No Arabic abstract
The repulsive fermionic Hubbard model is a typical model describing correlated electronic systems. Although it is a simple model with only a kinetic term and a local interaction term, their competition generates rich phases. When the interaction part is significant, usually in many strongly correlated, flat or narrow band systems, lots of novel correlated phases may emerge. One way to understand the possible correlated phases is to go beyond finite interaction and solve the infinite-$U$ Hubbard model. Solving infinite-$U$ Hubbard model is usually extremely hard, and a large-scale unbiased numerical study is still missing. In this Letter, we propose a projection approach, such that a controllable quantum Monte Carlo (QMC) simulation on infinite-$U$ Hubbard model may be done at some integer fillings where either it is sign problem free or surprisingly has an algebraic sign structure -- a power law dependence of average sign on system size. We demonstrate our scheme on the infinite-$U$ $SU(2N)$ fermionic Hubbard model on both square and honeycomb lattice at half-filling, where it is sign problem free, and suggest possible correlated ground states. The method can be generalized to study certain extended Hubbard models applying to cluster Mott insulators or 2D Morie systems, among one of them at certain non-half integer filling, the sign has an algebraic behavior such that it can be numerically solved within a polynomial time. Further, our projection scheme can also be generalized to implement the Gutzwiller projection to spin basis such that $SU(2N)$ quantum spin models and Kondo lattice models may be studied in the framework of fermionic QMC simulations.
Lattice Monte Carlo calculations of interacting systems on non-bipartite lattices exhibit an oscillatory imaginary phase known as the phase or sign problem, even at zero chemical potential. One method to alleviate the sign problem is to analytically continue the integration region of the state variables into the complex plane via holomorphic flow equations. For asymptotically large flow times the state variables approach manifolds of constant imaginary phase known as Lefschetz thimbles. However, flowing such variables and calculating the ensuing Jacobian is a computationally demanding procedure. In this paper we demonstrate that neural networks can be trained to parameterize suitable manifolds for this class of sign problem and drastically reduce the computational cost. We apply our method to the Hubbard model on the triangle and tetrahedron, both of which are non-bipartite. At strong interaction strengths and modest temperatures the tetrahedron suffers from a severe sign problem that cannot be overcome with standard reweighting techniques, while it quickly yields to our method. We benchmark our results with exact calculations and comment on future directions of this work.
Two-dimensional Hubbard model is very important in condensed matter physics. However it has not been resolved though it has been proposed for more than 50 years. We give several methods to construct eigenstates of the model that are independent of the on-site interaction strength $U$.
We demonstrate that the sign structure of the t-J model on a hypercubic lattice is entirely different from that of a Fermi gas, by inspecting the high temperature expansion of the partition function up to all orders, as well as the multi-hole propagator of the half-filled state and the perturbative expansion of the ground state energy. We show that while the fermion signs can be completely gauged away by a Marshall sign transformation at half-filling, the bulk of the signs can be also gauged away in a doped case, leaving behind a rarified irreducible sign structure that can be enumerated easily by counting exchanges of holes with themselves and spins on their real space paths. Such a sparse sign structure implies a mutual statistics for the quantum states of the doped Mott insulator.
The last decade has seen a large increase in the number of electronic-structure calculations that involve adding a Hubbard term to the local density approximation band-structure Hamiltonian. The Hubbard term is then solved either at the mean-field level or with sophisticated many-body techniques such as dynamical mean field theory. We review the physics underlying these approaches and discuss their strengths and weaknesses in terms of the larger issues of electronic structure that they involve. In particular, we argue that the common assumptions made to justify such calculations are inconsistent with what the calculations actually do. Although many of these calculations are often treated as essentially first-principles calculations, in fact, we argue that they should be viewed from an entirely different point of view, viz., as phenomenological many-body corrections to band-structure theory. Alternatively, they may also be considered to be just a more complex Hubbard model than the simple one- or few-band models traditionally used in many-body theories of solids.
We study the Hubbard model with non-Hermitian asymmetric hopping terms. The conjugate hopping terms are introduced for two spin components so that the negative sign is canceled out. This ensures that the quantum Monte Carlo simulation is free from the negative sign problem. We analyze the antiferromagnetic order and its suppression by the non-Hermiticity.