the left hand side it's a way to describe and we're going to use these diagrams throughout the presentation to unpack the power we're seeing a circle and two dots when the circle when the dot is in the north pole that's a zero when it's in the south pole is a one that's a classical bit to state now a cubit is the unit of information in quantum world and here we get to have let's depict it as a sphere a zero and a one and we can have them in this superposition state but we have a special trick that is available to us not only we can have them positioned in this sphere but imagine that each one of this now is like a little moon that is in the sphere and we can rotate the moon so here you're seeing that um the bottom sphere has been rotated slightly so that's why you see a little bit of blue and pink and we could rotate it all the way so here's a 180 degree shift so we have this notion of phase quantum affords us three super powers that are exhibited in the qubits they are the powers of superposition the power of interference and the power of entanglement so let's unpack those the principle of superposition is actually quite straightforward to understand so let's start on the left-hand side we have our north pole zero or south pole one and if we add those two states now we have a superposition qubit that has both a zero and a one simultaneously north and south pole now the right diagram is the same but now we're gonna do a little trick we're to apply a gate a quantum gate and what we did here is remember we get to change now we get to rotate these little moons right so we've rotated it 180 degrees and now we have a 1. so now when we do the addition of 0 plus one we end up with a superposition state but notice that it's different than the one on the left we have still a zero and a one but now with a rotated state for the one now that is going to be very powerful to then leverage a second really interesting thing which is the principle of interference and this is not an operation that we get to perform in classical bit world so here in this example we have a zero and a one on the left plus this new state that we created which is also a zero now one but with this rotated moon when we do this now we get to interfere those two states and notice something really interesting happens we've cancelled the one state and we've ended up with that zero state we've done that because we've interfered the states so the next opportunity that we have right now is how many states can we create and the beautiful thing in quantum is we get to create an exponential number of states so let's look at a system that has five qubits in this case we're depicting these qubits as independent qubits right you're seeing them on the left zeros and ones and they're getting multiplied we have five of them so the number of states that we can create now in our quantum computer depicted on the sphere on the right it's two to the power of the number of qubits we have in this case we have five so two to the power of 5 is 32 states so we're seeing 32 of those little moons notice that in this case we did not incorporate any phase they're all pink the powerful aspect of using this property called entanglement in the quantum world is that we can create now states like the one depicted on the right where we have these combinations of pink and blue where that end state cannot be described as the independent product of these five qubits in this case since we have 32 states what this tells us is actually something very profound and that is that you could have an n state in your quantum system that you're using to process information that cannot be described as the independent components of the qubits that make the whole state there's a lot in there it's a very profound statement and it is the basis of the power and the complexity of quantum machines so let's bring that all together to give you an intuition for how an algorithm works in a quantum computer what we do is that at the beginning let's say in this case we have a five qubit quantum computer is that we prepare the computer to be in a superposition of states 2 to the power of 5 32 you have those 32 dots that you see on the left in pink then we have to encode the problem we have to inject data into the quantum machine the way that happens is that that encoding gets done through entanglement so notice that what we've done now is that some of those states are now changing phase you're seeing those depicted as blue and some are pink now in our machines not only we have the superposition we have exploited the property of entanglement to do the encoding of information and the principle of interference comes from the fact that we can now take these states and combine them and interfere them with one another in such a way that we get to cancel things out and maximize the right answer so many things fall away and the right answer gets maximized we perform a measurement and we get a result notice how very different that is from classical bit based operations and there is a really important relationship between this entangled property and the amount of information that we can process and this relationship is exponential what this table shows is the number of classical bits zeros and ones that are required to represent that complex single entangled state that i just described and here's an amazing number by the time you have a hundred perfect qubits that are entangled with one another if you needed to describe them using zeros and ones you would need to devote every atom of planet earth to store those zeros on ones clearly that's not possible in fact by the time you have 280 cubits you will need every