ORCID ID

https://orcid.org/0000-0002-1953-1579

Date Awarded

2023

Document Type

Dissertation

Degree Name

Doctor of Philosophy (Ph.D.)

Department

Physics

Advisor

Shiwei SZ Zhang

Committee Member

Shiwei SZ Zhang

Committee Member

Enrico ER Rossi

Committee Member

Konstantinos KO Orginos

Committee Member

Seth SA Aubin

Abstract

The study of interacting quantum many-body systems poses one of the main challenges in areas including condensed matter physics, nuclear physics, cold atoms physics, quantum chemistry, and materials science. Currently, no general approach is capable of handling the full complexity of interacting quantum systems, providing systematically accurate results across different ranges of many-body models and materials. The continued development of more general and more accurate numerical methodologies is instrumental in meeting the challenges of understanding and predicting the properties of interacting quantum systems. Quantum Monte Carlo (QMC) methods represent an important class of many-body techniques extensively employed in studying correlated quantum systems. However, their effective application to non-trivial quantum systems requires specialized adaptations to ensure sufficient sampling efficiency. In this thesis, we introduce novel advancements and applications of the auxiliary-field quantum Monte Carlo (AFQMC) methods on Hubbard Model, atoms/molecules, and magic-angle twisted bilayer graphene. One of our development is the construction of pseudo-BCS wave functions using the one-body density matrix, offering a systematically improvable ansatz for correlated fermion systems. In AFQMC calculations, these pseudo-BCS wave functions serve as trial wave functions to control the sign/phase problem, using the two-dimensional Hubbard model as an example. Furthermore, we propose an interface between branching random walks and Markov chain Monte Carlo sampling, enabling seamless switching between these two. In the context of AFQMC, this interface facilitates a smooth transition from constrained-path sampling to constraint release, reducing systematic errors. We demonstrate this method in atoms and molecules, where improvements in accuracy can be clearly quantified and near-exact results are obtained. Leveraging the above developments, we apply AFQMC to investigate magic-angle twisted bilayer graphene (TBG) by studying a many-body Hamiltonian which consists of Bistritzer and MacDonald's single-particle model and realistic electron-electron Coulomb interactions. As our state-of-art AFQMC approach allows accurate treatment of the many-body Hamiltonian while controlling the sign problem, we determine the ground-state properties of this model across different integer fillings and quantify the errors from mean-field theory calculations. Overall, our developments in AFQMC offer broad applicability ranging from the Hubbard Model to Coulomb interactions which pave the way for intense studies encompassing abstract models, correlated materials with direct experimental comparisons, and what is in between involving both model and realistic interactions.

DOI

https://dx.doi.org/10.21220/s2-wr78-eg46

Rights

© The Author

Included in

Physics Commons

Share

COinS