Uncomputability of Phase Diagrams

TL;DR

This paper proves that determining the quantum phase diagram of a many-body Hamiltonian is fundamentally uncomputable, even for continuous, translationally invariant systems, highlighting limits of algorithmic predictability in condensed matter physics.

Contribution

It establishes the undecidability of the spectral gap for a continuous family of Hamiltonians, extending previous results to a broader, positive measure set of parameters.

Findings

Uncomputability of phase diagrams demonstrated for continuous Hamiltonian families.

Spectral gap undecidability proven for translationally invariant 2D spin systems.

Undecidability applies to a positive measure set of parameters.

Abstract

The phase diagram of a material is of central importance to describe the properties and behaviour of a condensed matter system. We prove that the general task of determining the quantum phase diagram of a many-body Hamiltonian is uncomputable, by explicitly constructing a one-parameter family of Hamiltonians for which this is the case. This work builds off recent results from Cubitt et al. and Bausch et al., proving undecidability of the spectral gap problem. However, in all previous constructions, the Hamiltonian was necessarily a discontinuous function of its parameters, making it difficult to derive rigorous implications for phase diagrams or related condensed matter questions. Our main technical contribution is to prove undecidability of the spectral gap for a continuous, single-parameter family of translationally invariant, nearest-neighbour spin-lattice Hamiltonians on a 2D square lattice: $H(\varphi)$ where $\varphi\in \mathbb R$. As well as implying uncomputablity of phase diagrams, our result also proves that undecidability can hold for a set of positive measure of a Hamiltonian's parameter space, whereas previous results only implied undecidability on a zero measure set.