Be responsible for your own professional development journey.

2020 marked the launch of a decade of explosive growth for the enterprise quantum computing market. This disruptive technological innovation is projected to reach $9.1B in annual revenue by 2030 (vs. $260M in 2020). Although, global quantum technology research reports reveal near-term trends with the potential to disrupt global commerce, the quantum workforce is not yet established to meet the anticipated demand for this important industry of the future.

Our new flagship Quantum Computing Education - Workforce Development Program is designed to empower our community of lifelong learners with quantum technology industry knowledge for global impact. At IEEE, we are dedicated to advancing technology for the benefit of humanity through educational activities. We aim to serve professionals involved in all aspects of science and technology that underlie modern civilization.

2021 Program Portfolio

 

Upcoming Courses

Check back soon for more upcoming courses.

 

Courses On-Demand

Quantum Engineering: Photonics in Quantum Computing and Quantum Networking
Wednesday, 28 July 2021 at 12pm ET

View this course on-demand

Abstract
This masterclass will review how photonics play a central role in several of the leading candidate technologies for building quantum computers and quantum networks. We will discuss trapped ions, trapped neutral atoms, optically active defects and quantum dots in solid-state materials, and purely photonic approaches for realizing quantum processors. We will also review how superconducting circuits, which don’t natively involve optics, can be coupled to photonics. The emphasis will be on giving a broad survey of the various photonics-related quantum technologies and the current state-of-the-art in each.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Dr. Peter McMahon, Assistant Professor, Cornell University School of Applied and Engineering Physics (AEP)

Dr. Peter McMahonPeter McMahon is an assistant professor of Applied and Engineering Physics at Cornell University. His research lab investigates how to harness physical systems to perform computations more energy-efficiently or faster (or both) than conventional computers. He works on both classical and quantum computing with a variety of platforms, including photonics and superconducting circuits. Peter received his Ph.D. from Stanford University in Electrical Engineering and performed his postdoctoral work at Stanford in Applied Physics before moving to Cornell. His is a CIFAR Azrieli Global Scholar in Quantum Information Science and won a Google Quantum Research Award in 2019.

 

 

Automation and Synthesis of Quantum Circuits
Wednesday, 12 May 2021 at 12pm ET

View this course on-demand

Download presentation slides (PDF, 46 MB)

Abstract
This class will review current limitations of designing quantum circuits, typically done at the gate level or using specific functional building blocks; introduce automation and computer-aided design (CAD) technologies for quantum algorithm design; and demonstrate how these technologies unlock new frontiers of creativity in quantum algorithm development.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Amir Naveh, Co-Founder and Head of Algorithms at Classiq Technologies

Amir NavehAmir Naveh is the co-founder and Head of Algorithms at Classiq Technologies, an exceptional quantum startup that recently received significant venture funding. Classiq enables the development of quantum algorithms through automation and synthesis. Amir is a former leader of large R&D teams and projects in the Israeli Ministry of Defence and Intelligence community and a "Talpiot" alumnus.

 

 

A Hands-On Approach to Quantum Computing Learning - Comics and Coding with Q#
Wednesday, 5 May 2021 at 10am ET

View this course on-demand

Download presentation slides (PDF, 13 MB)

Abstract
This course offers a unique opportunity to learn about quantum computing through an intuitive series of comics. It will provide new learners a fun way to understand key concepts. Learners already familiar with the fundamentals will experience a new perspective on quantum computing algorithms. Q# will be used to construct Grover’s algorithm hands-on.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Dr. Kitty Y. M. Yeung, Microsoft Quantum, Sr. Program Manager

Dr. Kitty Y. M. YeungDr. Kitty Yeung is a physicist, artist, maker, fashion designer and musician based in Germany. She works at Microsoft Quantum Systems as a Senior Program Manager on quantum computing education. Kitty is the producer of MS Learn quantum modules and the Quantum Learning website with customized learning materials, creator of comic series Quantum Computing through Comics, lecturer at HackadayU and Microsoft Reactor on Quantum Computing, founder & designer of sustainable and STEAM fashion brand, Art by Physicist, and creative technologist & lead of the Fashion Hack at Microsoft.

Kitty worked as a research scientist, hardware engineer and user experience designer at Intel, and Manager of the Microsoft Garage program in Silicon Valley, California. She received her Ph.D. in Applied Physics from Harvard University (Thesis: Engineering Plasmonic Circuits in 2-Dimensional Electron Systems) and a M.Sci., B.A. and M.A. in Natural Sciences from University of Cambridge. Kitty's career has been focusing on physics while pursuing the integration of technology, science, design and art. Kitty frequently gives technical and career talks reflecting her passion and experience in quantum computing, wearables industry, digital transformation, and internal startups. See her work on www.artbyphysicistkittyyeung.com

 

 

A New Approach to Quantum Machine Learning
Wednesday, 28 April 2021 at 12pm ET

View this course on-demand

Abstract
In recent years, there has been an increasing interest in combining the disciplines of quantum information theory and machine learning. One of the approaches is to translate ML or DL to quantum equivalents expecting "QUANTUM" speed up in training and inference. The other approach is to explore a new direction of machine learning which utilizes the nature of quantum computing. Though we are seeing the early signs of quantum advantage (e.g analytical performance), it is still uncharted and needs to be explored. This class will discuss simple and effective approaches to construct QML algorithms and explain some of the basic and key building elements and overall algorithm architecture.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Dr. Jae-Eun Park, Program Director for Quantum Industry Solution at IBM

Dr. Jae-Eun ParkJae Eun Park is managing IBM GBS Quantum Algorithm Acceleration Teams. Currently he is working on problems in Quantum ML, AI & Machine Learning, Optimization and Simulation for multiple industry application by blending quantum and classical advanced analytics together.

