mlsanityData to 14 Oct 2026

Generative Pseudo-Force Fields for Molecular Generation

arXiv 2605.19050 ↗First posted 18 May 2026cs.LG physics.chem-ph q-bio.QMcited by 1

On the map: Diffusion models → Flow matching, time series and science

Abstract (arXiv)

Generating stable molecular conformations typically forces a tradeoff between the physical realism of energy-based relaxation and the sampling efficiency of data-driven generative models. While machine learning force fields (MLFFs) can sample stable conformations by relaxing molecular geometries according to physical forces, they require costly ab-initio training data. Conversely, diffusion models (DMs) learn from equilibrium data alone but are dependent on noise schedules and time-step conditioning. In this work, we propose generative pseudo-force fields (GPFFs) to bridge these paradigms by training an MLFF on a quadratic pseudo-potential energy surface relative to reference equilibrium structures. Because no ab-initio calculations are required for the perturbed geometries, non-equilibrium training data can be generated on the fly by perturbing the equilibria with Gaussian noise. We show that GPFFs constitute a time-step-agnostic variant of variance exploding DMs: the score comes from the predicted pseudo-forces but because force magnitudes implicitly encode the noise level, no time-step conditioning is needed. Our GPFF can hence be used as a drop-in replacement in standard diffusion sampling (ancestral, Heun) but also facilitates more efficient, adaptive variants and an MLFF inspired direct denoising scheme. Our proposed sampling algorithms support arbitrary structural priors and geometric constraints. On QM9, GPFF has 100 % validity at 256 neural function evaluations (NFE) and over 50 % at just 6 NFE, outperforming diffusion baselines across all samplers. Combined with custom priors, we showcase the fast and accurate generation process of our method in a molecular editor for a drug design setting, where a molecule is generated in real time.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Flow matching, time series and science, one of 14 directions in Diffusion models. Builds on 53 of its 101 references that are in the map. First posted on arXiv 18 May 2026.

Builds on

53 of 101 references.

Closest ancestors · 13

Same direction, newest first

  1. 2026Enhanced Diffusion Sampling: Efficient Rare Event Sampling and Free Energy Calculation with Diffusion Models18 Feb
  2. 2025Consistent Sampling and Simulation: Molecular Dynamics with Energy-Based Diffusion Models20 Jun
  3. 2025Sampling 3D Molecular Conformers with Diffusion Transformers18 Jun
  4. 2025How to build a consistency model: Learning flow maps via self-distillation24 May
  5. 2025Energy Matching: Unifying Flow Matching and Energy-Based Models for Generative Modeling14 Apr
  6. 2023Equivariant flow matching26 Jun
  7. 2023Geometric Latent Diffusion Models for 3D Molecule Generation2 May
  8. 2023Two for One: Diffusion Models and Force Fields for Coarse-Grained Molecular Dynamics1 Feb
  9. 2022Structure-based Drug Design with Equivariant Diffusion Models24 Oct
  10. 2022Equivariant 3D-Conditional Diffusion Models for Molecular Linker Design11 Oct
  11. 2022MDM: Molecular Diffusion Model for 3D Molecule Generation13 Sep
  12. 2022Equivariant Diffusion for Molecule Generation in 3D31 Mar
1 more
  1. 2022GeoDiff: a Geometric Diffusion Model for Molecular Conformation Generation6 Mar

Other references · 40

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 9,122
  3. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,101
  4. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,761
  5. 2022Elucidating the Design Space of Diffusion-Based Generative Modelscited by 3,045
  6. 2023Consistency Modelscited by 1,822
  7. 2021E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentialscited by 708
  8. 2017SchNet: A continuous-filter convolutional neural network for modeling quantum interactionscited by 571
  9. 2023Consistency Trajectory Models: Learning Probability Flow ODE Trajectory of Diffusioncited by 459
  10. 2021Equivariant message passing for the prediction of tensorial properties and molecular spectracited by 441
  11. 2024One Step Diffusion via Shortcut Modelscited by 391
  12. 2021Knowledge Distillation in Iterative Generative Models for Improved Sampling Speedcited by 348
28 more
  1. 2023Common Diffusion Noise Schedules and Sample Steps are Flawedcited by 310
  2. 2022Torsional Diffusion for Molecular Conformer Generationcited by 263
  3. 2020Equivariant Flows: Exact Likelihood Generative Learning for Symmetric Densitiescited by 250
  4. 2020Machine Learning Force Fieldscited by 207
  5. 2023On the Importance of Noise Scheduling for Diffusion Modelscited by 194
  6. 2019Symmetry-adapted generation of 3d point sets for the targeted discovery of moleculescited by 172
  7. 2021E(n) Equivariant Normalizing Flowscited by 165
  8. 2022Pre-training via Denoising for Molecular Property Predictioncited by 124
  9. 2019Machine learning for molecular simulationcited by 110
  10. 2022Diffusion-based Molecule Generation with Informative Prior Bridgescited by 106
  11. 2021SpookyNet: Learning Force Fields with Electronic Degrees of Freedom and Nonlocal Effectscited by 104
  12. 2021Inverse design of 3d molecular structures with conditional generative neural networkscited by 93
  13. 2019Robust and interpretable blind image denoising via bias-free convolutional neural networkscited by 90
  14. 2023Timewarp: Transferable Acceleration of Molecular Dynamics by Learning Time-Coarsened Dynamicscited by 66
  15. 2023MolDiff: Addressing the Atom-Bond Inconsistency Problem in 3D Molecule Diffusion Generationcited by 65
  16. 2025Is Noise Conditioning Necessary for Denoising Generative Models?cited by 44
  17. 2022So3krates: Equivariant attention for interactions on arbitrary length-scales in molecular systemscited by 39
  18. 2022SchNetPack 2.0: A neural network toolbox for atomistic machine learningcited by 29
  19. 2025Equilibrium Matching: Generative Modeling with Implicit Energy-Based Modelscited by 21
  20. 2025Force-Free Molecular Dynamics Through Autoregressive Equivariant Networkscited by 15
  21. 2025FlashMD: long-stride, universal prediction of molecular dynamicscited by 12
  22. 2025How simple can you go? An off-the-shelf transformer approach to molecular dynamicscited by 10
  23. 2024Molecular relaxation by reverse diffusion with time step predictioncited by 9
  24. 2026Blind denoising diffusion models and the blessings of dimensionalitycited by 9
  25. 2026The Geometry of Noise: Why Diffusion Models Don't Need Noise Conditioningcited by 8
  26. 2026Learning Hamiltonian Flow Maps: Mean Flow Consistency for Large-Timestep Molecular Dynamicscited by 7
  27. 2025Disentangling Total-Variance and Signal-to-Noise-Ratio Improves Diffusion Modelscited by 4
  28. 2024Riemannian Denoising Model for Molecular Structure Optimization with Chemical Accuracycited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Representations from Pretrained Machine-Learning Interatomic Potentials as Coarse Coordinates for Material Generation and Evaluationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2021Learning Gradient Fields for Molecular Conformation Generation
  2. 2023Swallowing the Bitter Pill: Simplified Scalable Conformer Generation
  3. 2024Physics-informed generative model for drug-like molecule conformers
  4. 2025Guiding Diffusion Models with Reinforcement Learning for Stable Molecule Generation
  5. 2026Generating Physically Consistent Molecules with Energy-Based Models
  6. 2022DiffMD: A Geometric Diffusion Model for Molecular Dynamics Simulations
  7. 2026Elign: Equivariant Diffusion Model Alignment from Foundational Machine Learning Force Fields
  8. 2023Stochastic force inference via density estimation