Bridging Simulators with Conditional Optimal Transport

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zeghal, Justine, Remy, Benjamin, Hezaveh, Yashar, Lanusse, Francois, Levasseur, Laurence Perreault
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911237022941184
author Zeghal, Justine
Remy, Benjamin
Hezaveh, Yashar
Lanusse, Francois
Levasseur, Laurence Perreault
author_facet Zeghal, Justine
Remy, Benjamin
Hezaveh, Yashar
Lanusse, Francois
Levasseur, Laurence Perreault
contents We propose a new field-level emulator that bridges two simulators using unpaired simulation datasets. Our method leverages a flow-based approach to learn the likelihood transport from one simulator to the other. Since multiple transport maps exist, we employ Conditional Optimal Transport Flow Matching (COT-FM) to ensure that the transformation minimally distorts the underlying structure of the data. We demonstrate the effectiveness of this approach by bridging weak lensing simulators: a Lagrangian Perturbation Theory (LPT) to a N-body Particle-Mesh (PM). We demonstrate that our emulator captures the full correction between the simulators by showing that it enables full-field inference to accurately recover the true posterior, validating its accuracy beyond traditional summary statistics.
format Preprint
id arxiv_https___arxiv_org_abs_2510_24631
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Bridging Simulators with Conditional Optimal Transport
Zeghal, Justine
Remy, Benjamin
Hezaveh, Yashar
Lanusse, Francois
Levasseur, Laurence Perreault
Cosmology and Nongalactic Astrophysics
Machine Learning
We propose a new field-level emulator that bridges two simulators using unpaired simulation datasets. Our method leverages a flow-based approach to learn the likelihood transport from one simulator to the other. Since multiple transport maps exist, we employ Conditional Optimal Transport Flow Matching (COT-FM) to ensure that the transformation minimally distorts the underlying structure of the data. We demonstrate the effectiveness of this approach by bridging weak lensing simulators: a Lagrangian Perturbation Theory (LPT) to a N-body Particle-Mesh (PM). We demonstrate that our emulator captures the full correction between the simulators by showing that it enables full-field inference to accurately recover the true posterior, validating its accuracy beyond traditional summary statistics.
title Bridging Simulators with Conditional Optimal Transport
topic Cosmology and Nongalactic Astrophysics
Machine Learning
url https://arxiv.org/abs/2510.24631