Southwest Quantum Information and Technology

Scalable unfolding of measurement errors

Presenting Author: Siddarth Srinivasan, University of Washington
Contributing Author(s): Bibek Pokharel Gregory Quiroz Byron Boots

Measurement error mitigation (MEM) techniques are postprocessing strategies to counteract systematic readout errors on programmable quantum computers. Most MEM strategies model these errors as a linear stochastic map also called the response matrix, and aim to invert the effect of this map. Unfortunately, the response matrix scales exponentially with the number of qubits, and its inverse is not stochastic so the mitigated distribution can contain negative probabilities. There are strategies to address scalability concerns, e.g. by assuming that measurement errors are mostly uncorrelated and that reduced mitigation accuracy is tolerable. However, these scalable strategies still return quasiprobabilities. On the other hand, existing methods that guarantee a non-negative mitigated distribution are not scalable. In this work, we demonstrate a scalable MEM strategy that avoids quasiprobability distributions. In particular, we implement a scalable implementation of iterative Bayesian unfolding, a standard mitigation technique in high-energy physics experiments. We demonstrate our method with experimental preparation of GHZ states up to 127 qubits and the implementation of the Bernstein-Vazirani algorithm on up to 26 qubits.

(Session 5 : Thursday from 5:00 pm - 7:00 pm)