Macrocanonical Generator Networks: Data-Efficient Neural Surrogates for Amortized Physics Simulation

Abstract

Many scientific simulations are computationally expensive, limiting their use in simulation-based inference, uncertainty quantification, and decision making. We introduce macrocanonical generator networks, a framework for learning fast, data-efficient neural surrogates for amortized simulation of stationary physical processes by matching multiscale statistics of a target process. Inspired by microcanonical maximum-entropy synthesis, which matches feature statistics per sample by inner-loop optimization, our approach instead trains a neural generator to satisfy the same constraints in expectation, preserving physically realistic sample-to-sample fluctuations. We prove that gradient-descent training induces preconditioned gradient-descent dynamics in sample space that inherit the symmetry-preservation properties of the microcanonical gradient descent of Bruna and Mallat, amortizing its per-sample inner-loop optimization into a single training run after which every sample is generated in one forward pass. We further show that residual generator parameterizations initialized near the identity, trained with a small learning rate, preserve entropy locally during early optimization and mitigate mode collapse. Experiments on cosmology and fluid simulations show that macrocanonical generators recover realistic samples from extremely small training sets, in some cases a single training example, while generating new realizations more than 10⁶ times faster per sample than the reference simulator and more than 10³ times faster than microcanonical gradient descent. The trained generators can be reused for downstream pretraining, including simulation-based inference, providing a data-efficient route to multi-fidelity scientific machine learning workflows.

ENFR: Evidence Networks with Flow-Matched References for Amortized Simulation-Based Model Inference
Kai Lehman, Jameson Dong, Carolina Cuesta-Lazaro +5
arXiv16 Oct 2026PreprintCC BY 4.0Full textCosmology StatisticsarXiv
TNFlow: Amortized Posterior Inference for Trans-Neptunian Object Surface Composition
Agastya Gaur, Cristina M. Dalle Ore, Alessandra Ricca
arXiv3 Sep 2026PreprintCC BY 4.0Full textInstrumentation & methods Planetary science & exoplanetsarXiv
MUSHY: Multimodal Flow-Based Amortized Bayesian Inference for Spectroscopic Data Fusion
Jeff Shen
arXiv14 Sep 2026PreprintarXiv licenceFull textInstrumentation & methodsarXiv
Misspecification in amortized X-ray spectral inference: a detection benchmark, a gain shift three schemes cannot see, and marginalizing it out
Karan Akbari
arXiv30 Aug 2026PreprintCC BY 4.0Full textHigh-energy astrophysics Instrumentation & methodsarXiv
Generative Diffusion Surrogates with Analytical Variance Schedule
Patrick Reichherzer, Gianluca Gregori, David N. Hosking +1
arXiv1 Sep 2026PreprintCC BY 4.0Full textInstrumentation & methods Plasma physicsarXiv

Cite

BibTeX RIS

BibTeX

@misc{jeffrey2026macrocanonical,
  title = {{Macrocanonical Generator Networks: Data-Efficient Neural Surrogates for Amortized Physics Simulation}},
  author = {Niall Jeffrey and Benjamin D. Wandelt},
  howpublished = {arXiv},
  year = {2026},
  month = nov,
  doi = {10.48550/arxiv.2611.05157},
  eprint = {2611.05157},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.05157},
}

RIS

TY  - UNPB
TI  - Macrocanonical Generator Networks: Data-Efficient Neural Surrogates for Amortized Physics Simulation
AU  - Niall Jeffrey
AU  - Benjamin D. Wandelt
PY  - 2026
DA  - 2026/11/03
PB  - arXiv
DO  - 10.48550/arxiv.2611.05157
UR  - https://arxiv.org/abs/2611.05157
AB  - Many scientific simulations are computationally expensive, limiting their use in simulation-based inference, uncertainty quantification, and decision making. We introduce macrocanonical generator networks, a framework for learning fast, data-efficient neural surrogates for amortized simulation of stationary physical processes by matching multiscale statistics of a target process. Inspired by microcanonical maximum-entropy synthesis, which matches feature statistics per sample by inner-loop optimization, our approach instead trains a neural generator to satisfy the same constraints in expectation, preserving physically realistic sample-to-sample fluctuations. We prove that gradient-descent training induces preconditioned gradient-descent dynamics in sample space that inherit the symmetry-preservation properties of the microcanonical gradient descent of Bruna and Mallat, amortizing its per-sample inner-loop optimization into a single training run after which every sample is generated in one forward pass. We further show that residual generator parameterizations initialized near the identity, trained with a small learning rate, preserve entropy locally during early optimization and mitigate mode collapse. Experiments on cosmology and fluid simulations show that macrocanonical generators recover realistic samples from extremely small training sets, in some cases a single training example, while generating new realizations more than 10^6 times faster per sample than the reference simulator and more than 10^3 times faster than microcanonical gradient descent. The trained generators can be reused for downstream pretraining, including simulation-based inference, providing a data-efficient route to multi-fidelity scientific machine learning workflows.
ER  -

Abstract

Related

Cite