Pith. sign in

REVIEW 5 cited by

An ab initio foundation model of wavefunctions that accurately describes chemical bond breaking

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2506.19960 v1 pith:WVG2FALX submitted 2025-06-24 physics.chem-ph cs.AIstat.ML

classification physics.chem-phcs.AIstat.ML
keywords bondmoleculesmultireferentialquantumbreakingchemicalchemistrycommonalities
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Reliable description of bond breaking remains a major challenge for quantum chemistry due to the multireferential character of the electronic structure in dissociating species. Multireferential methods in particular suffer from large computational cost, which under the normal paradigm has to be paid anew for each system at a full price, ignoring commonalities in electronic structure across molecules. Quantum Monte Carlo with deep neural networks (deep QMC) uniquely offers to exploit such commonalities by pretraining transferable wavefunction models, but all such attempts were so far limited in scope. Here, we bring this new paradigm to fruition with Orbformer, a novel transferable wavefunction model pretrained on 22,000 equilibrium and dissociating structures that can be fine-tuned on unseen molecules reaching an accuracy-cost ratio rivalling classical multireferential methods. On established benchmarks as well as more challenging bond dissociations and Diels-Alder reactions, Orbformer is the only method that consistently converges to chemical accuracy (1 kcal/mol). This work turns the idea of amortizing the cost of solving the Schr\"odinger equation over many molecules into a practical approach in quantum chemistry.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Machine learning for sample-based quantum diagonalization: generative configuration recovery and the classical-simulability frontier

    quant-ph 2026-08 conditional novelty 6.0 of 10

    A critical review plus small exact-FCI experiments concludes that sample-based quantum diagonalization has not beaten classical selected CI and maps where, if anywhere, a quantum or generative advantage could survive.

  2. Rhombohedral Graphene: A Tale of Many Crystals

    cond-mat.str-el 2026-07 conditional novelty 6.0 of 10

    A Mexican-hat band dispersion in rhombohedral graphene is predicted to stabilize four electron-crystal phases—Wigner, nodal Wigner, self-doped 'phantom' crystal, and hole-lattice 'anticrystal'—mapped over density and ...

  3. Large Electron Model: A Universal Ground State Predictor

    cond-mat.str-el 2026-03 conditional novelty 6.0 of 10

    A parameter-conditioned Fermi Sets neural network, trained by variational Monte Carlo on a small grid of (λ,N), predicts ground-state wavefunctions of 2D quantum dots for unseen interaction strengths and particle numb...

  4. Attention-Based Foundation Model for Quantum States

    cond-mat.str-el 2025-12 conditional novelty 6.0 of 10

    A single attention-based transformer, conditioned on the parameters (V/t,N), learns and predicts ground-state wavefunctions of the t-V model across a 2D parameter plane from a few exact-diagonalization training points.

  5. Functional Neural Wavefunction Optimization

    cond-mat.str-el 2025-07 conditional novelty 5.0 of 10

    The authors show that stochastic reconfiguration is L2 gradient descent on the unit sphere, and they derive projected inverse iteration, a shifted inverse iteration method for neural-network wavefunctions, with faster...

Pith tools