Sample optimal tomography of quantum Markov chains

TL;DR

This paper establishes dimension-independent robustness of the Petz map for quantum Markov chains and derives optimal sample complexities for their tomography and certification, extending results to multipartite systems.

Contribution

It provides the first dimension-independent robustness analysis of the Petz map and derives optimal sample complexities for quantum Markov chain tomography and certification.

Findings

Robustness results are dimension-independent for infidelity and trace distance.

Sample complexity for quantum Markov chain tomography is near-optimal and explicitly characterized.

Extended tomography results to multipartite quantum systems with scalable sample complexity.

Abstract

A state on a tripartite quantum system $\mathcal{H}_{A}\otimes \mathcal{H}_{B}\otimes\mathcal{H}_{C} $ forms a Markov chain, i.e., quantum conditional independence, if it can be reconstructed from its marginal on $\mathcal{H}_{A}\otimes \mathcal{H}_{B}$ by a quantum operation from $\mathcal{H}_{B}$ to $\mathcal{H}_{B}\otimes\mathcal{H}_{C}$ via the famous Petz map: a quantum Markov chain $\rho_{ABC}$ satisfies $\rho_{ABC}=\rho_{BC}^{1/2}(\rho_B^{-1/2}\rho_{AB}\rho_B^{-1/2}\otimes id_C)\rho_{BC}^{1/2}$. In this paper, we study the robustness of the Petz map for different metrics, i.e., the closeness of marginals implies the closeness of the Petz map outcomes. The robustness results are dimension-independent for infidelity $\delta$ and trace distance $\epsilon$. The applications of robustness results are The sample complexity of quantum Markov chain tomography, i.e., how many copies of an unknown quantum Markov chain are necessary and sufficient to determine the state, is $\tilde{\Theta}(\frac{(d_A^2+d_C^2)d_B^2}{\delta})$, and $\tilde{\Theta}(\frac{(d_A^2+d_C^2)d_B^2}{\epsilon^2}) $. The sample complexity of quantum Markov Chain certification, i.e., to certify whether a tripartite state equals a fixed given quantum Markov Chain $\sigma_{ABC}$ or at least $\delta$-far from $\sigma_{ABC}$, is ${\Theta}(\frac{(d_A+d_C)d_B}{\delta})$, and ${\Theta}(\frac{(d_A+d_C)d_B}{\epsilon^2})$. $\tilde{O}(\frac{\min\{d_Ad_B^3d_C^3,d_A^3d_B^3d_C\}}{\epsilon^2})$ copies to test whether $\rho_{ABC}$ is a quantum Markov Chain or $\epsilon$-far from its Petz recovery state. We generalized the tomography results into multipartite quantum system by showing $\tilde{O}(\frac{n^2\max_{i} \{d_i^2d_{i+1}^2\}}{\delta})$ copies for infidelity $\delta$ are enough for $n$-partite quantum Markov chain tomography with $d_i$ being the dimension of the $i$-th subsystem.