atom of the known universe so that it tells you that this is exponential relationship between the special property of entanglement and the representation of information why does this matter well it matters because it turns out that despite how powerful or classical computers are the reality is that there's a vast set of problems in the world that they cannot tackle from an information theory perspective we would say that really classical computers can tackle easy problems things that don't have an exponential number of variables in them remember in the case of the virus as an example that molly nature is an example of an exponential problem but so are other problems like factoring and problems in optimization in fact there's a vast number of problems in the world of mathematics that have that character and those problems are deeply important for business the world over for optimization and in the future and machine learning and developing new molecules and chemistry and the life sciences and beyond now the reality of it is that you say well that's very interesting and the theoretical underpinning but can you actually build these machines and access these machines and we also recognize that for most of history to access computing power you had to build it and maintain it yourself but now we're in a situation where even these very special machines quantum computers can be accessed by anyone in the world in fact the system looks as follows you can sit in front of your terminal write your program you can send your zeros and ones and when they get here to the laboratory where we are as an example we convert those zeros and ones to microwave pulses they operate about five gigahertz and we send them dino cryostat that operates about 15 millikelvin temperature and we get to use those microwave pulses to manipulate these qubits to perform those superposition and entanglement and interference operations and then return the result to the user who is using a regular classical computer this is what the inside of these beautiful machines look like i say beautiful because they're really you know gorgeous pieces of of engineering and science and you're seeing the inside of this cryostat uh you know that uh golden chandelier and those wires are the means by which uh you know we send those microwave pulses down at the very bottom of that golden chandelier is where we have the quantum processor itself where we perform those operations so if we go and look inside this is what a quantum processor looks like in our case in ibm we use superconducting technology specifically we use a qubit device called the transmon qubit and what you're seeing is uh you know at the core of it if you look at the bottom right hand side we have a device which is called a joseph junction that is about 100 nanometers by 100 nanometers you know roughly the size of the the corona virus in fact so 100 nanometers by 100 nanometers and essentially allows us to create an artificial atom with a ground state and an excited state that is the basis then uh with which we can couple it to other qubits and you're seeing on the right hand side those wiggly lines that we've labeled microwave resonators and that's what allows us to couple the qubits with one another and then perform these entanglement and interference operations that give quantum computing its power we build these quantum chips here in the laboratory from which i'm talking to you today in yorktown heights now with those units of information the very important thing that is being created is a rich and vibrant world of quantum circuits this is what's really next in the world of quantum so let's explore them a little bit a quantum circuit is going to be the unit of value for your business the qubit is a unit of information but a circuit is the unit of computation so let's try to understand what is a quantum circuit and let's look at its anatomy this is what a quantum circuit looks like so let's decompose this into pieces the qubits are depicted by these horizontal lines in this case we're depicting a circuit that has four qubits that you see here shown now the gates are how we control the qubits and you read the gates from left to right so it's the sequence of placing these gates that is the basis of creating a circuit to construct this circuit we have many quantum gates available to control these circuits so this is analogous to the way we do controls in the world of bits so in the world of bits we're all accustomed to that you know behind the scenes we have these ands and ores and knots etc so similarly in the world of qubits we have a different set of operations remember some of those operations are the very idea of doing the rotation to one of our little moons that i was depicting before right introducing a phase change and i showed before a z operation that performed that rotation that allows us to do that interference operation so let's do a comparison between a classical circuit and a quantum circuit here in a classical circuit what you witness here is that you don't get to see blue moons right there's no rotations everything here is pink and that you've got these single states that in this case were seen moving the sphere but notice what a contrast is going to look like when we actually implement a quantum circuit so let's look at what a quantum circuit looks like so in this case we're implementing grover's algorithm which is a basis for performing search operations first of all you notice the complexity has increased greatly right we have these superposition states now you're seeing that we get to perform these operations where we do rotations so here you're seeing blue we're