Complexity of Quadratic Bosonic Hamiltonian Simulation: $\mathsf{BQP}$-Completeness and $\mathsf{PostBQP}$-Hardness

TL;DR

This paper investigates the computational complexity of simulating large bosonic quantum systems with quadratic Hamiltonians, establishing BQP-completeness for certain classes and PostBQP-hardness for more general cases.

Contribution

It introduces a broad class of quadratic bosonic problems that are BQP-complete and demonstrates that more general cases are PostBQP-hard, highlighting a complexity transition.

Findings

Certain quadratic bosonic problems are BQP-complete.

Extending to more general Hamiltonians leads to PostBQP-hardness.

Shows a sharp complexity transition in simulating quantum systems.

Abstract

The computational complexity of simulating the dynamics of physical quantum systems is a central question at the interface of quantum physics and computer science. In this work, we address this question for the simulation of exponentially large bosonic Hamiltonians with quadratic interactions. We present two results: First, we introduce a broad class of quadratic bosonic problems for which we prove that they are $\mathsf{BQP}$-complete. Importantly, this class includes two known $\mathsf{BQP}$-complete problems as special cases: Classical oscillator networks and continuous-time quantum walks. Second, we show that extending the aforementioned class to even more general quadratic Hamiltonians results in a $\mathsf{PostBQP}$-hard problem. This reveals a sharp transition in the complexity of simulating large quantum systems on a quantum computer, as well as in the difference in complexity between their simulation on classical and quantum computers.