Offloading Quantum Computation by Superposition Masking

TL;DR

This paper introduces a superposition masking technique to offload quantum computations to classical processors, reducing quantum overhead for certain tasks, but with limitations in uncomputing and inversion.

Contribution

The paper presents a novel superposition masking method enabling classical offloading of quantum tasks, improving efficiency for specific computations.

Findings

Achieves at least a constant-factor reduction in quantum operations for targeted problems.

Demonstrates applicability to modular inversion, root-finding, and matrix inversion.

Highlights limitations in uncomputing and inverting the superposition masking technique.

Abstract

Error correction will add so much overhead to large quantum computations that we suspect the most efficient algorithms will use a classical co-processor to do as much work as possible. We present a method to offload portions of a quantum computation to a classical computer by producing a superposition of masks which hide a quantum input. With the masks, we can measure the result without altering the original input and then perform classical computations on the measured output. If the task has enough structure, the classical computations will be equivalent to a quantum computation performed in superposition. We apply this technique to modular inversion, root-finding, division with remainder, sparse matrix inversion, and inverting generic group homomorphisms, achieving at least a constant-factor improvement in quantum operations for each. Unfortunately, it is difficult to uncompute or invert this technique because of the measurement, and thus we know of no useful algorithm which benefits from superposition masking.