From the Physics to the Computational Complexity of Multiboson Correlation Interference

TL;DR

This paper links the physics of multiboson interference with the computational difficulty of simulating such processes, demonstrating that MultiBoson Correlation Sampling is classically hard, highlighting quantum supremacy in optical networks.

Contribution

It establishes the classical hardness of MultiBoson Correlation Sampling for nonidentical photons, emphasizing the fundamental quantum computational advantage.

Findings

MBCS is computationally hard for classical algorithms.

Hardness holds even with nonidentical input photons.

Results support quantum supremacy in linear optical networks.

Abstract

We demonstrate how the physics of multiboson correlation interference leads to the computational complexity of linear optical interferometers based on correlation measurements in the degrees of freedom of the input bosons. In particular, we address the task of MultiBoson Correlation Sampling (MBCS) from the probability distribution associated with polarization- and time-resolved detections at the output of random linear optical networks. We show that the MBCS problem is fundamentally hard to solve classically even for nonidentical input photons, regardless of the color of the photons, making it also very appealing from an experimental point of view. These results fully manifest the quantum computational supremacy inherent to the fundamental nature of quantum interference.