performing operations of interference you're seeing some of those little moons getting bigger and bigger in size that is telling us we're converting to some important answer and that is how we get finally to to the answer so if we put them side by side just visually you can get a sense that a classical circuit can never exploit the vast computational space that a quantum circuit can it's it's the visualizations give you this intuition of the richness with which we can access this large state and the beautiful part of this is that we can have many classes of circuits to do different functions so for example this is a circuit that is designed to model financial risk here's an example of a circuit for the world of chemistry right to measure the energy of lithium hydride that is important for the development of new battery technology is a circuit that is a unique form of a classifier a new approach to do machine learning applications so what we're very excited to share with all of you is the fact that libraries of these circuits are emerging and that there's going to be this richness of circuits that allow you to perform different applications that we are going to be able to embed in the programs that we all use and create to create business value libraries of circuits that get embedded in classical programs that now combine the best of classical programming and the best of quantum power it is important to realize that not all circuits have equal value and that there will be circuits that you can implement classically far far better because if you're not exploiting that computational space if you're not exploiting these ideas of superposition and interference and entanglement in the proper way you don't get there now we're very fortunate that some of the best minds in the world are thinking really hard about this problem about what are the classes of circuits that give us a path to quantum advantage and i'll highlight this seminal paper that was published last year in science by sergey bravi and his collaborator now sergey works here in the yorktown lab and he is absolutely one of the top theorists in the world in quantum and this paper what uh him and his collaborators shown is the fact that you can have shallow circuits meaning these kind of quantum circuits that don't have a massive amount of depth to them that have a provable theoretical advantage compared to the classical implementation this is the first time where we have a mathematical proof that these classes of circuits have proven advantages compared to classical ones so guided by these theoretical insights that are so profound it is how we are chartering the path of creating these libraries of circuits and this is how you consume that you're not going to need to learn a new quantum programming language you can use your favorite language and from there call upon these circuits that implement these special functions from there map them to the right quantum systems that we have available on ibm cloud and then return the answer back to you as the programmer now i know you hear that some are telling you that you got to learn you know a new quantum programming language we believe that we need to exploit the richness of all the programming languages that we have today whether it's python or c plus plus or whatever really you love and that circuits is what will provide you the quantum value and behind the ibm cloud you will get to benefit from both the library and a large portfolio of hardware designed to implement those circuits efficiently this is not an aspirational future we really have many quantum systems on ibm cloud here is the growth of the number of quantum systems we have made available on ibm cloud and to the community since we launched the world's first service in may 2016. you can see that at present we have 18 quantum systems that we've made available through the cloud so let's go take a look at a few so that you can see what they look like so welcome to uh one of our quantum computation centers here in yorktown uh remember i showed you that golden chandelier before uh where you saw those wires that went down all the way to the quantum processor well this is what a quantum computer looks like a superconducting one when it's encased we have to protect it for the environment so that's why it's shielded and all the way at the very bottom here is where the quantum processor sits and as i mentioned this is one of the coldest places in the universe right down here 15 millikelvin let's listen for a minute to the sound of a quantum computer you hear like this chirping sound tintin tintin tintin that is linked to the flow of gases that we pump through the system to be able to extract energy from the system and cool it to the very low temperatures let's go take a look at some of the other key components of a quantum computer in this case is the control electronics here this custom-made control electronics is how we convert the zeros and ones to microwave pulses at about five gigahertz and we send those pulses into the quantum computer to perform the quantum operations and to execute our quantum circuits so let's just quickly take a final look and some of the other systems here are the four systems out of the 18 that i mentioned to you and let's take one last look and one of the other ones this is one of my favorite places in the world right it is so wonderful to see a whole new paradigm of computation and one of the very first quantum computation centers that have ever been created and this is how the community has grown since may 2016. what you're seeing here is a visualization of the number of quantum circuits executed in the ibm cloud quantum