Describing correlated electron systems near phase transitions has been a major challenge in computational condensed-matter physics. In this paper, we apply highly accurate fixed-node quantum Monte Carlo techniques, which directly work with many-body wave functions and simulate electron correlations, to investigate the metal-to-insulator transition of a correlated hydrogen lattice. By calculating spin and charge properties, and analyzing the low-energy Hilbert space, we identify the transition point and identify order parameters that can be used to detect the transition. Our results provide a benchmark for density functional theories seeking to treat correlated electron systems.
Describing correlated electron systems near phase transitions has been a major challenge in computational condensed-matter physics. In this paper, we apply highly accurate fixed-node quantum Monte Carlo techniques, which directly work with many-body wave functions and simulate electron correlations, to investigate the metal-to-insulator transition of a correlated hydrogen lattice. By calculating spin and charge properties, and analyzing the low-energy Hilbert space, we identify the transition point and identify order parameters that can be used to detect the transition. Our results provide a benchmark for density functional theories seeking to treat correlated electron systems.
Describing correlated electron systems has been a major challenge in computational condensed-matter physics. Quantum Monte Carlo, a powerful computational tool for the study of correlated systems, solves electron correlation problems explicitly. It has been taken as a benchmark method for understanding the correlated systems. Instead of making approximations to Hamiltonian, QMC methods work with the wave functions, and the computational cost scales well with the system size. With the development of parallel computing, QMC calculations on large systems are becoming more and more feasible. We have investigated two correlated systems with highly accurate fixed node QMC techniques. The first system is a correlated hydrogen model system near the metal to insulator transition. We have successfully identified the transition point by calculating spin and charge properties and analyzing the low energy Hilbert space. The second one is a strongly correlated Fe/O system. Calculations on the Fe atoms, O atoms, and FeO molecules are conducted with multiple highly accurate many-body techniques. The source of errors has been disentangled by comparing the results of the many body techniques with the experimental results. For the Fe and O atoms, the calculated properties coincide well with previous experimental results. For the basis-based techniques, the performance is mainly limited by the basis set. The calculated equilibrium bond length, excitation energy and vibrational frequency of the FeO molecules are also in close agreement with the known values from previous experiments.
The many-body correlation between electrons is the origin of many fascinating phenomena in condensed matter systems, such as high temperature superconductivity, superfluidity, fractional quantum Hall effect, and Mott insulator. Strongly correlated systems have been an important subject of condensed matter physics for several decades, especially after the discovery of high temperature cuprate superconductors. In this thesis, we apply first principles quantum Monte Carlo (QMC) method to several representative systems to study the electron correlations in transition metal oxides (vanadium dioxide) and low dimensional electronic systems (graphene and graphene-like two dimensional systems). Vanadium dioxide (VO2) is a paradigmatic example of a strongly correlated system that undergoes a metal-insulator transition at a structural phase transition. To date, this transition has necessitated significant post-hoc adjustments to theory in order to be described properly. We apply first principles quantum Monte Carlo (QMC) to study the structural dependence of the properties of VO2. Using this technique, we simulate the interactions between electrons explicitly, which allows for the metal-insulator transition to naturally emerge, importantly without ad-hoc adjustments. The QMC calculations show that the structural transition directly causes the metal-insulator transition and a change in the coupling of vanadium spins. This change in the spin coupling results in a prediction of a momentum-independent magnetic excitation in the insulating state. While two-body correlations are important to set the stage for this transition, they do not change significantly when VO2 becomes an insulator. These results show that it is now possible to account for electron correlations in a quantitatively accurate way that is also specific to materials. Electron correlation in graphene is unique because of the interplay of the Dirac cone dispersion of pi electrons with long range Coulomb interaction. The random phase approximation predicts no metallic screening at long distances and low energies because of the zero density of states at Fermi level. It is thus interesting to see how screening takes place in graphene at different length scales. We addressed this problem by computing the structure factor S(q) and S(q, ω) of freestanding graphene using ab initio fixed-node diffusion Monte Carlo and the random phase approximation. The X-ray measured structure factor is reproduced very accurately using both techniques, provided that sigma-bonding electrons are included in the simulations. Strong dielectric screening from sigma electrons are observed, which redshifts the pi plasmons resonance frequency at long distance and reduces the effective interactions between pi electrons at short distance. The short distance screening makes suspended graphene a weakly correlated semimetal which otherwise would be an insulator. The third piece of works is dedicated to studying the low energy excitation of many-body systems using extended Koopmans' theorem (EKT). The EKT provides a straight forward way to compute charge excitation spectra, such as ionization potentials, electron affinities from any level of theory. We implemented the EKT within the QMC framework, and performed systematic benchmark studies of ionization potentials of the second- and third-row atoms, and closed- and open-shell molecules. We also applied it to compute the quasiparticle band structure of solids (graphene). For complex correlated systems, identifying relevant low energy physics degrees of freedom is extremely important to understanding the system's collective behavior at different length scales. In this sense, bridging the realistic systems to lower energy effective lattice models that involve fewer but important degrees of freedom is significant to understanding correlated systems. We have formulated three ab initio density matrix based downfolding (AIDMD) methods to downfold the ab initio systems into effective lattice models. We have demonstrated the successfulness of these methods by applying them to molecules (H2) and periodic systems (hydrogen chain and graphene).
