A computational phase diagram for the transverse field Ising model
We study the transverse field Ising model, defined by the Hamiltonian $H =\frac{1}{2}\sum_{i, j\in [n]} J_{ij} Z_i Z_j +\sum_{i=1}^n h_i^z Z_i + η\sum_{i} X_i$ where $J $ is the symmetric interaction matrix, and $η$ is the transverse field strength. Let $Δ(J)=λ_{\max}(J)-λ_{\min}(J)$ be the spectral width of $J.$ When the inverse temperature $β\geq0$ satisfies $Δ(J)\cdot\frac{\tanh(βη)}η\leq1$, we give a randomized classical algorithm that approximates the partition function $Z(β)=\operatorname{Tr}(e^{-βH})$ to a given relative error $ε\in(0,1)$ in time polynomial in $n$, $β$, the model parameters, and $ε^{-1}$. When $ Δ(J) \cdot \frac{\tanh(βη)}η > 1 ,$ we show that approximating $ Z(β)$ within an $\exp(o(n))$-multiplicative factor is $\textbf{NP}$-hard, and thus unlikely to admit an efficient classical or quantum algorithms under standard complexity theoretic assumptions. Furthermore, in the regime $Δ(J)\cdot \frac{\tanh(βη)}η\leq 1,$ we provide an efficient randomized classical algorithm that approximates Pauli string observables of the Gibbs state $ ρ_β= \frac{e^{-βH}}{\operatorname{Tr}(e^{-βH})}$ within an arbitrarily small additive error. In the special case when the observable is also diagonal in the $X$-basis, i.e. $P \in \{I, X\}^{\otimes n}$, the algorithm further achieves arbitrarily small relative error.