He has 19 years of industry experience in business strategy, advanced analytics & modeling, and artificial intelligence. He has managed IBM Research teams in AI industry focusing on commerce sector including supply chain and marketing. Previously, he managed Research strategy and offering team for IBM Research.

He holds MBA from The New York University of Stern School of Business with Stern Scholar honor and Ph.D in Electrical Engineering and Computer Science from Arizona State University.

 

 

Quantum Engineering Bootcamp: Module 2 - Integrating Your Findings - Best Practices and Pitfalls for Setting Up a Quantum HW Test Program
Thursday, 22 April 2021 at 12pm ET

View this course on-demand

Abstract
When considering a test program comprised of multiple projects and research thrusts, it is important to take an integrated approach to ensure progress. Processes and procedures can serve as guidelines to help ensure said progress and avoid common pitfalls associated with emerging technologies such as quantum computers. This class will address the best practices and pitfalls of setting up an integrated test program.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Will Madsen, Quantum Systems Engineering and Architecture Manager, Rigetti Computing

Will MadsenWill leads systems engineering and integration efforts within the technical organization at Rigetti Computing and manages its portfolio of Department of Defense (DOD) programs. Before joining Rigetti, Will was a Developmental Engineer for the United States Air Force where he led engineering teams in flight testing and space launch operations. He holds a BS in Systems Engineering from the US Air Force Academy.

 

 

Quantum Engineering Bootcamp: Module 1 - Test and Evaluation for Quantum Devices
Tuesday, 20 April 2021 at 12pm ET

View this course on-demand

Abstract
Test and evaluation is at the heart of the push to advance the state of the art in quantum devices. Understanding and adapting test best practices from other industries is imperative to ensure efficiency and speed-up learning cycles. This class will address key considerations with test and evaluation of NISQ era hardware and will aim to educate attendees on how to best think through planning HW tests.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Will Madsen, Quantum Systems Engineering and Architecture Manager, Rigetti Computing

Will MadsenWill leads systems engineering and integration efforts within the technical organization at Rigetti Computing and manages its portfolio of Department of Defense (DOD) programs. Before joining Rigetti, Will was a Developmental Engineer for the United States Air Force where he led engineering teams in flight testing and space launch operations. He holds a BS in Systems Engineering from the US Air Force Academy.

 

 

Overview of Quantum Machine Learning Algorithms
Wednesday, 14 April 2021 at 10am ET

View this course on-demand

Download presentation slides (PDF, 2 MB)

Abstract
Quantum computers make it feasible to solve some problems that are computationally intractable on classical computers. Machine learning (ML) uses big data and statistical/mathematical modelling to solve problems and often the best ML algorithm is the one that scales best as the input grows in size. This makes quantum computers a natural fit for machine learning.

During this talk, we will give an overview of some of the algorithms that exist within quantum machine learning discussing requirements, speedup and possible limitations. This includes HLL and several applications of HLL. Our goal is to give a flavor of how quantum machine learning works and why it is a promising application for quantum computers.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructors
Dr. Troels Steenstrup Jensen, Head of KPMG Global Quantum and Head of Machine Learning at KPMG Denmark
Marco Ugo Gambetta, NewTech Consultant at KPMG Denmark

Dr. Troels Steenstrup JensenTroels is Head of Machine Learning and Quantum Technologies at KPMG Denmark and head of KPMG's Global Quantum Hub. He has a PhD in theoretical quantum mathematics and works at the intersection of mathematics, statistics, physics, computer science and business. He has been working with Machine Learning for more than 10 years and with Quantum Technologies for more than 3 years. Troels has a deep passion for technology and for bringing theory to practice - seeing technology solutions come to life at clients is his main driver. He combines a strong theoretical foundation with business understanding and pragmatic solution design in order to create value for clients.

 

 

Quantum Computing 101: Introduction to Quantum Computing for Non-Technical Learners
Wednesday, 31 March 2021 at 12pm ET

Abstract
Quantum computers could create new industries because of their unique ability to generate extraordinary power that speeds up certain types of complex calculations of great importance in a way that is simply not possible with today’s ordinary computers. They are a more powerful type of computer because they are designed to drastically improve information processing power by taking advantage of special properties of quantum mechanics. This class is designed to introduce quantum computing to non-technical learners who want to have massive quantum fun while learning about this important technology. Register today if you are planning for a career in quantum computing or if you are simply curious about quantum computing because it could shape our future.

Instructor
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Maeva GhondaMaëva Ghonda is a scientist with the unique ability to explain complex information in a manner that is easy to understand. Maëva fell for her true love -- Quantum -- while working as Quantum Scholar for the Joint Quantum Institute (JQI) for a National Institute of Standards and Technology (NIST) Fellow. She began to discover what is possible with quantum -- i.e. Quantum Teleportation and Quantum Money -- while reading intricate details of novel quantum-enabled inventions hidden in global patent documents to uncover valuable quantum technology innovations. Before this fantastic quantum meet-cute, Maëva was an engineer in aerospace where she worked on the production of the 3D printed parts for the autonomous CST-100 Starliner for NASA’s Commercial Crew Program. Moreover, she has also held cybersecurity risk management roles in healthcare and financial services. In addition to her passion for Quantum Computing and Quantum Cryptography, Maëva Ghonda is also quite obsessed with Quantum Teleportation and, of course, Quantum Money.

