Quantum state tomography with incomplete data: Maximum entropy and variational quantum tomography

Abstract

Whenever we do not have an informationally complete set of measurements, the estimate of a quantum state can not be uniquely determined. In this case, among the density matrices compatible with the available data, it is commonly preferred that one which is the most uncommitted with the missing information. This is the purpose of the Maximum Entropy estimation (MaxEnt) and the Variational Quantum Tomography (VQT). Here, we propose a variant of Variational Quantum Tomography and show its relationship with Maximum Entropy methods in quantum tomographies with incomplete set of measurements. We prove their equivalence in case of eigenbasis measurements, and through numerical simulations we stress their similar behavior. Hence, in the modified VQT formulation we have an estimate of a quantum state as unbiased as in MaxEnt and with the benefit that VQT can be more efficiently solved by means of linear semidefinite programs.PACS number(s): 03.65.Wj, 03.67.-a *

A quasi-current representation for information needs inspired by Two-State Vector Formalism

“…Thus, the RρR algorithm [26] can be used to solve the optimization problem. The RρR algorithm works in most of cases, but it has 310 no theoretical guarantee of convergence, regardless the dataset and the initial point [17]. So, in this paper, we decide to use another algorithm [17] proposed under a line search procedure with Armijo condition, namely "Diluted RρR".…”

“…During the optimization process, the search direction is a combination of 315 two ascent directions controlled by the step size t. In the paper [17], it has shown that an inexact line search method to determine t is enough for finding a value to guarantee the global convergence. The initial density matrix…”

“…In contrast with [6] which utilizes as much session information as possible to enhance the retrieval performance, we focus on finding necessary information to represent users' current IN from existing session data inspired by Two-State Vector Formalism (TSVF). In addition, we choose another different maximum likelihood estimation (MLE) method which 130 can ensure the global convergence and obtain a Hermitian density matrix [17], see more detail in Section 4.1 and 4.2.…”

Least-bias state estimation with incomplete unbiased measurements

“…In the same spirit, several algorithms for estimating quantum states from incomplete measurement data have been reported recently [19][20][21][22]. Here, we revisit the problem in the context of mutually unbiased (MU) measurements, which are known to be optimal for state reconstruction [23] if a complete set of bases is to be measured.…”

Quantum state tomography and fidelity estimation via Phaselift

“…Recently, efficient quantum state tomography which may overcome these difficult challenges has attracted great attention [19]. The maximum entropy approach can give an approximate estimation for general states [20]. Part parameters, which are much less than the full (d 2 − 1) parameters, are enough to characterize matrix product states (MPS) [19,21,22].…”