Study of the triangular-lattice Hubbard model with constrained-path quantum Monte Carlo
Shu Fay Ung, Ankit Mahajan, David R. Reichman · 2026
Read Study of the triangular-lattice Hubbard model with constrained-path quantum Monte Carlo by Shu Fay Ung, Ankit Mahajan, David R. Reichman (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-14808.txt.
Introduction
Ung, Mahajan and Reichman benchmark constrained-path Monte Carlo on the triangular-lattice Hubbard model. Symmetry-adapted trial wavefunctions drive the accuracy gains here. Away from half-filling, symmetry-keeping free-electron trials stay within about 1% of exact results. Exact diagonalization and DMRG runs supply the reference values. At half-filling, frustration in mid to strong coupling breaks simple trials. Symmetry-projected trials restore matching accuracy there. Cost grows only polynomially with system size under projection. Frustrated ground-state contests gain a practical tool, they conclude.
Catalog record
| CREATOR | Shu Fay Ung, Ankit Mahajan, David R. Reichman |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 57407 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.14808
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Study of the triangular-lattice Hubbard model with constrained-path quantum Monte Carlo Shu Fay Ung, Ankit Mahajan, David R. Reichman Abstract We benchmark constrained-path Monte Carlo (CPMC) on the triangular-lattice Hubbard model for several fillings and U U values and show that symmetry-adapted trial wave functions are essential for quantitative accuracy. Away from half-filling, simple free-electron-based trials that preserve the ground state symmetry yield energy deviations ≲ 1 % \lesssim 1\% from exact diagonalization and density matrix renormalization group results. At half-filling, strong frustration in the intermediate to large U U regimes necessitates symmetry-projected trials to reach comparable accuracy, where both free-electron and symmetry-broken Hartree-Fock trials incur substantial constraint bias. Since the computational cost of CPMC with symmetry projection scales polynomially with system size, our results motivate its use as a practical route for studying competing ground states in strongly correlated, frustrated systems. I Introduction The Hubbard model has long served as a central paradigm for understanding the emergence of strongly correlated phenomena, ranging from magnetic order and metal-insulator transitions to superconductivity 36 . In recent years, the discovery of triangular moiré superlattices in graphene and transition-metal dichalcogenide (TMD) heterostructures has reinvigorated interest in the triangular-lattice Hubbard model 33 ; 78 ; 29 ; 84 ; 23 ; 60 , which has been shown to capture the low-energy physics of TMD heterobilayers 69 . Extensive study of the model was previously spurred by the discovery of superconductivity and spin-liquid states in organic salts 48 ; 17 ; 73 ; 71 ; 16 ; 72 ; 74 , as it offers a minimal framework for interpreting experiments on strongly correlated, frustrated systems. Theoretically, the triangular-lattice Hubbard model is an important benchmark for many-body computational methods due to the complex interplay of geometric frustration and electronic correlations. At half-filling in the limit of large on-site repulsion U ( i.e. U\to\infty ), the model maps onto the Heisenberg model, which was originally proposed by Anderson to host a quantum spin-liquid ground state 2 . Subsequent numerical studies find that the ground state retains the 120^{\circ} Néel order, albeit with a significantly reduced order parameter compared to the classical prediction 15 ; 67 . Introducing additional frustration, for example via next-nearest-neighbor interactions or ring-exchange terms, has been shown to melt the magnetic order and stabilize spin-liquid phases 31 ; 45 ; 20 ; 86 ; 6 . In the limit of small U , the ground state is metallic; the intermediate- U regime, where charge fluctuations remain, is still under intense debate. The existence of a non-magnetic insulating phase has been predicted by several computational methods including the path-integral renormalization group 30 ; 76 , variational cluster approximation 42 , exact diagonalization (ED) 22 ; 21 , and the ladder dual-fermion approach 26 . Recent DMRG studies 57 ; 58 ; 4 provide evidence that the intermediate phase is a chiral spin liquid, while earlier DMRG results 49 challenged this and variational Monte Carlo studies 63 ; 64 instead suggest the absence of a spin-liquid phase. A recent “multi-method, multi-messenger” approach 68 also finds competing chiral and magnetic orders at intermediate U . In the doped regime relevant to many triangular-lattice materials, DMRG calculations have