Uncovering Physical Drivers of Dark Matter Halo Structures with Auxiliary-Variable-Guided Generative Models

2026-02-26

AI Summary

This work addresses the limited interpretability of deep generative models in compressing high-dimensional astrophysical data, where latent spaces often entangle physically meaningful factors. To resolve this, the authors propose a guided framework that incorporates halo mass and concentration as auxiliary variables, uniquely integrating a lightweight alignment penalty with an improved Latent Conditional Flow Matching (LCFM) approach. This enables high-fidelity generation of thermal Sunyaev–Zel’dovich (tSZ) images while achieving disentangled, physically interpretable latent representations. The resulting latent dimensions explicitly correspond to meaningful astrophysical quantities, successfully reproducing the known mass–concentration scaling relation and revealing latent outliers potentially indicative of anomalous halo formation histories. Consequently, the latent space is transformed into an effective diagnostic tool for probing cosmic structure.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Deep generative models (DGMs) compress high-dimensional data but often entangle distinct physical factors in their latent spaces. We present an auxiliary-variable-guided framework for disentangling representations of thermal Sunyaev-Zel'dovich (tSZ) maps of dark matter halos. We introduce halo mass and concentration as auxiliary variables and apply a lightweight alignment penalty to encourage latent dimensions to reflect these physical quantities. To generate sharp and realistic samples, we extend latent conditional flow matching (LCFM), a state-of-the-art generative model, to enforce disentanglement in the latent space. Our Disentangled Latent-CFM (DL-CFM) model recovers the established mass-concentration scaling relation and identifies latent space outliers that may correspond to unusual halo formation histories. By linking latent coordinates to interpretable astrophysical properties, our method transforms the latent space into a diagnostic tool for cosmological structure. This work demonstrates that auxiliary guidance preserves generative flexibility while yielding physically meaningful, disentangled embeddings, providing a generalizable pathway for uncovering independent factors in complex astronomical datasets.

Problem

• dark matter halos
• disentangled representation
• generative models
• physical drivers
• latent space

Innovation

• disentangled representation
• auxiliary-variable guidance
• latent conditional flow matching
• dark matter halos
• generative modeling