arXiv subjects

Halley Fritze

Publications and source records attributed to Halley Fritze.

5 recordsLinked to original sources

arXiv2026

Population-scale Ancestral Recombination Graphs with tskit 1.0

Ancestral recombination graphs (ARGs) are an increasingly important component of population and statistical genetics. The tskit library has become key infrastructure for the field, providing an expressive and general representation of ARGs together with a suite of efficient fundamental operations. In this note, we announce tskit version 1.0, describe its underlying rationale, and document its stability guarantees. These guarantees provide a foundation for durable computational artefacts and support long-term reproducibility of code and analyses.

q-bio.PE↗

arXiv2025

Multiscale 2-Mapper -- Exploratory Data Analysis Guided by the First Betti Number

The Mapper algorithm is a fundamental tool in exploratory topological data analysis for identifying connectivity and topological clustering in data. Derived from the nerve construction, Mapper graphs can contain additional information about clustering density when considering the higher-dimensional skeleta. To observe two-dimensional features, and capture one-dimensional topology, we construct 2-Mapper. A common issue in using Mapper algorithms is parameter choice. We develop tools to choose 2-Mapper parameters that reflect persistent Betti-1 information. Computationally, we study how cover choice affects 2-Mapper and analyze this through a computational Multiscale Mapper algorithm. We test our constructions on three-dimensional shape data, including the Klein bottle.

cs.CG↗

arXiv2024

Faithful Reeb Graph Reconstruction of a Tectonic Subduction Zone from Earthquake Hypocenters

An important problem in topological data analysis (TDA)$\unicode{x2014}$of both theoretical and practical interest$\unicode{x2014}$is to reconstruct the topology and geometry of an underlying (usually unknown) metric graph from possibly noisy data sampled around it. Reeb graphs have recently been successfully employed in abstract metric graph reconstruction under Gromov$\unicode{x2013}$Hausdorff noise: the sample is assumed to be metrically close to the ground truth. However, such a strong global density guarantee is often unavailable, making the existing Reeb graph-based methods unusable. A very different yet more relevant paradigm focuses on the reconstruction of metric graphs$\unicode{x2014}$embedded in the Euclidean space$\unicode{x2014}$from Euclidean samples that are only Hausdorff-close. We relax the density assumption to give provable geometric reconstruction schemes, even when the sample is metrically close only locally, but still provide provable guarantees for the successful geometric reconstruction of Euclidean graphs under the Hausdorff noise model. We apply our graph reconstruction techniques to reconstruct earthquake plate tectonic boundaries from the global earthquake catalog. The SLAB2.0 model is a comprehensive spatial summary of all known subduction zone slabs on Earth. We reconstruct parts of the SLAB2.0 model from possibly noisy earthquake hypocenter data.

cs.CG↗

arXiv2024

ICML Topological Deep Learning Challenge 2024: Beyond the Graph Domain

This paper describes the 2nd edition of the ICML Topological Deep Learning Challenge that was hosted within the ICML 2024 ELLIS Workshop on Geometry-grounded Representation Learning and Generative Modeling (GRaM). The challenge focused on the problem of representing data in different discrete topological domains in order to bridge the gap between Topological Deep Learning (TDL) and other types of structured datasets (e.g. point clouds, graphs). Specifically, participants were asked to design and implement topological liftings, i.e. mappings between different data structures and topological domains --like hypergraphs, or simplicial/cell/combinatorial complexes. The challenge received 52 submissions satisfying all the requirements. This paper introduces the main scope of the challenge, and summarizes the main results and findings.

cs.LG↗

arXiv2024

TopoBench: A Framework for Benchmarking Topological Deep Learning

This work introduces TopoBench, an open-source library designed to standardize benchmarking and accelerate research in topological deep learning (TDL). TopoBench decomposes TDL into a sequence of independent modules for data generation, loading, transforming and processing, as well as model training, optimization and evaluation. This modular organization provides flexibility for modifications and facilitates the adaptation and optimization of various TDL pipelines. A key feature of TopoBench is its support for transformations and lifting across topological domains. Mapping the topology and features of a graph to higher-order topological domains, such as simplicial and cell complexes, enables richer data representations and more fine-grained analyses. The applicability of TopoBench is demonstrated by benchmarking several TDL architectures across diverse tasks and datasets.

cs.LG↗