 

 

Spin Qubit System Integration with Advanced Semiconductor Manufacturing
Wednesday, 24 March 2021 at 12pm ET

View this course on-demand

Download presentation slides (PDF, 6 MB)

Abstract
As the field of quantum computing burgeons, many technology platforms are now in contention for realizing the dream of a useful quantum computer, which can help tackle problems conventional computers cannot. At Intel, we are developing spin qubit systems using our advanced semiconductor manufacturing facilities on 300mm wafers with recent results showing long qubit coherence times. With each wafer that reaches our measurement labs, we have more than 10,000 quantum dot test structures that can be measured. This could represent a large measurement bottleneck if it were not for first-of-a-kind tools such as our 300mm cryoprober, which is able to measure what would have normally been weeks-worth of data within a single day. The statistical data measured by this cryoprober enables critical feedback to integration and manufacturing for improvement of wafer uniformity, device performance, and process stability. By measuring at ~1.6 Kelvin, we can provide this feedback close to the final operating temperature of the spin qubit devices. Additionally, our manufacturing facilities allows us to develop a custom cryogenic CMOS control system, our Horse Ridge control chip. All these devices and systems represent a large cross section of the advanced semiconductor manufacturing process. An overview of spin qubits and this manufacturing and testing infrastructure will be given to demonstrate the role advanced semiconductor manufacturing can serve towards a useful quantum computer.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Dr. Lester Lampert, Quantum Computing Engineer at Intel Corporation

Dr. Lester LampertLester joined the Quantum Computing program at Intel in 2017 and currently works in the Quantum Computing Measurement Lab with a focus on spin qubit control and system integration. Before joining Intel, Lester spent many years studying two-dimensional materials systems, specializing in long-distance spin transport and wafer scale manufacture of graphene. He holds a Ph.D. in Applied Physics from Portland State University and a B.S. in Engineering Physics from University of Wisconsin-Platteville.

 

 

Implications of Quantum Technologies for Cybersecurity
Wednesday, 17 March 2021 at 10am ET

View this course on-demand

Download presentation slides (PDF, 3 MB)

Abstract
Quantum computers pose a threat to cyber security as the core algorithms of current public-key infrastructure are easily broken by a sufficiently large quantum computer. Further escalating the issue, information that needs to stay secure for some time in the future is at risk of "harvest now and decrypt later" attacks. Fortunately, quantum technology also bring means to overcome this challenge with three key solutions to restore security: Quantum Key Distribution, Post Quantum Cryptography and Quantum Random Number Generation. In this class, we will introduce the cyber security challenge and the above-mentioned key means to restore security, including how companies are approaching the situation.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructors
Dr. Troels Steenstrup Jensen, Head of KPMG Global Quantum and Head of Machine Learning at KPMG Denmark
Marco Ugo Gambetta, NewTech Consultant at KPMG Denmark

Dr. Troels Steenstrup JensenTroels is Head of Machine Learning and Quantum Technologies at KPMG Denmark and head of KPMG's Global Quantum Hub. He has a PhD in theoretical quantum mathematics and works at the intersection of mathematics, statistics, physics, computer science and business. He has been working with Machine Learning for more than 10 years and with Quantum Technologies for more than 3 years. Troels has a deep passion for technology and for bringing theory to practice - seeing technology solutions come to life at clients is his main driver. He combines a strong theoretical foundation with business understanding and pragmatic solution design in order to create value for clients.

 

 

Bringing a Quantum Computer to Life with RF Pulses: Fundamental Aspects of Pulse Level Control
Wednesday, 10 March 2021 at 9am ET

View this course on-demand

Download presentation slides (PDF, 2 MB)

Abstract
The qubits in a quantum computer, once they are assembled in place and ready to operate, are inert and in their ground state. To execute a quantum circuit, we must send an intricate and complex sequence of pulses and perform measurements that will determine the state of the qubits. All of this is accomplished using the quantum hardware controller. In this class, we will see how the hardware controller fits into the quantum stack, provide examples on how various quantum gates are translated to pulses, and discuss the evolution of the quantum hardware controller from its roots using lab test equipment into the sophisticated machines that are being built and used today. In particular, we will focus on the OPX, the unique controller offered by quantum machines. Then, we will discuss: (1) the various design and architecture challenges that go into building quantum hardware controllers and (2) why it is essential (a) to build an entirely new classical processing architecture from the ground up in order to maximize the potential of the quantum hardware controller and (b) to create specialized programming languages for pulse level control. We will also demonstrate an example with a particular pulse language named: QUA.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Dr. Lior Ella, Research and Product Team Leader, Quantum Machines

Dr. Lior EllaLior is a physicist with extensive experience in the development of quantum devices and techniques. He holds advanced degrees in both electrical engineering and in physics. He is currently a research and product team leader at Quantum Machines, working on system, product and architecture engineering of the next generation of quantum hardware controllers.