portfolio and look how it has grown over the years over 180 billion quantum circuits have been executed with over 230 000 users all over the world if you look now from an institutional perspective of companies and universities who have joined the ibmq network by far the largest community of institutions collaborating and exploiting the power of quantum computing you can also see the fantastic growth and the excitement that is behind the scenes and what are the cases that are being explored by this community of institutions and businesses well a broad variety of problems ranging from chemistry and developing new materials to optimization to artificial intelligence and to scenario simulation let me give you just a few examples daimler uh is a wonderful partner in the ibmq network and in their case they're very interested in the power of quantum to look at advanced battery materials and the possibilities of using quantum for manufacturing defect inspection in this case i'm highlighting on the bottom right the example an example of one of the circuits that are being used to perform these use cases or let's take out the case still still staying in the field of chemistry and materials of exxon mobil in this case they're very interested in uh you know modeling the the thermal reaction modeling for petrochemical r d and here we're seeing some of the example uh of the circuit that is being used to perform these calculations and by the way we develop these circuits through a collaboration right between ibm and and our partners as as we pursue this work let me give another example for a different industry this is in the case of the you know our wonderful partner j.p morgan chase where we work on the problem of optimization and here we're looking at the potential to find better solutions to more efficiently than what we can do classically to improve things like options pricing and fraud detection or portfolio optimization and you're seeing here an example of circuits that have been developed for european derivative pricing now when you look at the capability today i think it's also a very important question to say well what is the road map you know how much better are these systems and how much more complex are going to be the circuits that we're able to execute and for that we have developed a very important metric called quantum volume sometimes you may read or hear in the popular press the number of qubits and you know and as a measure of progress i'll tell you straight that just the number of qubits in the machine is not a good measure of the power of a quantum computer there's another axis that is vitally important and that is the error rate in these qubits these qubits and these cubic machines are not perfect today and they have errors that are introduced by the coupling of the system to the external world the very power of these qubits of entanglement and those properties are also make them both powerful and delicate to the intersection and connection to the outside world so you need to lower the error rate while you increase the qubit count so that's what's depicted here what you really want to do is you want to move in the diagonal and quantum volume is a measure that incorporates both of those metrics and some key additional considerations and the good news is that we are in a new exponential here and what we have demonstrated experimentally and by building systems that we have deployed to all the clients of the ibmq network is that we are at least doubling quantum volume every year we've gone from a quantum volume of 4 to 8 to 16 to 32 on a yearly basis and we are committed to at least double it right year-on-year basis if we keep this space or even faster we are going to see really spectacular results with quantum computing let me close with a reflection on the urgency of science the urgency of tackling some of the most complex problems in the world and i'll put that reflection that a revolution in discovery and a revolution on mission critical applications is going to be power by the most exciting time in computing in 60 years where what we're witnessing is the convergence of bits neurons and qubits bits the classical world that we've all known and the computers we've known qubits this new fundamental way to process information that we explore today and the world of neurons and neural architectures that are the basis of artificial intelligence it is not that one will eat the other the most profound implication of what is happening today in computing is the convergence of bits neurons and qubits now this convergence is going to be orchestrated through a hybrid cloud architecture and on top of it to mask the complexity of the underlying infrastructure we are going to be assisted by artificial intelligence in the way we program ai assisted programming the consequence of all of this coming together will be nothing short than a revolution on how science itself is practiced and the rate at which we could perform accelerated discovery and a whole new class of intelligent mission critical applications thank you

Channel: IBM Research

Views: 136,391

Rating: 4.9010563 out of 5

Keywords: IBM, IBMResearch, quantum computer, quantum computing explained, quantum computing 2020, quantum computing for computer scientists, quantum computing, quantum computers, quantum computing, how quantum computing works, how quantum computing will change the world, why quantum computers are faster, future of quantum computing quora, quantum computers for dummies, future of quantum computing, future of computers, quantum computer explained

Id: zOGNoDO7mcU

Length: 28min 14sec (1694 seconds)

Published: Fri May 08 2020