Quantum state tomography: Mean squared error matters, bias does not
arXiv:1405.5350
Abstract
Because of the constraint that the estimators be bona fide physical states, any quantum state tomography scheme - including the widely used maximum likelihood estimation - yields estimators that may have a bias, although they are consistent estimators. Schwemmer et al. (arXiv:1310.8465 [quant-ph]) illustrate this by observing a systematic underestimation of the fidelity and an overestimation of entanglement in estimators obtained from simulated data. Further, these authors argue that the simple method of linear inversion overcomes this (perceived) problem of bias, and there is the suggestion to abandon time-tested estimation procedures in favor of linear inversion. Here, we discuss the pros and cons of using biased and unbiased estimators for quantum state tomography. We conclude that the little occasional benefit from the unbiased linear-inversion estimation does not justify the high price of using unphysical estimators, which are typically the case in that scheme.
7 pages, 4 figures, 1 table, 10 references
Cited by in corpus (5)
- Pseudo-Bayesian Quantum Tomography with Rank-adaptation
- Optimal error intervals for properties of the quantum state
- Investigating Bias in Maximum Likelihood Quantum State Tomography
- An efficient adaptive MCMC algorithm for Pseudo-Bayesian quantum tomography
- Least-bias state estimation with incomplete unbiased measurements