 

 

Fundamental Concepts in Quantum Error Correction
Wednesday, 24 February 2021 at 10am ET

View this course on-demand

Access course on the IEEE Learning Network
Earn 1 PDH / 0.1 CEUs

Download presentation slides (PDF, 4 MB)

Abstract
The long-term vision of quantum computing relies on building systems that implement Quantum Error Correction (QEC), which enable computations to be robust to physical qubit errors. In this class, I will break down QEC into basic concepts that will help you grasp the ongoing research and development in the field. I will discuss why error correction is a necessity for scalable systems, what stabilizers are and how they detect and protect quantum states from error, the basics of error correcting codes such as the Surface Code, and how measurements are used to decode and remove quantum errors. I will not cover the basics of qubits, quantum circuits, or linear algebra, which you will need to get the most out of this course. On a personal level, I find QEC to be engaging for its mix of quantum circuit manipulation, computer science concepts, and fundamental quantum mechanics with deep implications on the future of quantum computing. Hopefully you will find QEC as fun as I do!

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Dr. Julian Kelly, Research Scientist, Google AI Quantum

Dr. Julian KellyDr. Julian Kelly is a Research Scientist at Google AI Quantum. He is the lead for the System Control Team which is responsible for building the hardware and software to operate and manipulate Quantum Computers. He began his career in Quantum Computing in 2008 where he joined John Martinis' physics research group at UCSB as an undergraduate and researched qubit control and benchmarking techniques. Julian stayed at UCSB and completed his PhD in 2015 in experimental quantum computing. His thesis focused on the development of highly controllable, coherent, and scalable "Xmon" transmon systems that demonstrated record fidelity entangling gate and measurement operations, culminating in a demonstration of experimental quantum error correction. Since joining Google, Julian worked to improve, scale, and integrate quantum processors and was the lead designer for the 72 qubit Bristlecone processor. Julian also developed the automated calibration framework "optimus" which is a software backbone of operating quantum processors at Google. The above technologies were critical in the team's 2019 demonstration: "Quantum supremacy using a programmable superconducting processor.”

 

 

Quantum Systems Engineering for Scientists
Wednesday, 10 February 2021 at 4pm ET

View this course on-demand

Abstract
Dr. Martinis would like to invite you to a talk on Quantum systems engineering for scientists. As the field of quantum Computing has advanced building complex machines it seems like a good time to talk about some of the organizational principles that one might use for such a large effort. System engineering concepts have been well developed for other technologies, so here he has focused on quantum computers. This special emphasis comes from the need for engineering discipline for the many physicists and scientists on the project who typically don't have an engineering background, so his talk will cover some basic principles. He will also discuss some of the unusual constraints that are found for quantum computers such as the inability to copy information and the large amount of information that is needed to control qubits. Here's an example of an interesting principle that scientists should know. Although the scientific method is the foundation of all technology, it is well-known that strictly following the scientific method for project management will cause failure so you will want to know why. This is an important subject for the field of Quantum Computing so please come and bring a lot of questions since Dr. Martinis would like to learn from you through active engagement.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Dr. John Martinis, Physics Professor, UCSB

Dr. John MartinisJohn Martinis did pioneering experiments in superconducting qubits in the mid 1980’s for his PhD thesis. He has worked on a variety of low temperature device physics during his career, focusing on quantum computation since the late 1990s. He was awarded the London Prize in Low temperature physics in 2014 for his work in this field. From 2014 to 2020 he worked at Google to build a useful quantum computer, culminating in a quantum supremacy experiment in 2019.

 

 

Systems Engineering Approaches and Challenges in Quantum Computing
Wednesday, 20 January 2021 at 10am ET

View this course on-demand

Access course on the IEEE Learning Network
Earn 1 PDH / 0.1 CEUs

Abstract
NISQ era quantum computers can perform useful applications today. But, realizing the full potential of these systems will require both advances and close collaboration from a broad swath of science and engineering disciplines. Traditional systems engineering models, typically adapted from aerospace and defense industries, are often too prescriptive in defining requirements and use-cases. Furthermore, enterprise systems engineering methods and tools are often focused on how to best prepare an enterprise for change, not how to vector the development of specific systems that will disrupt enterprises. Systems engineering professionals need to consider augmenting best-practices while building lower TRL emerging technologies, such as quantum computing, that require more flexible planning. This course will address possible approaches and challenges, especially for systems with many potential use-cases of strategic interest.

Host
Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program

Instructor
Will Madsen, Quantum Systems Engineering and Architecture Manager, Rigetti Computing

Will MadsenWill leads systems engineering and integration efforts within the technical organization at Rigetti Computing and manages its portfolio of Department of Defense (DOD) programs. Before joining Rigetti, Will was a Developmental Engineer for the United States Air Force where he led engineering teams in flight testing and space launch operations. He holds a BS in Systems Engineering from the US Air Force Academy.

 

 

The Quantum Computing Education - Workforce Development program is brought to you by:

IEEE Quantum IEEE Standards Association

 

This page is based on presentations at ICRC 2019 in San Mateo, CA. The views expressed by the speakers are theirs alone.

 

How is quantum computer performance assessed?

Benchmarks for conventional computers are standardized methods that test and evaluate hardware, software, and systems for computing. The results from these tests are expressed using metrics that measure features and behaviors of the system such as speed and accuracy. With the advent of quantum computers, new benchmarks are needed to address these same metrics while also accounting for differences in the underlying technologies and computational models.

