Southwest Quantum Information and Technology

How to benchmark quantum computers using any quantum algorithm

Presenting Author: Stefan Seritan, Sandia National Laboratories
Contributing Author(s): Antonio Russo, Aidan Wilber-Gauthier, Kenneth Rudinger, Timothy Proctor, Robin Blume-Kohout, Andrew Baczewski

Quantum computers promise better solutions to real-world problems, such as those arising in quantum chemistry, but only if they can successfully run large, complex programs implementing specific quantum algorithms. There is thus an urgent need for benchmarks that measure how well quantum computers can execute such programs. However, this appears impossible because quantum programs solving difficult problems are (1) too big to fit on current-generation quantum testbeds, and (2) seemingly impossible to verify on classical computers. We overcome both problems by applying circuit mirroring and subcircuit snipping to generate the first scalable, efficiently verifiable application-inspired volumetric benchmarks. We demonstrate our technique on a key subroutine for quantum chemistry algorithms: the application of a block-encoded second quantized Hamiltonian. While this is more resource intensive than near-term approaches to the same task, it is more representative of what will be executed on future fault-tolerant systems. Experiments were performed for small instances, i.e., minimal basis H2, HeH+, and LiH, using two different fermion-to-qubit mappings. We compare the performance of several IBM Q devices on our application-inspired benchmarks to their performance on prior mirror circuit benchmarks. We also validate our results against simulations using error models containing coherent and stochastic noise. SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.

(Session 13 : from 4:45 pm - 5:15 pm)