Semiconductor artificial graphene nanostructures where the Hubbard model parameter U/t can be of the order of 100, provide a highly controllable platform to study strongly correlated quantum many-particle phases. We use accurate variational and diffusion Monte Carlo methods to demonstrate a transition from antiferromagnetic to metallic phases for an experimentally accessible lattice constant a=50 nm in terms of lattice site radius $\ensuremath{\rho}$, for finite-sized artificial honeycomb structures nanopatterned on GaAs quantum wells containing up to 114 electrons. By analyzing spin-spin correlation functions for hexagonal flakes with armchair edges and triangular flakes with zigzag edges, we show that edge type, geometry, and charge nonuniformity affect the steepness and the crossover $\ensuremath{\rho}$ value of the phase transition. For triangular structures, the metal-insulator transition is accompanied with a smoother edge polarization transition.
In this thesis we apply recently developed, as well as sophisticated quantum Monte Carlo methods to numerically investigate models of strongly correlated electron systems on honeycomb structures. The latter are of particular interest owing to their unique properties when simulating electrons on them, like the relativistic dispersion, strong quantum fluctuations and their resistance against instabilities. This work covers several projects including the advancement of the weak-coupling continuous time quantum Monte Carlo and its application to zero temperature and phonons, quantum phase transitions of valence bond solids in spin-1/2 Heisenberg systems using projector quantum Monte Carlo in the valence bond basis, and the magnetic field induced transition to a canted antiferromagnet of the Hubbard model on the honeycomb lattice. The emphasis lies on two projects investigating the phase diagram of the SU(2) and the SU(N)-symmetric Hubbard model on the hexagonal lattice. At sufficiently low temperatures, condensed-matter systems tend to develop order. An exception are quantum spin-liquids, where fluctuations prevent a transition to an ordered state down to the lowest temperatures. Previously elusive in experimentally relevant microscopic two-dimensional models, we show by means of large-scale quantum Monte Carlo simulations of the SU(2) Hubbard model on the honeycomb lattice, that a quantum spin-liquid emerges between the state described by massless Dirac fermions and an antiferromagnetically ordered Mott insulator. This unexpected quantum-disordered state is found to be a short-range resonating valence bond liquid, akin to the one proposed for high temperature superconductors. Inspired by the rich phase diagrams of SU(N) models we study the SU(N)-symmetric Hubbard Heisenberg quantum antiferromagnet on the honeycomb lattice to investigate the reliability of 1/N corrections to large-N results by means of numerically exact QMC simulations. We study the melting of phases as correlations increase with decreasing N and determine whether the quantum spin liquid found in the SU(2) Hubbard model at intermediate coupling is a specific feature, or also exists in the unconstrained t-J model and higher symmetries.
This article describes the variational and fixed-node diffusion quantum Monte Carlo methods and how they may be used to calculate the properties of many-electron systems. These stochastic wave-function-based approaches provide a very direct treatment of quantum many-body effects and serve as benchmarks against which other techniques may be compared. They complement the less demanding density-functional approach by providing more accurate results and a deeper understanding of the physics of electronic correlation in real materials. The algorithms are intrinsically parallel, and currently available high-performance computers allow applications to systems containing a thousand or more electrons. With these tools one can study complicated problems such as the properties of surfaces and defects, while including electron correlation effects with high precision. The authors provide a pedagogical overview of the techniques and describe a selection of applications to ground and excited states of solids and clusters.
The metal–insulator transition (MIT) is a fundamental phenomenon in condensed matter physics and a hallmark of strong electronic correlations. Hydrogen-based systems offer a simple yet powerful model for investigating the MIT, as their insulating behavior arises purely from electron–electron interactions. In this work, we study finite hydrogen clusters with cubic geometries using Natural Orbital Functional Theory (NOFT), a method capable of accurately describing correlated systems beyond mean-field approaches. We focus on two key signatures of the MIT: the fundamental energy gap and the harmonic average of the atomic one-particle reduced density matrix. Our results show that NOFT captures the transition from insulating to metallic behavior as the interatomic distance decreases. By extrapolating the energy gap to the thermodynamic limit, we estimate a critical distance rc ~1.2 Å, in excellent agreement with quantum Monte Carlo benchmarks. These findings demonstrate the reliability of NOFT for describing strong correlation effects in large-scale models.