While conventional computers rarely fail to operate as designed, current quantum computing systems are susceptible to large error rates. This dramatic difference in technological readiness influences the design and purpose of the quantum computing benchmarks, which are currently focused on identifying the scaling behavior with respect to size and error rates. This may include measuring coherence times or scattering rates specific to quantum technologies alongside the speed and accuracy familiar from conventional computers.

For example, ion trap technology for quantum computing, the stability of the trap frequency, the duration of a gate operation, and the stability of the control lasers all represent unique measures of performance. Similarly, for superconducting technologies, the precision of the Josephson, anhamonicity, and gate duration, are equally relevant to performance. The operation of these devices emphasize additional concerns in the controllability and scalability of the device and system, which can be addressed through more holistic benchmark methods based on gate error rates and circuit fidelities.

Benchmarks and metrics must also account for the differences between the underlying computing model. Currently, the gate- and adiabatic/annealing models of quantum computing represent different approaches to designing and operating quantum computing devices, and the benchmarks are tailored to these architectures in order to account for those differences. These differences also give rise to a different in size of the available quantum computers -- gate-models support less than 100 qubits currently while quantum annealers support up to 2,048 qubits. These differences in system capacity underscore the need for a diversity of approach to metrics and benchmarks for quantum computing.

See introductory video, Travis Humble, Kevin Young, Cathy McGeoch. 2:13 to 17:13

 

What is quantum supremacy and cross-entropy benchmarking?

The material on this section cross-entropy benchmarking and quantum supremacy was recently published in Nature by authors from Google.1

To benchmark quantum computers, you run quantum circuits from a set of single- and two-qubit gates and measure the resulting data string outputs of zeros and ones. Some data strings will be more probable than others, with the set of probabilities comprising the output of the benchmark that have to be compared against the correct values. We use this method because we know that if you choose the gates at random, it is a hard computational problem for classical computers, but not as hard for quantum computers.

Cross-entropy benchmarking and a quantum supremacy experiments use two different meanings of the word “benchmarking.” One is trying to measure the worst-case equivalent computational power of the quantum computer. The second is to assess the fidelity or accuracy of the experimental quantum computer, which is important to help design the next generation system as well as to learn the “control map” that adjusts control signals on the fly to compensate for manufacturing variance.

So what is the cross-entropy metric? We compute the probability of some data stream Q by sampling the output of the experimental quantum computer millions of times. There are other observables, such as the heavy output score introduced by Aaronson and Chen, and we use a new chip the hardware group has developed called Sycamore with 54 Xmon qubits, Xmons are a version of transmons. The new feature in Sycamore is that the couplers between the qubits are adjustable, which is how we create random gates without remanufacturing the chip.

Google’s road map is to create a fault tolerant computer, which will require error rates about 10 times lower than the error correction threshold, which is around 10-2 error rate, so we will need 10-3. After this, we will scale the number of qubits.

We already perform cross-entropy benchmarking the whole system, not just simultaneous two-qubit gates. We fix the circuit’s depth to 14 cycles of two-qubit gates with single qubit gates in between, which means the quantum circuit is really 28 clock cycles. An equivalent supercomputer simulation takes 5 hours for one circuit and there were 10 simulations running in parallel on a million cores, so this is 50,000,000 core hours—which we ran on Google clusters.

In terms of classical simulation algorithms for the run time comparison, there is the Schrödinger algorithm, which stores the entire quantum state in RAM. Nobody has implemented such an algorithm that goes beyond that references 48 qubits, but IBM is exploring the use disk.

Quantum Supremacy Data

We expect quantum hardware will also keep improving, so the race is on for both more capable quantum chip and larger classical simulations. So, it will be very important for IBM to run its improved classical simulation algorithms, which will face challenges due to scale such as failures, checkpointing to recover from failures, high energy consumption, and other nightmares that arise for any large supercomputer run.

See video by Sergio Boxio, Cross Entropy Benchmarking and Quantum Supremacy, duration 34:45

 

What is quantum algorithmic breakeven for quantum annealers?

Rather than focusing on single- and two-qubit errors, the larger scale of quantum annealers is sufficient to actually look ahead and talk about speedup. This leads to a new concept called quantum algorithmic breakeven.

Quantum Algorithmic Breakeven

To understand quantum speedup, let’s first clarify that it is not an about actual runtime. Actual runtime is highly dependent on which classical supercomputer you measure your quantum system against. Instead, quantum speedup is about scaling with respect to the time to solution. A quantum speedup could be exponential scaling for a classical algorithm versus a polynomial scaling for a quantum algorithm--or a lower exponent when we're experiencing exponential scaling in both cases—or maybe a lower degree polynomial if we're talking about polynomial scaling and the quantum exponent is less than the classical exponent.

The recent spat between Google and IBM it's an example of an argument about IBM’s claim of 2½ days classical simulation time versus Google’s claim of 10,000 years. However, this is an argument about actual run time, which according to the definition discussed here is not really related to quantum speedup.

Imagine the following dream scenario. The quantum computer’s time to solution on a semi log scale would be polynomial for some algorithm, such as quantum simulation, in the ideal case where there's no decoherence and no control errors. However, the classical time to solution would be exponential. We would see a beautiful separation on the graph where the quantum scaling advantage is clear for every problem size. But unfortunately, this has not been observed on any quantum hardware to date.

A more realistic scenario is we have some decoherence and then the quantum polynomial scaling is somewhat worse. The exponent could be higher, so perhaps we would not see a speedup when the number of qubits is small and the quantum scaling advantage only becomes clear for larger N. It's reasonable to assume current NISQ-era quantum systems follow this scenario.