revealed a rich phase diagram comprising chiral metallic, spin density wave, and enhanced pair-pair correlation phases as doping and interaction strength are varied 85 . Significant effort has focused on superconductivity and its pairing symmetries 5 ; 11 ; 7 ; 66 , driven by the discovery of superconductivity in organic salts 73 ; 71 and cobaltates 59 . Given the complexity of the phase diagram, it is valuable to compare results across complementary numerical approaches. Quantum Monte Carlo (QMC) techniques have played a central role in elucidating the Hubbard model, especially on the square lattice, where the fermion sign problem can be eliminated at half-filling 27 . Due to its non-bipartite nature, one generically encounters a sign problem on the triangular lattice, including at half-filling. Constrained-path Monte Carlo (CPMC), a projector QMC method, is particularly well-suited for the Hubbard interaction and has been extensively benchmarked on the square lattice 80 ; 82 ; 3 ; 47 ; 46 ; 50 ; 38 ; 37 ; 83 . In contrast, its performance has not been systematically assessed to a comparable extent on the triangular lattice, with only a few recent studies 75 ; 77 . In this paper, we present a detailed CPMC study of the triangular-lattice Hubbard model at several fillings. We show that the use of appropriate symmetry-adapted trial wave functions is essential for obtaining accurate results, especially at half-filling where strong frustration makes the problem particularly challenging. Our results help clarify the ground-state properties of the triangular-lattice Hubbard model and demonstrate the effectiveness of symmetry projection in CPMC simulations of frustrated systems. This provides a potentially powerful route to exploring emergent states of the model in a controlled manner as the thermodynamic limit is approached. The paper is organized as follows. We first introduce the model, the CPMC method, and our symmetry-adapted trial states in Sec. II . We then present results in Sec. III and conclude with a discussion in Sec. IV . II Model and methods II.1 The Hubbard model (a) XC 4\times 4 (b) YC 4\times 4 Figure 1 : Triangular-lattice clusters (left) and corresponding one-electron spectra (eigenvalues of \hat{H}_{1} ) grouped by D_{2} irreps (right). The black outline shows a 4\times 4 region with periodic boundaries across the dashed lines. Note that while the symmetry groups of the XC 4\times 4 and YC 4\times 4 lattices are D_{4} and D_{8} , respectively, one can label the orbitals using irreps of the D_{2} subgroup (see App. C for details). The Hubbard model is described by the lattice Hamiltonian \displaystyle\hat{H}=\underbrace{-t\sum_{\sigma}\sum_{\braket{ij}}\left[\hat{c}^{\dagger}_{i\sigma}\hat{c}_{j\sigma}+\text{h.c.}\right]}_{\hat{H}_{1}}+\underbrace{U\sum_{i}\hat{n}_{i\uparrow}\hat{n}_{i\downarrow}}_{\hat{H}_{2}}, (1) where \hat{c}_{i,\sigma}^{\dagger}\left(\hat{c}_{i,\sigma}\right) is the creation (annihilation) operator for an electron at site i and spin \sigma\in\{\uparrow,\downarrow\} , \hat{n}_{i\sigma}=\hat{c}_{i\sigma}^{\dagger}\hat{c}_{i\sigma} is the number operator, t\geq 0 is the hopping parameter, U\geq 0 is the interaction strength, and \sum_{\braket{ij}} enumerates nearest-neighbor sites only. We define the filling factor \nu=(N_{\uparrow}+N_{\downarrow})/N_{s} , where N_{\sigma} denotes the number of spin- \sigma electrons and N_{s} the number of sites. In this work, we focus on ground states in the \hat{S}_{z}=0 spin sector ( i.e. N_{\uparrow}=N_{\downarrow} ) and the hole-doped to half-filling regime \nu\leq 1 . We set t=1 in all calculations reported. We perform our CPMC calculations on finite cylinders with periodic (open) boundary conditions and N_{y} ( N_{x} ) sites along the y - ( x -) direction, giving a total number of N_{s}=N_{x}N_{y} sites. This geometry facilitates direct comparison with DMRG results and reduces open-shell effects associated with degeneracies in the one-electron spectrum, although it is not strictly required for CPMC. For the triangular lattice, two commonly considered configurations are the XC n and YC n cylinders, where n denotes the number of bonds along the y -direction 49 . Illustrative examples are depicted in Fig. 1 . II.2 Constrained-path Monte Carlo CPMC is a projector quantum Monte Carlo method that filters out the ground state \mathinner{|{\Psi_{0}}\rangle} from an initial state \mathinner{|{\Phi^{(0)}}\rangle} (such that \braket{\Psi_{0}|\Phi^{(0)}}\neq 0 ) via imaginary-time evolution, \displaystyle\mathinner{|{\Psi_{0}}\rangle}\propto\lim_{\tau\to\infty}e^{-\tau\hat{H}}\mathinner{|{\Phi^{(0)}}\rangle}. (2) In practice, the long-time propagator is first discretized into n time steps \Delta\tau=\tau/n and Trotterized to obtain