Cosmology with Topological Deep Learning

TL;DR

This paper introduces topological neural networks (TNNs) that leverage higher-order structures and symmetry invariance to improve modeling of cosmic structures, outperforming traditional graph neural networks in key cosmological parameter predictions.

Contribution

The paper presents a novel TNN framework that incorporates higher-order topological information and $E(3)$-invariance, enhancing the ability to analyze large-scale cosmic structures beyond traditional GNNs.

Findings

ClusterTNNs improve $ ext{Ω}_m$ and $ ext{σ}_8$ predictions by up to 22% and 34%.

FullTNN achieves up to 60% improvement in $ ext{σ}_8$.

Results on CAMELS are comparable to existing GNN benchmarks.

Abstract

The standard cosmological model with cold dark matter posits a hierarchical formation of structures. We introduce topological neural networks (TNNs), implemented as message-passing neural networks on higher-order structures, to effectively capture the topological information inherent in these hierarchies that traditional graph neural networks (GNNs) fail to account for. Our approach not only considers the vertices and edges that comprise a graph but also extends to higher-order cells such as tetrahedra, clusters, and hyperedges. This enables message-passing between these heterogeneous structures within a combinatorial complex. Furthermore, our TNNs are designed to conserve the $E(3)$-invariance, which refers to the symmetry arising from invariance against translations, reflections, and rotations. When applied to the Quijote suite, our TNNs achieve a significant reduction in the mean squared error. Compared to our GNNs, which lack higher-order message-passing, ClusterTNNs show improvements of up to 22% in $\Omega_{\rm m}$ and 34% in $\sigma_8$ jointly, while the best FullTNN achieves an improvement of up to 60% in $\sigma_8$. In the context of the CAMELS suite, our models yield results comparable to the current GNN benchmark, albeit with a slight decrease in performance. We emphasize that our topology and symmetry-aware neural networks provide enhanced expressive power in modeling the large-scale structures of our universe.