Now let’s look at algorithmic speedup. The only current quantum systems of sufficient scale to assess algorithmic speedup quantum annealers. So for a particular work we published last year,2 we actually demonstrated a speedup of time to solution on a D-wave, or showing time to solution as a function of problem size for scaling better for D-Wave was better than classical simulated annealing. Now why was this not published in Nature and made it to the New York Times?

Because the classical algorithm was not the best one. And Unfortunately for D-Wave, when we compared its performance against other algorithms, it lost.

However, there's something you can do for annealers that is similar to randomized techniques, called dynamical coupling, it is a very primitive type of error suppression method.

So let's go back to the shape of the graph discussed above and see if there is hope for quantum error correction on annealers? The goals would simply be to demonstrate that quantum error correction can take uncorrected quantum scaling, push it down a little bit, so that perhaps it is still not better than a classical computer, but you are demonstrating that error correction helps in terms of the time to solution metric relative to the uncorrected setting. So, what is algorithm breakeven with quantum error correction? It’s the idea of demonstrating a corrected quantum scaling that is better than the uncorrected quantum scaling, but it doesn't necessarily have to be better than classical—that bar is too high for NISQ-era devices.

We observed the time for solution as a function of problem size increasing on a semi-log scale roughly exponentially without error correction. The clincher is that scaling improved with error correction.

We reached this point 5 years ago but have made progress on a more challenging goal since then. Five years ago we only had 500 qubits and didn't look at that control errors, so now we have 2,048 qubits and can now account for control errors.

After many millions of experiments without error correction, the number of runs or the time to solution required to find a certain ground state, plotted on a semi log scale as function of the number of qubits, with added noise, showed that as you add more noise the scaling increases dramatically, i.e. super exponential. What happens with error correction, with the quantum annealing correction method described above, is what happens to the scaling. So once again this is an example of algorithmic breakeven in the sense that the uncorrected versus the corrected scaling is better you can eyeball it for every value of noise and for every value of qubits it's true that that the result is better for the corrected setting.

So, we have introduced a notion of quantum algorithmic breakeven, which is the idea that we should demonstrate an error corrected scaling that is better than the uncorrected scaling on a non-trivial computational problem. It’s early days for NISQ processors but its moving in the direction where these kind of demonstrations should become possible.

See Quantum Algorithmic Breakeven: on scaling up noisy qubits, Daniel Lidar. 55:13 to 1:34:05

 

How is Quantum Computer Performance Measured?

This talk covers volumetric randomized benchmarking and mirror networks. This work on scalable benchmarking originated in the Quantum Performance lab at Sandia National Laboratories. The talk first discusses why we need to do benchmarking quantum computers in the first place. Volumetric benchmarking is the framework of the solution discussed, but you also need circuits to actually run in the framework, for which Sandia developed mirror circuits. The talk also shows experimental results.

What are we trying to benchmark? Unlike standard classical computers, the components in a quantum computer fail frequently, forming the main limiting factor with quantum computers.

 

What are some examples of quantum computer errors?

  • Bit flips, just like in classical computers
  • Failure modes unique to quantum computers, such as errors that add up coherently
  • Drift in performance over time, based on the fact that quantum computers are analog
  • Cross talk
  • Integration failures where a device doesn't actually perform as well as the sum of its parts

Real quantum computer errors have to be explained in the context of real devices, and we’ll use the publicly available IBM Q Tenerife from IBM quantum experience as an example. The device’s website shows the name of the device and graph showing its 5 qubits in a particular layout and a spec. For example, the gate error rate for qubit 0 is 0.1% for gates on just that qubit and a 2-qubit gate between qubit 0 and qubit 1 has an error rate of 0.3%. Multiply those numbers together and subtract from 1 using basic and probability theory and you get 15%. So, for this particular circuit you would think that the chance of failure is 15%.

The reality is that this doesn't work generally because it ignores almost all structure in the circuit, or where different errors interact with different structures in complicated ways, such as emergent noise, crosstalk, and so on.

So we need more than low-level benchmarks. This talk presents some approaches, but other recent developments include cycle benchmarking presented elsewhere at this workshop.

Sandia developed a volume metric benchmarking framework, inspired by IBM's work on quantum volume. IBM recognized that adding more qubits doesn't increase computational power if there's going to be an error before all the qubits can interact with each other. You essentially have a smaller device than your thought you had. Furthermore, decreasing error rate won't actually increase computational power if you can already access all the states. So, IBM defined the effective size of the device as the largest number of qubits for which you can access the entire state. In more detail, the quantum circuits have both width and depth equal to D and they’re a type of scrambling circuit. The quantum volume is defined as 2D*, where D* is the largest D for which the circuit computes the correct answer with acceptable reliability.

Quantum volume is a pretty nice way to benchmark a quantum computer, but it does not give a complete measure device performance. Unfortunately, real programs process data in accordance with an algorithm, which often has different properties than a D×D random scrambling network. For example, the straightforward implementation of Shor’s algorithm has order n qubits and order n3 depth, yet other algorithms that have lower depth than the number of qubits. Another complication is that we don't know yet what algorithms will be important for quantum computers.

