QMBLthe quantum many-body leaderboard

All instances / Heisenberg

Heisenberg kagome 6x6 (108 sites)

Heisenberg/kagome-6x6_108_P

No record

every variational row is flagged (rules §6). Every row is still listed below.

The Hamiltonian

H = ∑⟨ij⟩ σi·σj

Nearest-neighbour Heisenberg antiferromagnet, J = 1, written with Pauli matrices.

lattice
kagome-6x6
sites
108
boundary
periodic
energies
2

Stored as a total energy in the convention of VarBench; quoted here as E/N (S.S), which is what the papers report (how the conversion works).

Every published energy

Grouped by what the number is. Ranking happens only inside the first group.

Strict variational upper bounds (1)

The record is the lowest eligible energy in this group, and only this group.

E/N (S.S)σVar(E)V-score methodsourceyear
-0.4461111 n/a n/a n/a GCNN, spinon pair-density-wave state (chi0, P_Z2 = -1, q = (pi, pi/sqrt3))
flagged: sampling-nonergodic peer reviewed
Đurić et al. (2024) 2024

Extrapolations (1)

Zero-variance, bond-dimension or Trotter-error extrapolations. Not a bound: no ansatz ever reached the number, so it cannot hold the record.

E/N (S.S)σVar(E)V-score methodsourceyear
-0.4383(3) 3.0e-4 n/a n/a DMRG, truncation-error extrapolated (Torus 6)
peer reviewed
Depenbrock et al. (2012) 2012

Flagged rows

A flag withholds the record and nothing else: the row stays listed, in rank order. Rules §10 is how a flag is lifted or upheld.

sampling-nonergodic — GCNN, spinon pair-density-wave state (chi0, P_Z2 = -1, q =..., -0.4461111

Finding. Published in Phys. Rev. X 15, 011047 (2025) as a ground state 1.78% below the DMRG benchmark. A Comment (arXiv:2605.28861) shows the single-spin-flip update used at this size does not conserve total magnetisation, so the Markov chains freeze and the low energy is a sampling artifact. Under ergodic sampling the same ansatz converges ~3.5% ABOVE DMRG.

Diagnosis. The reported energy is an artifact of non-ergodic Monte Carlo sampling, not a property of the ansatz. The kagome Heisenberg Hamiltonian is SU(2) symmetric and its ground state lies in a fixed total-magnetisation sector, so a single-spin-flip update - which changes S^z_tot - is incompatible with the symmetry. As the network concentrates on the physical S^z_tot = 0 sector the acceptance rate collapses to exactly zero beyond 5000 iterations and the chains freeze, so the reported average is taken over a non-representative set of configurations. With the magnetisation-preserving exchange update the same architecture and optimizer converge stably to E ~ -45.6 against a DMRG value of E ~ -47.3; re-evaluating the spin-flip-optimised PARAMETERS under ergodic sampling raises the energy further, to E ~ -42.6. The published -48.18 lies below all of them.

Ruled out. NOT a variational-principle violation, and that is the point: DMRG at finite bond dimension is itself an upper bound, so an energy below it is not on its own evidence of error - it would ordinarily just be a better state. Nothing about the number alone identifies it as wrong. The refutation had to come from the sampler.

Evidence. arXiv:2605.28861 (Kamal, Kufel, Vu, Laumann & Yao), Fig. 1(a) acceptance-rate collapse and Fig. 1(b) energy convergence; Ðurić et al., Phys. Rev. X 15, 011047 (2025), arXiv:2401.02866, Sec. IV.

How these numbers were read

Rows added after the VarBench import record where the number came from.

-0.4461111 ○ — GCNN, spinon pair-density-wave state (chi0, P_Z2 = -1, q =...

Checked 2026-09-14 · arXiv PDF text extracted locally with pypdf, no LLM transcription

Reported as. -0.4461111111111111 per site in S.S units

Sec. IV: "The obtained ground state energy E0 = -48.18(0) is ~1.78% lower than the ground state energy obtained with DMRG calculations (E_DMRG = -47.33964)". Total in S.S units; E/N = -0.4461111. The stated error (0) is not a usable error bar, so no sigma is recorded. DISPUTED - see the defect flag.

-0.4383(3) — DMRG, truncation-error extrapolated (Torus 6)

Checked 2026-09-14 · arXiv PDF text extracted locally with pypdf, no LLM transcription

Reported as. -0.4383 per site in S.S units

Table I row "Torus 6 -0.4383(3)", the 108-site torus ("we also consider tori of up to 108 sites"). E/N x 108 = -47.336, consistent with the -47.33964 that Ðurić et al. quote as the DMRG benchmark. Extrapolated in the truncation error, so not a strict bound (RULES.md 4).

Coverage

Newest published number here: 2024. 2 rows added from the 2026-09 literature sweep, on top of the VarBench snapshot of 2024-10-22.

This instance as data

JSON · source file on GitHub · open an issue about this instance