Classical simulability of noisy boson sampling

TL;DR

This paper presents an algorithm that simulates noisy boson sampling efficiently, showing that current photonic technology cannot surpass the noise threshold needed to demonstrate quantum advantage, thus challenging its viability as a quantum supremacy test.

Contribution

The authors develop a classical simulation algorithm for noisy boson sampling, establishing a noise threshold for quantum advantage demonstration and assessing current technology against this benchmark.

Findings

Current photonic systems are below the noise threshold for quantum advantage.

Classical simulation becomes efficient at certain noise levels.

Boson sampling may not be suitable for demonstrating quantum supremacy under realistic noise conditions.

Abstract

Quantum mechanics promises computational powers beyond the reach of classical computers. Current technology is on the brink of an experimental demonstration of the superior power of quantum computation compared to classical devices. For such a demonstration to be meaningful, experimental noise must not affect the computational power of the device; this occurs when a classical algorithm can use the noise to simulate the quantum system. In this work, we demonstrate an algorithm which simulates boson sampling, a quantum advantage demonstration based on many-body quantum interference of indistinguishable bosons, in the presence of optical loss. Finding the level of noise where this approximation becomes efficient lets us map out the maximum level of imperfections at which it is still possible to demonstrate a quantum advantage. We show that current photonic technology falls short of this benchmark. These results call into question the suitability of boson sampling as a quantum advantage demonstration.