Your Guide to 100 Qubits: Quantum Computing in Practice

hello my name is Olivia Lan and I am a researcher and an educator at IBM Quantum this video marks the launch of a brand new series on the kkit YouTube channel and our learning platform called Quantum Computing in practice in short in this series I'm going to show you how to use a quantum computer that exists right now to its full potential there are billions of videos already on the internet about quantum computers and their future potential or on theoretical Concepts behind Quantum computation which are also just as important see understanding Quantum information and computation

hosted by John watus for example but in this course we are going to assume the learner maybe knows a little bit more than we have in the past and we will go very far technically speaking and that's because over the past year and at the time of this recording it is 2024 we have seen a shift a phase transition if you will into a new era of computation quantum computers are finally approaching the point where they can be useful even though experimenting with smallscale circuits and quantum computers was necessary to educate our community on how

to utilize these machines no one was running experiments that couldn't also be performed on regular old classical computers but as of today we have entered an era of quantum utility and what is utility you might ask it is different than Quantum Advantage which can be understood at the point at which quantum computers are outperforming classical computers at a meaningful task in instead utility means that they can start to compete with one another and this is the natural next step in the technological Evolution IBM published a paper that showed for the first time last year a

quantum computer could be used to simulate a physical model and compete with state-of-the-art classical simulation techniques it even bested the Brute Force attempts and let researchers have a new Avenue that they can use for research so why is this all happening now well in the past few months IBM has made Quantum processors of over 100 cubits widely available if you're already a user you might have noticed that we took down a lot of our smaller processors to make way for these larger chips and we felt like it was really important to do so because some

numbers of cubits can still be simulated classically without a problem so there's no real advantage that can be gained by using these chips however at 100 cubits this is no longer the case they cannot be simulated classically it is also important to note that there are these things called simulators which are essentially computer programs and they can mimic some behavior of quantum systems including noise however please don't miss the fact that I said mimic simulators are not quantum computers they cannot produce the results that quantum computers can at this scale but they are still an

important part of our testing workflow so that was a very lengthy answer to the question at the top of this slide what is practical Quant Quantum Computing in this course I'm going to teach you about realistic potential use cases for Quantum and the best practices for running and experimenting on these larger scale processors today so who is this course aimed at let's begin by doing some level setting at the beginning this course is for everyone that aims to develop new applications for quantum computers wants to know how to scale up their current work or learn

how to use quantum in their existing workflows I previously mentioned that this course will be very Hands-On and we won't be discussing theoretical aspects nearly as much we expect you to come in with a basic understanding of what Quantum computation is but if you're not quite there yet no worries please just watch the first few episodes of John waters's course and you'll be ready we also want to make sure that folks watching know that this isn't just a course for physicists or even computer scientists these two subjects have really been hogging Quantum Computing for the

past decade or so we want to make sure that others who work with big data or simulations such as Engineers chemists and material scientists know that this course is also for them in summary this course is directed towards a broad audience and we hope that you'll join us for its entirety upon completing this course you will be able to understand how to create and run a utility scale job through kuit runtime including the entire workflow you will be able to understand the main types of error mitigation and also understand likely application areas in the short

term you will also be able to per all of these algorithms in kit and lastly we expect you to be able to understand noise and how to deal with it so let's get started Quantum Computing is an exciting new technology in an early stage of development but it's just one chapter in the story of computation which goes back thousands of years I think it's important to understand this context because it makes what we're aiming to do here a little bit more intuitive predictable and less sci-fi since ancient times we as humans have needed to perform

computations or in other words process information according to certain rules and constraints in order to enable communication construction Commerce science and all other aspects of Our Lives we've looked to the physical world for assistance and through ingenious discoveries we've constructed devices to help us compute long ago devices were made from wood bone knotted ropes and they stored information and facilitated calculations much differently than we do today Mechanical Devices built from levers gears and other Machinery advance from early astronomical clocks to calculators to sophisticated Computing devices such as differential analyzers that solved equations using wheels and

rotating discs even the technology of writing has played an important part in the story by allowing people to perform computation that they wouldn't be able to otherwise when we think about computers today we tend to think about electronic digital computers but this is actually a fairly recent technology electronic digital computers were first built in the 1940s the earliest known tool actually for computation is the Sumerian Abacus which experts think was invented in Babylon in 27 to 2300 BC that's a really long time ago but we have come to understand more and more about the natural

world from Archimedes to uid to Einstein our needs for more computational resources has exploded today we have supercomputers which are basically many very powerful computers hooked up in parallel they are one of the best tools that mankind has for solving difficult problems but there are still plenty of problems that even these behem of computers will never be able to solve more on this later so we've established that humans have been using physics and nature to help us compute for thousands of years and to help us with our NeverEnding quest to compute more and bigger things

you might already be familiar with Moore's Law it's actually not a law so much it is an observation the law part is a little tongue and cheek in my opinion the max number of transistors in an integrated circuit basically a computer chip doubles about every 2 years over the past 5 decades or so we've not only observed this but have made great use of it with more transistors we can perform more faster computations this is essentially what is responsible for making computers more and more powerful and smaller Over time however there's some bad news Mo's