So. this inspires the volumetric benchmarking framework we've been developing. Sandia defines a volumetric benchmark as a map from widths and depths to a measure of success for an ensemble of circuits at each width and depth. The plot below shows exemplary data just to demonstrate how the method allows a person to visualize the data from these methods. Each data point shows weather the circuit succeeded or failed with a binary success or failure measure. The blue squares are where circuits were successful and white where it failed. The frontier is where we can no longer run the circuit successfully. We can compare the predictions of this spec sheet, for example from IBM, Rigetti, or some new device you’re evaluating.

Volumetric Benchmarking Qubits

 

What are Mirror Circuits?

Sandia developed a specific class of circuits called mirror circuits to facilitate volumetric benchmarking. Their general structure is based on motion reversal and consists of:

  • A general D-input unitary in 2 layers, i.e. a quantum gate network
  • A layer of local randomization on each qubit
  • The inverse of the previous unitary in D over 2 more layers

Within this framework there may different things you could do:

  • The unitary could be a random circuit, and then you run the random circuit in reverse, undoing the calculation and giving the original input back. The output should be the same as the input, making it easy to check correctness.
  • The unitary could be based on a structured circuit, such as one layer or perhaps a few layers in repetition to see how structure impacts performance

We have principally considered circuits that only contain only Clifford gates and so that and the randomization is a Pauli, which means that the outcome is a fixed bit stream that a classical computer can easily calculate.

We benchmarked 12 different quantum computers, with results from 7 shown below, including devices from IBM and Rigetti Computing. For Agave, for example, the black line that shows the threshold beyond which we can no longer run these circuits. So, for example, we can run out to depth six for two qubits but we can't go out to depth ten.

Benchmarking 7 Quantum Computers with Randomized Mirror Circuits

 

Structured Circuits

Structured circuits could be very different in general, so let’s combine ideas and consider a set of structured mirror circuits. The plot below shows what we get when we run on IBM Q Ourense with their recent devices, showing random best average and worst case again just like before and then there's structured worst case and what you can see is that the structured performance is much worse than the other cases, so the structure is really causing a big impact and this is a clear sign of coherent errors.

Structured Mirror Circuits

See the video for more information. Essentially, holistic benchmarking is clearly important as many sources of error only emerge at scale, making current types of performance data insufficient to predict performance. We developed a general framework around volumetric benchmarking and some specific classes of circuits for the framework including randomized and structured mirror circuits. The circuits scale and we've run them on all the quantum computers that are currently publicly available. The results show the predictive power of spec sheets varies a lot between devices but they're generally not very predictive and they’re overoptimistic of how good devices how well devices actually perform. Our work also shows that circuit structure matters, so just giving a set of gate error rates is not enough. In addition, performance on standard randomized benchmarks doesn't guarantee good performance on real application circuits. It's going to be very important to carefully benchmark computers in the near term and we think mirror circuits are good candidate for this.

See Demonstrating Scalable Benchmarking of Quantum Computers, Tim Proctor.

 

References

[1] Arute, Frank, et al. "Quantum supremacy using a programmable superconducting processor." Nature 574.7779 (2019): 505-510.

[2] Albash, Tameem, and Daniel A. Lidar. "Demonstration of a scaling advantage for a quantum annealer over simulated annealing." Physical Review X 8.3 (2018): 031016.

 

AtomThis online community is intended to help educate and inspire the next generation of QIS scientists and engineers.

Quantum Information Science (QIS) is at a historically important juncture.

  • Its laboratory bona fides have been firmly established; now, scientists and engineers at scores of companies and institutes are racing to transform lab projects into scalable, production-ready systems that can be turned loose on real-world problems.

The situation gives rise to a number of questions.

  • What expectations should society, especially those involved in setting government policy, have for the near- and longer-term future of quantum machines?
  • What should today's science and engineering students be taught about the growing body of quantum information sciences?
  • Might a shortage of skilled workers hamper the roll-out of robust quantum systems?

There is rapid funding now available in quantum computing, like the billion-dollar National Quantum Initiative and likely industry co-investment.

  • The immediate problem is that there are not enough workers trained in quantum information to effectively spend the projected funds. These immediate needs could be addressed by incremental education, such as one or two courses that would allow a skilled circuit designer, for example, to design quantum circuits -- or a materials researcher to be able to study quantum information behaviors in qubits

 

Quantum Computing Education - Workforce Development

Quantum Computing Education - Workforce DevelopmentQuantum Engineering: Photonics in Quantum Computing and Quantum Networking
28 July 2021 | Virtual Event

View the latest course in our Quantum Computing Education - Workforce Development series, "Quantum Engineering: Photonics in Quantum Computing and Quantum Networking," instructed by Dr. Peter McMahon, Assistant Professor, Cornell University School of Applied and Engineering Physics (AEP), and hosted by Maëva Ghonda, IEEE Chair, Quantum Computing Education for Workforce Development Program.

Access on-demand

 

Summary of the 2019 IEEE Workshop on Benchmarking Quantum Computational Devices and Systems

2019 IEEE Workshop on Benchmarking Quantum Computational Devices and Systems7 November 2019 | San Mateo, California, USA

A summary and speaker presentations on the topics of quantum supremacy and quantum computer performance are now available from our half-day workshop on benchmarking quantum computational devices and systems. The workshop was held in conjunction with the 2019 IEEE International Conference on Rebooting Computing (ICRC) and was part of IEEE Rebooting Computing Week 2019.

