Research map / Probabilistic and scientific modeling / Variational and generative sampling
Learning Boltzmann Generators via Constrained Mass Transport
Christopher von Klitzing, Denis Blessing, Henrik Schopmans, Pascal Friederich, Gerhard Neumann
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use CMT instead of TA-BG | normalizing-flow Boltzmann generator · ELIL tetrapeptide | ESS (%): 13.75 -> 26.06 | Improved |
| use CMT instead of TA-BG | Boltzmann generator · Alanine hexapeptide | ESS (%): 18.22 -> 29.63 | Improved |
| use CMT instead of FAB | Boltzmann generator · Alanine dipeptide | ESS (%): 94.80 -> 97.69 | Improved |
| use CMT instead of reverse KL training | Boltzmann generator · ELIL tetrapeptide | Ram TV w. RW: 1.67 -> 1.68 | Improved |
| add trust-region constraint to no-constraint forward KL (ablation) | CMT · Alanine tetrapeptide | EUBO: -333.61 -> -333.99 | Improved |
| add entropy constraint to no-constraint forward KL (ablation) | CMT · Alanine tetrapeptide | EUBO: -333.61 -> -333.96 | Improved |
| add entropy constraint to trust-region-only CMT (ablation) | CMT · Alanine tetrapeptide | EUBO: -333.99 -> -334.00 | Improved |
| add trust-region constraint to entropy-only CMT (ablation) | CMT · Alanine tetrapeptide | EUBO: -333.96 -> -334.00 | Improved |
| add trust-region constraint to no-constraint forward KL (ablation) | CMT · Alanine tetrapeptide | Ram KL: 7.36 -> 2.16 | Improved |
| add trust-region constraint to no-constraint forward KL (ablation) | CMT · Alanine dipeptide | Ram KL: 1.47 -> 1.45 | No clear effect |
| add both constraints to no-constraint forward KL (ablation) | CMT · Alanine dipeptide | Ram KL w. RW: 1.46 -> 1.35 | Improved |
Closest papers by meaning
- Temperature-Annealed Boltzmann Generators (ICML 2025 · similarity 0.91)
- Scalable Equilibrium Sampling with Sequential Boltzmann Generators (ICML 2025 · similarity 0.89)
- BoltzNCE: Learning likelihoods for Boltzmann Generation with Stochastic Interpolants and Noise Contrastive Estimation (NeurIPS 2025 · similarity 0.88)
- Fast training and sampling of Restricted Boltzmann Machines (ICLR 2025 · similarity 0.88)
- Transferable Boltzmann Generators (NeurIPS 2024 · similarity 0.88)
- Transport meets Variational Inference: Controlled Monte Carlo Diffusions (ICLR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
