Tom de CONINCK

PhD Student

Resarch activities at CPHT: Condensed Matter

Research interests:  Condensed Matter Physics, Correlated Quantum Systems, Quantum Computing, Computational Physics

Thesis: "Hybrid quantum/classical methods for the study of strongly correlated quantum systems."
Advisors: Michel Ferrero and Thomas Ayral

Abstract

The study of strongly correlated fermion systems, such as Hubbard model, is a challenging problem in condensed matter physics, particularly due to the emergence of the "sign problem" when computing the ground state. When using Monte Carlo methods, this problem manifests itself through the exponential growth of the variance of statistical estimators. The so-called "constrained path" approximation, used in conjunction with the AFQMC (Auxiliary Field Quantum Monte Carlo) algorithm, makes it possible to eliminate this phenomenon, at the cost of an approximation whose accuracy increases the closer the chosen reference state is to the ground state of the system. The classical implementation of this method is, however, restricted to quantum states that are easily representable and usable on a classical processor, such as Slater determinants. This thesis investigates how the use of quantum algorithms such as the so-called "shadow tomography" protocol can improve the classical CP-AFQMC method applied to the Hubbard model by diversifying the usable reference states. The objective is thus to study the influence of an improvement of the reference quantum state on the constrained path approximation in the framework of the quantum many-body problem.