View the summary

 

 

Resources

Satellite-Based Continuous-Variable Quantum Communications: State-of-the-Art and a Predictive Outlook (Open Access)
IEEE Communications Surveys & Tutorials, Volume 21, Issue 1

IEEE GLOBECOM 2016 Tutorial: Quantum Communications (PDF, 6 MB)
Rob Malaney, UNSW

 

Videos

 

Introduction to Quantum

Tech Pioneers Weigh in on the Vast Potential of Quantum ComputingTech Pioneers Weigh in on the Vast Potential of Quantum Computing, 15 July 2019
A Beginner’s Guide to Quantum Computing | Talia Gershon, IBMA Beginner’s Guide to Quantum Computing | Talia Gershon, IBM, 31 May 2017
The Mathematics of Quantum Computers | PBS Infinite SeriesThe Mathematics of Quantum Computers | PBS Infinite Series, 16 Feb 2017
Quantum Computing Explained: An IntroductionQuantum Computing Explained: An Introduction to Quantum Computers, 30 May 2014
Quantum Computing Explained | Lawrence KraussQuantum Computing Explained | Lawrence Krauss, 26 Aug 2013

 

University Lectures

Quantum Supremacy: Checking a Quantum Computer with a Classical Supercomputer | John MartinisChecking a QC with a Classical Supercomputer | John Martinis, 12 Jun 2018
Quantum Computing: Untangling the HypeQuantum Computing: Untangling the Hype - The Royal Institution, 2 May 2018
What Will We Do With Quantum Computing? | Aram Harrow, MITWhat Will We Do With Quantum Computing? | Aram Harrow, MIT, 13 Feb 2018
The Future of Quantum Computing | Prof. Seth LloydThe Future of Quantum Computing | Prof. Seth Lloyd, 2 Jan 2018
Quantum Computing and the Entanglement Frontier | John Preskill, CalTechQuantum Computing and the Entanglement Frontier | John Preskill, CalTech, 7 Mar 2017
The Quantum Computational Universe - 1 of 2 | Patrick HaydenThe Quantum Computational Universe - 1 of 2 | Patrick Hayden, 2 Dec 2016
Quantum Computing Concepts - Quantum Hardware | Andrea Morello, UNSW AustraliaQC Concepts - Quantum Hardware | Andrea Morello, UNSW Australia, 20 Apr 2016
Quantum Entanglements, Part 1 (Stanford) | Leonard SusskindQuantum Entanglements, Part 1 (Stanford) | Leonard Susskind, 25 Sep 2006

 

Innovations in Industry

Quantum Computing: Don’t Count Your Qubits Before They’re Hatched | Robert Sutor, IBMDon’t Count Your Qubits Before They’re Hatched | Robert Sutor, IBM, 28 Sep 2018
Fujitsu Forum 2017 Keynote - Quantum Computing | Joseph RegerFujitsu Forum 2017 Keynote - Quantum Computing | Joseph Reger, 9 Nov 2017
Quantum Computing - Top 3 Microsoft Breakthroughs with Krysta SvoreQuantum Computing - Top 3 Microsoft Breakthroughs with Krysta Svore, 2 Oct 2017
Quantum computing explained with a deck of cards | Dario Gil, IBM ResearchQuantum computing explained with a deck of cards | Dario Gil, IBM, 22 Jun 2017
Quantum Hardware at Google | Josh MutusQuantum Hardware at Google (Chapman University) | Josh Mutus, 19 Oct 2016
Quantum Computing: Artificial Intelligence Is Here | Geordie RoseQuantum Computing: Artificial Intelligence Is Here | Geordie Rose, 25 Aug 2015

 

TEDx Talks

Quantum computing demystified | Ilyas Khan | TEDxCityUniversityLondonQC demystified | Ilyas Khan | TEDxCityUniversityLondon, 20 Dec 2017
Introduction to Quantum Computing | Koen Bertels | TEDxAntwerpIntroduction to Quantum Computing | Koen Bertels | TEDxAntwerp, 16 Nov 2017
What Quantum Computing Isn't | Scott Aaronson | TEDxDresdenWhat Quantum Computing Isn't | Scott Aaronson | TEDxDresden, 1 Nov 2017
How quantum computers are different! | Alireza Shabani | TEDxUCLAHow quantum computers are different! | Alireza Shabani | TEDxUCLA, 8 Jul 2016
Can we make quantum technology work? | Leo Kouwenhoven | TEDxAmsterdamCan we make quantum work? | Leo Kouwenhoven | TEDxAmsterdam, 29 Nov 2015
Knots, World-Lines, and Quantum Computation | Steve Simon | TEDxOxfordKnots, World-Lines, & Quantum Computation | Steve Simon | TEDxOxford, 4 Aug 2015
Quantum computers | David DiVincenzo at TEDxEutropolisQuantum computers | David DiVincenzo at TEDxEutropolis, 24 Oct 2013
Quantum computing, the story of a wild idea | Andris Ambainis at TEDxRiga 2013Quantum computing, the story of a wild idea | Andris Ambainis at TEDxRiga 2013, 9 Aug 2013
Quantum Computing Update | Ray Laflamme at TEDxWaterloo 2013QC Update | Ray Laflamme at TEDxWaterloo 2013, 30 May 2013
Nanoelectronics and Quantum Computation | Charlie Marcus - TEDxCaltechNanoelectronics and Quantum Computation | Charlie Marcus - TEDxCaltech, 18 Feb 2011

Subcategories