law law is dying experts disagree about when Mo's law will come to its eventual end but some say it already has although we may not be exactly sure when we will reach our limit we know that it is eventually inevitable because some transistor Parts cannot be made physically any smaller you can't make a transistor that is smaller than an atom for instance this is of course completely unphysical sounds silly but those are the limits that we're actually working with so what will we do just give up and say hey that's as good as it gets

no humans have never done that luckily quantum mechanics has entered the conversation quantum mechanics was discovered in the 1920s and since then the world has literally never been the same and we could spend hours talking about the revolutionizing discoveries that were made but that's not really the focus of this course instead I want to make it clear that we know that quantum mechanics and Computing are already compatible with with one another the findings and understandings of nature that we learned from quantum mechanics have in part made our modern-day computers possible without quantum mechanics we would

not understand or probably have even been able to invent the solid state hard drive for example Quantum computation is simply the next natural step in this Evolution but computation doesn't just exist in a vacuum information requires a medium computing information harnesses the physical world and it affects it as well this places limitations on what is even physically possible but understanding nature and the physical limitations is actually the aim of physics as a discipline so in a way physics and computation have always been intertwined Ralph Landau a computer scientist and an ibmr recognized decades ago that

information is physical putting it another way information is not an abstract entity but exists only through physical representation landow's principle establishes a connection between information and the laws of thermodynamics but in fact there are many connections between physics and computation classical physics was all we were able to explore up until recently but with the discovery of quantum mechanics we are now on the precipice of being able to create an entirely new way to compute and this is great because classical computers can only approximate what actually occurs in nature and in some cases that approximation is

very limited that's because it's really really hard to get machines that follow the laws of classical physics to approximate the laws of quantum physics and nature is not classical it is quantum one analogy that might help you understand is a wind tunnel fluid dynamics are notoriously hard to simulate and predict mathematically even something as mundane as a car driving through some wind car manufacturers found that it was actually too costly and impractical to try to simulate this physical interaction when they were designing cars instead what they ended up doing is creating a literal tunnel that

blows wind through it and just drive a car into the tunnel and see how it performs this is analogous to what I mean when I say classical computers only poorly approximate nature and sometimes it's too costly to even attempt to have them do it instead of simulating the wind researchers just recreate it similarly quantum computers don't have to approximate the laws of nature on a molecular level because they follow those laws intrinsically they don't have to simulate anything they emulate it and the same difficulties arise when we're simulating batteries or chemical reactions which leads us

to our next section Quantum Computing what is it good for so why would we even want to think about combining quantum mechanics and Computing well there are already plenty of problems that are too difficult too computationally costly for even our cuttingedge supercomputers to solve maybe these are problems that only researchers are super focused on but their impacts could touch all parts of the globe these problems that classical computers get stuck on are things that we as a Society are becoming more and more reliant on technology for and really care about like battery efficiency Financial modeling

and optimization and even how to build better planes so why turn to quantum computers well we have very good reasons to think there's a lot of potential here and we can open doors to new computation that is currently Out Of Reach for even the most powerful supercomputers in theory some Quantum algorithms provide advantages over the best possible classical algorithms for some applications for other applications there is mounting evidence that even with some noise and Imperfections quantum computers can still be used as a tool to compete with classical brute force solutions basically it is a brand

new tool and we want to figure out even more cool things we can do with it but we can also look to Fan's original idea for them which is to use them to understand physics and nature in and of itself after all I already told you that nature is quantum not classical and this field is incredibly Rich fascinating and interdisciplinary it brings together physics computer science information processing and of course the specific domains of application areas that we might be interested in however unlike classical computers not everyone has a quantum computer sashed away in their

back pocket until very recently if you wanted to experiment with a quantum computer you had to build and maintain one yourself usually in a sad basement lab in a university or research facility and trust me I would know no longer is this the case however in 2016 IBM put the first Quantum processor on the cloud it had only five cubits at the time and fairly High rates of errors but that was only 8 years ago and look at how far we've come so where do we stand today there are a lot of parallels to be

drawn to the very early days of the modern computer like I mentioned this is a very hard Challenge and it was only within the last 8 years that quantum computers became widely available besides for those who are working in a Cutting Edge lab or facility but in those past eight years we have made progress on every front processors are orders of magnitude larger in Cubit size gates are far improved and we have introduced methods of reducing or mitigating errors intrinsic to the quantum systems even while we push forward to create fault tolerant Quant quantum computers

now you and anyone else can use one of these numerous IBM Quantum processors that are hosted on the cloud almost all of which are the newest latest and greatest systems with over 100 cubits but isn't just the size of the processors that is important this is only one metric that we care about and that makes sense right if we have more cubits that's obviously better provided that increasing the number of cubits doesn't degrade performance however this isn't always is the case we need more good quality cubits that don't interfere with one another through cross talk

