All Or Nothing: No-Downfolding Theorems For Quantum Simulation

TL;DR

This paper proves that exact downfolding of quantum Hamiltonians into lower dimensions, which would simplify quantum simulations, is fundamentally impossible for the most common formulations, impacting future quantum information research.

Contribution

The paper establishes a no-go theorem showing the impossibility of exact eigenvalue-preserving downfolding for parameter-dependent Hamiltonians in quantum systems.

Findings

Exact quantum downfolding cannot recover all eigenvalue functions.

The result applies to Hamiltonians of the form A + xB with parameter x.

Implications for quantum simulation and information processing.

Abstract

The physics of a quantum system with many degrees of freedom is often approximated by downfolding: most of the degrees of freedom are "folded into" a much smaller number of degrees of freedom, resulting in an effective Hamiltonian that still captures the essential physics. Approaches of this sort are particularly relevant for quantum information, where exact downfolding would allow eigenstates in a large Hilbert space to be simulated with fewer qubits. Very little work has been done to prove the existence of such an exact downfolding for general systems or even particular cases. In this letter we prove that exact quantum downfolding is impossible for what is perhaps the most commonly-used formulation of the task. Specifically, for a non-trivial Hamiltonian $\mathbf{H}(x)\equiv\mathbf{A}+x\mathbf{B}$ that depends on some parameter $x$ (e.g. an electric field, bond length or interaction strength) it is not possible to construct a lower-dimension effective Hamiltonian $\mathbf{h}(x)\equiv\mathbf{a}+x\mathbf{b}$ that exactly recovers \emph{any} of the eigenvalue functions $E_i(x)$ of $\mathbf{H}(x)$. We discuss several generalizations of this result and the impacts of these findings on future directions in quantum information.