PatchNet: A hierarchical approach for neural field-level inference from Quijote Simulations

2025-09-03

AI Summary

This study addresses the challenge of tightening cosmological parameter constraints from nonlinear dark matter fields for next-generation surveys (e.g., Euclid, DESI, Vera Rubin), where conventional analytic statistics—such as the power spectrum and bispectrum—fail to capture strong non-Gaussian structures. To overcome this limitation, we propose a hierarchical fusion framework based on simulation-based inference (SBI), integrating PatchNet—a patch-wise neural network architecture—for localized inference with global power spectrum and bispectrum modeling. This design circumvents memory bottlenecks associated with full-sky training while efficiently fusing multi-scale information. At a resolution of 7.8 Mpc/h, our method achieves Fisher information comparable to wavelet-based statistics, substantially outperforms traditional statistics, and approaches the theoretical information limit of the dark matter density field.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Search and Optimization: Non-convex Optimization
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

extit{What is the cosmological information content of a cubic Gigaparsec of dark matter? } Extracting cosmological information from the non-linear matter distribution has high potential to tighten parameter constraints in the era of next-generation surveys such as Euclid, DESI, and the Vera Rubin Observatory. Traditional approaches relying on summary statistics like the power spectrum and bispectrum, though analytically tractable, fail to capture the full non-Gaussian and non-linear structure of the density field. Simulation-Based Inference (SBI) provides a powerful alternative by learning directly from forward-modeled simulations. In this work, we apply SBI to the extit{Quijote} dark matter simulations and introduce a hierarchical method that integrates small-scale information from field sub-volumes or extit{patches} with large-scale statistics such as power spectrum and bispectrum. This hybrid strategy is efficient both computationally and in terms of the amount of training data required. It overcomes the memory limitations associated with full-field training. We show that our approach enhances Fisher information relative to analytical summaries and matches that of a very different approach (wavelet-based statistics), providing evidence that we are estimating the full information content of the dark matter density field at the resolution of $sim 7.8~mathrm{Mpc}/h$.

Problem

• Extracting cosmological information from non-linear dark matter distribution
• Overcoming limitations of traditional summary statistics like power spectrum
• Estimating full information content of dark matter density field

Innovation

• Hierarchical method combining patches with large-scale statistics
• Simulation-Based Inference from Quijote dark matter simulations
• Overcomes memory limits of full-field training efficiently