Simulation-Efficient Cosmological Inference with Multi-Fidelity SBI

2025/07/01 by Thiele, Leander, Bayer, Adrian E., Takeishi, Naoya · 1 citation
#Cosmology and Nongalactic Astrophysics (astro-ph.CO) #FOS: Computer and information sciences #FOS: Physical sciences #Machine Learning (cs.LG)

paper · doi:10.48550/arxiv.2507.00514

Abstract

The simulation cost for cosmological simulation-based inference can be decreased by combining simulation sets of varying fidelity. We propose an approach to such multi-fidelity inference based on feature matching and knowledge distillation. Our method results in improved posterior quality, particularly for small simulation budgets and difficult inference problems.

Citations

• Transfer learning for multifidelity simulation-based inference in cosmology
• Modeling Galaxy Surveys with Hybrid SBI
• How many simulations do we need for simulation-based inference in cosmology?
• Multifidelity Simulation-based Inference for Computationally Expensive Simulators
• Cosmological Analysis with Calibrated Neural Quantile Estimation and Approximate Simulators
• Simulation-Based Inference with Quantile Regression
• Hybrid SBI or How I Learned to Stop Worrying and Learn the Likelihood
• Neutrino mass constraint from an Implicit Likelihood Analysis of BOSS voids
• Benchmarking Simulation-Based Inference
• Optuna: A Next-generation Hyperparameter Optimization Framework
• Neural Spline Flows
• Automatic Posterior Transformation for Likelihood-Free Inference
• Sequential Neural Likelihood: Fast Likelihood-free Inference with Autoregressive Flows
• Flexible statistical inference for mechanistic models of neural dynamics
• Revisiting Classifier Two-Sample Tests
• Distilling the Knowledge in a Neural Network

Cited by

• Transfer Learning Beyond the Standard Model
• Multilevel neural simulation-based inference

Related

• Transfer learning for multifidelity simulation-based inference in cosmology
• Simulation-based stacking
• Cosmological Parameter Estimation with Sequential Linear Simulation-based Inference
• Dynamic SBI: Round-free Sequential Simulation-Based Inference with Adaptive Datasets
• Cosmic Inference: Constraining Parameters With Observations and Highly Limited Number of Simulations