Southwest Quantum Information and Technology

Tensor network simulations of variational Bayesian metrology with correlated noise

Presenting Author: Tyler Thurtell, University of New Mexico CQuIC
Contributing Author(s): Akimasa Miyake

We consider variational metrology in the global, or Bayesian, framework. We first introduce a new family of ansatzes which we call arbitrary-axis twist ansatzes. This family of ansatzes amounts to taking both the encoding and decoding circuits to be an alternating sequence of rotations and one-axis twists about arbitrary directions. We find that this family of ansatzes can preform at least as well, and in some cases better, than previous approaches despite having fewer entangling one-axis twists. We also study these ansatzes in the presence of spatially correlated noise. This breaks the permutation symmetry of the noiseless dynamics meaning that symmetric subspace techniques cannot be used. To facilitate this numerical study, we introduce a matrix product operator based simulation scheme. As long as the matrix product operator associated with the noise has a bond dimension that is at most polynomial in the size of the system, the result is an exact simulation algorithm that has cost polynomial in the system size but exponential in the number segments of the evolution that break the permutation symmetry. In addition, we use these techniques to study the effect of various types of spatially correlated noise in twist-untwist protocols.

(Session 5 : Thursday from 5:00 pm - 7:00 pm)