Please Note: The 2020 American Physical Society (APS) March Meeting that was to be held in Denver, Colorado from March 2 through March 6 has been canceled. The decision was made late Saturday (February 29), out of an abundance of caution and based on the latest scientific data available regarding the transmission of the coronavirus disease (COVID-19). See our official release on the cancelation for more details.
DENVER, COLO., FEBRUARY 28, 2020 -- Google made headlines in late 2019 with an experiment that demonstrated quantum supremacy for the first time. Their quantum computer, the Sycamore Processor, took a mere 200 seconds to perform a computation that would have taken a traditional computer 10,000 years. Members of the research team--Pedram Roushan, Zijun Chen, and Kevin Satzinger--will discuss this groundbreaking feat at the 2020 American Physical Society March Meeting in Denver.
Roushan will report the results of this experiment, discuss its computational complexity, and show how the team could verify the results of their quantum computation. Looking to the future, He will also demonstrate the programmability of the Sycamore Processor and how various algorithms can be implemented on it.
When the Google research team performed their experiment, they measured the outputs of the Sycamore Processor's 53 superconducting qubits simultaneously. This is one of the largest simultaneous measurements of a superconducting quantum computer ever done. Google researcher Zijun Chen will present the new strategies used by the research team to get precise results from their unprecedentedly large experiment and reduce measurement errors. Such strategies could be crucial to operating larger quantum computers..
"As far as I know, no one's ever done anything quite like this," Chen said.
Kevin Satzinger, also a Google researcher, will discuss the engineering advances of the Sycamore Processor device which allowed it to accomplish its record-breaking feat. He will also demonstrate the benchmarking methods used to evaluate its performance--which, themselves, were impossible methods to implement on a classical computer and therefore demonstrated quantum supremacy. Lastly, Satzinger will present the digital error model used by the researchers, which was verified by the experiment. This digital error model represents a step forward toward solving the problem of quantum error correction, which is the major obstacle to scaling up quantum computers.
Quantum computing scientists refer to the current period of time as the Noisy Intermediate-Scale Quantum (NISQ) era. Quantum computers of over 50 qubits have been created, entering an intermediate scale where we might expect to see evidence of quantum supremacy: a quantum computer solving a problem that a classical computer might take millennia to process. However, measurements of these quantum computers will be noisy because 50-100 qubits is not enough to implement a self-correcting algorithm that would eliminate or reduce noise.
Two presenters in the session will be discussing their research on quantum error correction in the NISQ era, when quantum computers don't have enough qubits to implement self-correction. Ramis Movassagh of IBM Research tackled a theoretically complex problem: how do you quantify the difficulty of a problem posed to a quantum computer, when that difficulty needs to include the inherent noise? Previous work suggested that the task of Random Circuit Sampling (RCS) has a certain level of difficulty for a quantum computer as the number of qubits becomes arbitrarily large. This task was run on Google’s Sycamore Processor for 53 qubits. However, the previous mathematical work left no room for noise and errors, which might decrease the problem’s difficulty. Movassagh, however, proved that RCS is still a difficult task when including some specific additive errors, thereby significantly advancing toward a proof of the quantum supremacy conjecture. This work could provide a new way of considering problem difficulty in an era of noisy quantum computers.
Gian Giacomo Guerreschi from Intel Corporation and collaborators from Carnegie Mellon University have tested a noise-robust algorithm that is a hybrid of quantum and classical computing: the Quantum Approximate Optimization Algorithm (QAOA). "Its back-and-forth between quantum and classical computing should get rid of systematic errors," says Guerreschi. Because it's a mix of two computing types, he prefers to call it a protocol rather than an algorithm, and says they are developing a way to tailor QAOA to different kinds of problems.
Sebastian Deffner, of the University of Maryland Baltimore County, will present a new theory of quantum thermodynamics at the quantum supremacy session. Since the 1960s, scientists have had a good understanding of how much energy computers require to process a given amount of information. This understanding is missing for quantum computers.
"Let's say you take a selfie and you uploaded it to Snapchat," he said. "You can calculate exactly how much energy has been drawn from the battery of your phone. For quantum computers, we don't have that yet."
Theoretically, quantum computers may be exponentially better than classical computers at solving certain problems; but that might mean they require exponentially more power to do so. But work and energy operate differently on the quantum scale, where states can be in superposition and positions can have set uncertainties. "We cannot rely on old concepts anymore, but we need something new," Deffner said.