QMBLthe quantum many-body leaderboard

The best published ground-state energies, in one table

QMBL is a record book for variational quantum many-body simulation: one row per published energy, ranked within each Hamiltonian instance, every row citing the paper that produced the number and declaring what kind of quantity it is.

The frontier

The instances the field actually competes on. Energies are per site, in the convention the papers use (units and conventions). Bold is the record under the ranking rules.

instancerecordmethodclosest challenger
J1-J2 square 8x8, J2 = 0.5 -0.498963(2) RBM+PP with momentum (K=0), spin-parity (even... Nomura & Imada (2021) -0.498666 RBM with momentum (K=0)... Nomura (2021)
J1-J2 square 10x10, J2 = 0.5 -0.4976939(2) CNN-MPS (h,D,l)=(32,20,20), Marshall sign... Fan et al. (2026) -0.4976923 T-MPS Fan et al. (2026)
J1-J2 square 16x16, J2 = 0.5 -0.4969140(5) CNN-MPS Fan et al. (2026) -0.4967860 T-MPS Fan et al. (2026)
J1-J2 square 20x20, J2 = 0.5 -0.4967987(6) CNN-MPS (h,D,l)=(32,15,20) Fan et al. (2026) -0.4967320 ViT with symmetry restoration Viteritti et al. (2026)
Hubbard square 16x16, U = 8, n = 0.875 -0.75865(3) VAFQMC stripe length 8 Sorella (2023) -0.75730 ○ ACE (16 conv layers), no symmetry... Gu et al. (2026)
Hubbard 4x16, U = 8, n = 0.875 -0.76623(1) ACE (16 conv layers) + full symmetry... Gu et al. (2026) -0.76560 NNBF, 32 determinants + free... Loehr & Clark (2025)
Triangular Heisenberg, 196 sites, open -0.5138(1) 2D Gated RNN Hibat-Allah et al. (2022) -0.5108 DMRG (Bond dimension = 2000) run script, no paper cited
Kagome Heisenberg, 192 sites -0.42987(1) VMC with Dirac spin liquid + Jastrow run script, no paper cited none
Pyrochlore Heisenberg, 256 sites -0.48310(7) mVMC with SU(2) and symmetry projections Astrakhantsev et al. (2021) none

we found no error metric in the source we read, neither an error bar nor an energy variance. a sampled energy with a variance but no error bar. Both rows stay in the table and in rank order; the marker is a question for whoever can close it, not a criticism. See error metrics.

All 322 instances →

How current this is

67 of 322 instances carry a result published in 2025 or 2026; the rest stand where the VarBench compilation left them on 2024-10-22. Every instance page says which of the two it is, because a website that says nothing about its own currency reads as more authoritative than it is.

Separately, 31 sampled variational energies across 26 instances carry no error bar, so they are listed and rank for nothing — and 5 of them sit below their instance's current record. If one of those is your paper, the error bar is the only thing missing.

What a row claims

Every row is one published claim about one Hamiltonian instance: the energy, its error bar, the method, the primary reference, and a declared bound_type saying what the number actually is — a strict variational bound, a projected or fixed-node estimate, a zero-variance extrapolation, or a numerically exact result. Only strict variational bounds compete for a record, which is what keeps an extrapolated number from beating a measured one. The rules are in RULES.md, the row format in DATA.md.

Citing QMBL

Cite the dataset by its concept DOI, which always resolves to the latest release:

10.5281/zenodo.22753734

Individual energies should cite the primary paper named on the row, not this site. The references here were bootstrapped from VarBench (Wu et al., Science 386, 296 (2024)), which is credited as a source of references rather than as a parent work.