arXiv:2609.18761
Exact and fast series expansions for quantum models with long-range interactions
Antonia Duft, Patrick Adelhardt, Jan Alexander Koziol, Andreas A. Buchheit, Kai Phillip Schmidt
Abstract
Over the past decade, high-order series expansions based on linked-cluster methods have become an important tool for studying low-energy properties of gapped quantum systems with long-range interactions. We introduce a deterministic framework that removes a central computational bottleneck of this method. Our graph zeta method replaces the costly and statistically noisy Monte Carlo (MC) evaluation of high-dimensional lattice sums by a systematic, high-precision computation that delivers series coefficients within minutes on standard desktop hardware. The full momentum-dependent series is obtained in a single calculation, enabling high-resolution excitation spectra throughout the Brillouin zone. Building on the companion paper [1], the method reformulates graph-embedding sums as graph zeta functions and decomposes them into blocks classified by their treewidth tw. Low-treewidth blocks (tw$\leq2$) admit closed expressions based on Epstein zeta functions, while higher-treewidth blocks (tw$>2$) are evaluated using tensor-network bucket elimination. We benchmark the approach for transverse-field Ising models with power-law interactions in 1d, 2d, and 3d, reproducing previous MC results at a fraction of the computational cost while enabling substantially denser parameter sampling. An open-source implementation makes the method directly applicable to general interactions and large parameter scans. As an application, we compare microscopic interaction models for the stacked quasi-2d transverse-field Ising triangular-lattice antiferromagnet KTmSe$_2$ and find that a model including dipolar interactions best describes existing experimental data. The graph zeta method thus turns high-order linked-cluster expansions into a practical and deterministic tool for fast quantitative momentum-resolved modeling of short- and long-range quantum matter.
AI Summary
Linked-cluster expansions for long-range quantum models require high-dimensional lattice-embedding sums, previously evaluated using costly Monte Carlo sampling. This paper develops a deterministic replacement for translationally invariant interactions on Bravais lattices, targeting ground-state energies and one-quasiparticle dispersions in the thermodynamic limit. An exact combinatorial “softcore mapping” removes vertex-exclusion constraints through Möbius inversion over graph contractions. The resulting graph zeta functions factorize at articulation points into reusable blocks. Blocks of augmented treewidth $\mathrm{tw}\leq2$ are evaluated through Epstein zeta functions and their products and convolutions; higher-treewidth blocks use tensor-network bucket elimination and extrapolation. Mathematical proofs and convergence analysis are delegated to a companion paper. An open-source implementation computes the full momentum-grid dispersion in one calculation. Benchmarks cover power-law transverse-field Ising models on chain, square, triangular and cubic lattices, reaching order 13 for ground-state energies and order 11 for dispersions, with maximum treewidth four. Coefficients generally agree with published Monte Carlo estimates within reported uncertainties. Representative three-dimensional calculations take about 200 seconds on one desktop core. Reduced costs enable denser interaction-exponent scans. Fits to digitized neutron-scattering spectra of $\mathrm{KTmSe}_2$ favor nearest-neighbor exchange plus dipolar interactions over the tested short-range descriptions. “Numerically exact” means sampling-free, not free of discretization error. Perturbative truncation and critical-point extrapolation remain limitations. Semi-analytic singularity treatment is unavailable above treewidth two; multi-site unit cells and additional observables remain future extensions. Three-dimensional anisotropic dipolar modeling reaches only order five with experimental code, and material-fit precision is limited by extraction from published intensity plots.
Error found: The reported benchmarks do not pair minute-scale runtimes with quantified remaining discretization errors, leaving the advertised high-precision performance insufficiently established rather than disproved. The abstract overstates the implementation's immediate applicability to general interactions: documented library support covers power-law and short-range combinations, while other kernels require adaptations and the anisotropic dipolar extension remains experimental.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 33 pages, 17 figures