Predicting robust and sensitive outputs in real-world noisy quantum simulators

April 20, 2021 - Pablo Poggi, PhD


(a) Schematic of a dynamical quantum simulator subject to errors. (b) Diagram of the experimental device. (c) Energy level diagram corresponding to the hyperfine ground manifold of cesium, indicating the application of magnetic radio-frequency (rf) and microwave (μW) control fields. (d) Quantum simulation scheme through discrete-time evolution.  

One of the most exciting prospects of near-term quantum technologies is to enable simulation of complex quantum systems such as exotic superconducting materials, large molecules for pharmaceutical discovery, or high-energy particle physics. Quantum simulators are devices that are designed to serve this special purpose and are thus expected to work with less-stringent requirements than those needed for universal error-corrected digital quantum computers. However, quantum simulators are intrinsically analog computers, subjected to a continuum of errors. Can we trust the output of a noisy analog quantum simulator?  It is essential to understand how errors affect the performance and reliability of these devices. 

Recent work by the long-standing theory-experiment CQuIC collaboration between the University of New Mexico and the University of Arizona focused on studying how the unavoidable imperfections that are present in any real-world quantum device affect its ability to reliably extract information about a system. By proposing a model in which small random perturbations uncontrollably affect the evolution of the device in each run, the authors established a quantitative theoretical model which characterizes the simulation output in terms of robust and fragile observables.  

These theoretical predictions were then put to the test in a state-of-the-art small highly accurate quantum (SHAQ) simulator at the University of Arizona, the development of which was published in Physical Review Letters in 2020 [1]. The results of the quantum simulator, which are inevitably affected by noise and imperfections, were compared to the ideal expected result obtained numerically for different choices of output observables. The comparison revealed that the theory model was able to accurately predict which observables are more robust and which are more fragile, even without having to do the hard work of unraveling the actual, physical errors that happen on the specific hardware.  

The findings of this work, published in PRX Quantum [2], are expected not only to find applications in other quantum information processing platforms, but also to tackle other problems related to quantum tomography and equilibration in closed quantum systems. 


[1] N. Lysne, K. Kuper, P. Poggi, I. Deutsch and P. Jessen, Phys. Rev. Lett. 124 230501 (2020). 

[2] P. Poggi, N. Lysne, K. Kuper, I. Deutsch and P. Jessen, PRX Quantum 1 020308 (2020).