QRCTM: iPEPS ground state data for the spin-1/2 Heisenberg antiferromagnet on the square lattice (D=4–11)

Read full text Summarise with VIRAT
Authors: Qi Yang
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Reproducibility data package for infinite Projected Entangled Pair State (iPEPS) calculations of the spin-1/2 Heisenberg antiferromagnet on the square lattice, $H = \sum_{\langle ij \rangle} \mathbf{S}_i \cdot \mathbf{S}_j$ ($S=1/2$), with bond dimensions $D = 4$ through $D = 11$. Method: Automatic differentiation (AD) iPEPS optimized with L-BFGS and Wirtinger-metric preconditioner. Environments computed via Corner Transfer Matrix Renormalization Group (CTMRG). Finite-$\chi$ energies extrapolated to $\chi\to\infty$ using the finite correlation length scaling (FCLS) scheme. D-ramping strategy used to initialize larger bond dimensions from converged smaller-$D$ solutions. Key results ($E/N$ at $\chi_\mathrm{max}$, QMC reference $-0.6694421(4)$, Sandvik 1997): D chi_max xi E(chi_max) m^2 4 256 10.43 -0.6692299926 0.1080 5 392 6.44 -0.6694046273 0.1017 6 576 8.60 -0.6694299200 0.0995 7 776 10.52 -0.6694354227 0.0985 8 1024 11.39 -0.6694369616 0.0981 9 1288 12.88 -0.6694380245 0.0978 10 1600 13.74 -0.6694384832 0.0977 11 1864 16.72 -0.6694401201 0.0971 FCLS extrapolation ($D\to\infty$, $\chi\to\infty$): E_0 = -0.6694416(2), QMC: -0.669441857(7) [Sandvik, arXiv:2601.20189], tension 1.3 sigma m = 0.30735(30), QMC: 0.307447(2) [Sandvik, arXiv:2601.20189], tension 0.3 sigma Package contents: - tensors/ -- Optimized iPEPS tensors $M$ (NumPy, float64) for $D = 4$--$11$, one best-seed tensor per $D$. From $M$ alone, all observables can be recomputed via CTMRG. - analysis/ -- Chi-sweep data (energy, correlation length, magnetization vs $\chi$) and $\chi\to\infty$ extrapolations for each $D$. - config/ -- Full optimization parameters: D-ramping pipeline config, per-step summaries, per-seed configs. - logs/ -- Computational logs for the $D = 11$ optimization (with metric preconditioner) and chi sweep. - plots/ -- Plotting scripts for FCLS analysis, energy/magnetization convergence, and publication figures. - code/ -- Source code snapshot (qrpeps, commit 09d3a28) and the historical codebase that produced these results. - README.md -- Full documentation with reproduction instructions and data format specifications. Reproduction: Load any M.npy with PyTorch, run CTMRG (100 warmup steps), and evaluate the Heisenberg energy. Verified to reproduce stored energies to $\sim 10^{-7}$ precision. See README.md for step-by-step instructions. Computational resources: NVIDIA H100 GPUs on BSC MareNostrum 5.

Fields of Study & Subjects

Quantum many-body systems Advanced Condensed Matter Physics Physics of Superconductivity and Magnetism

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
22 March 2026
Language
EN
Document type
Book