when we don't want them to how the cubits are connected to one another is also something important to keep in mind in superc conducting circuits it's a challenge to figure out how to best physically connect the cubits another hugely important metric is two Cubit gate Fidelity Gates that run on single cubits are not as prone to airor two Cubit gates are the much bigger area of concern and usually one of the go-to metrics that we look for to see how performance is improving over time two cbid gates are crucial to get right because they are

responsible for creating entanglement between the cubits which is one of the physics phenomena that gives quantum computers their power lastly we look for Speed and efficiency we need to make sure that the time spent running a program on the quantum chip is executed as quickly and as efficiently as possible we refer to these three metrics together as scale quality and speed all of them are vital to improve performance now I want to show you some quick plots to back up what I've been saying about how the technology itself really is evolving quite rapidly the cubic

count itself is the easiest to understand I think and IBM has published a road map of where we hope to be in the next few years in terms of Cubit count and processor size for 2024 when we're filming this video we are right on track with our processor herin and Condor next quality this plot shows the median entangling gate error rate for IBM Quantum processors over time keep in mind that this is a log scale so while the progress looks vaguely linear it's actually much more rapid than that lastly we have speed this graph shows

the progress of a metric that we call kops which stands for circuit layer operations per second crucially kops encapsulates both the time it takes to run circuits and the required real and neartime classical compute basically kops is a measure of how quickly our processors can run specific circuits in series acting as a me measure of holistic system speed and incorporating Quantum and classical Computing right now at the time of recording IBM is the only Quantum cloud provider that is providing access to 100 plus Cubit machines as I mentioned a few slides ago 100 cubits or

right around there is where classical simulation begins to totally fail so large scale jobs being run on these machines often have no classical equivalent it isn't possible there is some nuance and some tips and tricks you can play to simulate some parts of a circuit and then other parts separately but in general we know that right around this number is where quantum computers can begin to do things that classical computers just can't so IBM decided that giving access to these larger devices was super important for the wider Community because anything being done on smaller devices

is classically simulatable and so it can't lead to Quantum Advantage the 100 Cubit devices we have now are still flawed and they're still noisy so we need to be smart about how we R run jobs and circuits on them naively just writing our circuit and pressing shift enter probably isn't going to lead to anything spectacular but again there are tips and tricks that we can Implement and best practices we can follow to squeeze even more performance out of these devices which will hopefully lead to Quantum Advantage sooner even while we continue to work and build

up our fa fault tolerant machines as well this plot shows the simulation cost or the amount of time it takes to solve a problem on classical computers and quantum computers equipped with error mitigation and error correction so the classical computation line scales exponentially as the circuit complexity grows for simple circuits down here that doesn't matter matter but it's obviously bad as the complexity grows because this makes the simulation cost very expensive very quickly air mitigation scales better so when the complexity of the circuit grows this is where we want to be we are currently here

with classical methods and Quantum methods sort of on par with one another for some applications and error correction might be the most preferable and scales really nicely we're just not there yet in the meantime we will continue to focus on error mitigation and getting to this area of quantum utility as soon as possible and by making use of several error mitigation schemes which we will learn about in this series we aim to find application areas that can make use of quantum utility very soon and that is really what the heart of this course is we

have these machines we think they're capable of doing some really interesting research so how do we use them in this last section of the episode let's briefly discuss application areas for practical Quantum Computing there are three main branches that we think we're most excited to explore and find avenues for Quantum utility the first is simulating nature this has to do with simulating molecular or Atomic interactions and this is really important for areas of material science and chemistry and for reasons we discussed earlier we think quantum computers are going to be a lot better at this

than their classical computer counterparts next is search and optimization the main algorithm of Interest here is called qaoa a hybrid classical Quantum algorithm qaoa uses a classical computer to train and find Optimal parameters for a Quantum circuit and we'll hopefully be able to do this a lot more efficiently than either computer could alone lastly we have mathematics and processing data with structure quantum computers are really good at finding structure in messy data where you or I or a typical computer might not see anything Quantum machine learning holds promise in this area and we aim to

use this for tasks such as feature extraction and pattern recognition it really is a very exciting time for those of us that work in the field of quantum Computing for the first time in history we can begin to explore a region of computing that lies Beyond classical Computing those of you watching this might possibly play a role similar to the early pioneers of computing and they had no conception of what was to follow t J Watson of IBM even famously said he thought there might be a market for only two or three computers ever and

wow was he wrong we are all those early pioneers of computing but in this new field of quantum Computing and Quantum is often contrasted with classical Computing as something distinctly different from it and often in competition with it but my aim for this introductory lecture was to hopefully convince you that from an eagle eye view we can see Quantum Computing as simply the next chapter and a long story of computing it is our nature as humans to seek out new ways to compute and to harness the power of the natural world and to figure out

what to do with it and we've been doing this for centuries Quantum Computing offers us a new tool in this endeavor and it is up to us to discover how we can leverage the power it offers us in the next lesson in the series we will talk a little bit more about 100 Cubit systems and give an overview of kcit our qu software SDK and how to best interact with these machines right now thanks for watching and we'll see you next time