Quantum Computing

Noise Amplification Squeezes More Computational Accuracy From Today’s Quantum Processors

Share this post:

Conventional wisdom holds that two wrongs don’t make a right. However counter intuitive this may seem, many wrongs could be the key to improving the performance of noisy quantum computers today.

The problem

Computers that rely on the power of quantum mechanical phenomena to perform calculations are also extremely susceptible to “noise” from their environment – which leads to errors in the computation. Even at the extreme cold temperatures of a dilution refrigerator where the quantum processors operate, our physical computing elements, superconducting qubits, have coherence times on the order of a few hundred microseconds at best, which sets the timescales over which quantum information is lost. While a major challenge to advancing quantum computers today involves increasing these qubit coherence times, the end goal is to build a fully fault tolerant quantum computer capable of detecting and correcting errors. However, these architectures are likely several years away.

A near-term solution

Single qubit trajectories measured at different noise levels (red, green) are used to estimate the error-mitigated trajectory.

Single qubit trajectories measured at different noise levels (red, green) are used to estimate the error-mitigated trajectory (blue). Figure modified from A. Kandala et al Nature, 2019

In a new research paper, published in Nature, entitled “Error mitigation extends the computational reach of a noisy quantum processor,” we demonstrate a way to improve the accuracy of quantum computations on today’s noisy hardware. We show that repeating the computation at varying levels of noise lets us estimate what the quantum computer would calculate in the absence of noise. To do this involves a bit of a “stretch” – the microwave pulses used to perform quantum operations on the qubits are stretched in time in order to controllably amplify the noise.

The error mitigation technique dubbed ‘zero-noise extrapolation’ is readily accessible for existing quantum computers since it doesn’t require any additional hardware modifications.

Longer computations

Computations on noisy quantum hardware are limited by the competition between decoherence and circuit depth, a measure of the number of sequential operations performed on the processor. Increasing circuit depth can help create more complex quantum states, and in the context of chemistry simulation, this may allow for a better representation of the energy states of the molecules considered. However, increasing circuit depth on a noisy quantum computer typically implies increased errors from decoherence. However, with the technique developed in this work, our ability to mitigate the effect of decoherence enables us to access more complex and accurate computations that benefit from increased circuit depth.

Four-qubit molecular simulations of LiH using noisy quantum processors: The experimental data (white circles) are compared with the exact energy curve (green dotted line). The left panel shows the results from the 2017 paper “Hardware-efficient Variational Quantum Eigensolver for Small Molecules and Quantum Magnets.” The right panel depicts our most recent results from the paper “Error mitigation extends the computational reach of a noisy quantum processor.” In the right panel, the error-mitigated estimates (white circles) are obtained by an extrapolation of results from experiments of varying noise (colored circles) and display far superior accuracies, without significant hardware improvements to the processors used for these computations.

General-purpose technique

Our method is fairly broad in its applicability and can be used to improve any quantum computation that relies on expectation values. For example, in this work, we use it to demonstrate improvements in the accuracy of quantum simulations initially considered in our 2017 Nature study Hardware-efficient variational quantum eigensolver for small molecules and quantum magnets.” Using up to four qubits of our 5-qubit superconducting processor, we compute the lowest energy state of the molecules H2 and LiH, as well as archetypal spin models in solid-state physics. The error mitigation technique was also employed to enhance the success of binary classification in the recent paper “Supervised learning with quantum enhanced feature spaces,” which was featured on the March 14 cover of Nature.

The path ahead

While our technique enabled computational accuracies that were otherwise inaccessible to the hardware, it is important to note that the improvements are not indefinite and are ultimately limited by the coherence properties of the processor. As we march towards systems with increasing Quantum Volume, improved coherence, quantum control, and circuit compilation will all contribute to extending the reach to even longer quantum circuits with more qubits.

Error mitigation extends the computational reach of a noisy quantum processor
Abhinav Kandala, Kristan Temme, Antonio D. Corcoles, Antonio Mezzacapo, Jerry M. Chow, Jay M. Gambetta
doi: 10.1038/s41586-019-0980-2

IBM Fellow and Vice President, Quantum Computing

Abhinav Kandala

Research Staff Member, Experimental Quantum Computing

More Quantum Computing stories

IBM Supports Q2Work Education Initiative

The need for a future workforce with a robust set of quantum computing skills drives our support for Q2Work, the National Science Foundation-funded initiative led by the University of Illinois and the University of Chicago to provide quantum education, programs, tools, and curricula to K-12 students.

Continue reading

Quantum Conundrum: Clifford Group Investigation Unexpectedly Reveals New Quantum Advantage Proof

When we began our current line of investigation, the goal was to study the structural property of the Clifford group, describing a set of transformations that generate entanglement, play an important role in quantum computing error correction, and are used in (randomized) benchmarking. In a series of one-thing-leads-to-another findings, however, we ended up discovering a new mathematical proof of quantum advantage – the elusive threshold at which quantum computers outperform classical machines in certain use cases.

Continue reading

IBM Roundtable: Building a Quantum Workforce Requires Interdisciplinary Education and the Promise of Real Jobs

The ability to harness quantum-mechanical phenomena such as superposition and entanglement to perform computation obviously poses a number of difficulties. Add in the need to make these systems perform meaningful work, and you’ve raised the stakes considerably. Creating a pipeline of talented, well-trained academics and professionals who can meet those challenges was the subject of IBM’s July 28 virtual roundtable, “How to Build a Quantum Workforce.” Watch the replay, here.

Continue reading