http://arxiv.org/api/wGTeCwMvqqVfV4w6MVdzFPDa13w 2025-05-15T00:00:00-04:00 16586 510 15 http://arxiv.org/abs/2404.15225v2 2025-02-17T07:09:37Z 2024-04-23T16:54:56Z

Link prediction (LP), inferring the connectivity between nodes, is a significant research area in graph data, where a link represents essential information on relationships between nodes. Although graph neural network (GNN)-based models have achieved high performance in LP, understanding why they perform well is challenging because most comprise complex neural networks. We employ persistent homology (PH), a topological data analysis method that helps analyze the topological information of graphs, to interpret the features used for prediction. We propose a novel method that employs PH for LP (PHLP) focusing on how the presence or absence of target links influences the overall topology. The PHLP utilizes the angle hop subgraph and new node labeling called degree double radius node labeling (Degree DRNL), distinguishing the information of graphs better than DRNL. Using only a classifier, PHLP performs similarly to state-of-the-art (SOTA) models on most benchmark datasets. Incorporating the outputs calculated using PHLP into the existing GNN-based SOTA models improves performance across all benchmark datasets. To the best of our knowledge, PHLP is the first method of applying PH to LP without GNNs. The proposed approach, employing PH while not relying on neural networks, enables the identification of crucial factors for improving performance.

Junwon You Eunwoo Heo Jae-Hun Jung http://arxiv.org/abs/2502.06670v2 2025-02-16T17:22:04Z 2025-02-10T16:55:16Z

In this paper we characterize the sectional category of subgroup inclusions and the $r^{th}$-sequential topological complexity of aspherical spaces of a group G in terms of the A-genus in the sense of Clapp-Puppe and Bartsch for a suitable one-element family of G-spaces A, and we discuss some of the consequences of such characterization, including new ideas about notions of category-like invariants with respect to proper actions of groups.

Arturo Espinosa Baro Several additions and improvements, a gap in a proof filled. Preliminary version, all comments are welcomed! http://arxiv.org/abs/2304.08906v3 2025-02-16T14:07:31Z 2023-04-18T11:25:17Z

Analyzing flight trajectory data sets poses challenges due to the intricate interconnections among various factors and the high dimensionality of the data. Topological Data Analysis (TDA) is a way of analyzing big data sets focusing on the topological features this data sets have as point clouds in some metric space. Techniques as the ones that TDA provides are suitable for dealing with high dimensionality and intricate interconnections. This paper introduces TDA and its tools and methods as a way to derive meaningful insights from ATM data. Our focus is on employing TDA to extract valuable information related to airports. Specifically, by utilizing persistence landscapes (a potent TDA tool) we generate footprints for each airport. These footprints, obtained by averaging over a specific time period, are based on the deviation of trajectories and delays. We apply this method to the set of Spanish' airports in the Summer Season of 2018. Remarkably, our results align with the established Spanish airport classification and raise intriguing questions for further exploration. This analysis serves as a proof of concept, showcasing the potential application of TDA in the ATM field. While previous works have outlined the general applicability of TDA in aviation, this paper marks the first comprehensive application of TDA to a substantial volume of ATM data. Finally, we present conclusions and guidelines to address future challenges in the ATM domain.

Manuel Cuerno Luis Guijarro Rosa María Arnaldo Valdés Fernando Gómez Comendador 36 pages. Accepted for publication in PLOS ONE http://arxiv.org/abs/2502.11119v1 2025-02-16T13:30:44Z 2025-02-16T13:30:44Z

In this Note, we provide the comments on the paper [A note on the hit problem for the polynomial algebra in the case of odd primes and its application] which was published in the RACSAM [Rev. Real. Acad. Cienc. Exactas. Fis. Nat. Ser. A-Mat. 118, 22 (2024)]. We will show that some of the results in this paper are not new and others are false.

Nguyen Sum 4 page http://arxiv.org/abs/2502.10964v1 2025-02-16T02:52:33Z 2025-02-16T02:52:33Z

We define the equivariant degree and local degree of a proper $G$-equivariant map between smooth $G$-manifolds when $G$ is a compact Lie group and prove a local to global result. We show the local degree can be used to compute the equivariant Euler characteristic of a smooth, compact $G$-manifold and the Euler number of a relatively oriented $G$-equivariant vector bundle when $G$ is finite. As an application, we give an equivariantly enriched count of rational plane cubics through a $G$-invariant set of 8 general points in $\mathbb{C}\mathbb{P}^2$, valued in the representation ring and Burnside ring of a finite group. When $\mathbb{Z}/2$ acts by pointwise complex conjugation this recovers a signed count of real rational cubics.

Candace Bethea Kirsten Wickelgren http://arxiv.org/abs/2406.04259v2 2025-02-15T20:51:42Z 2024-06-06T17:04:50Z

We consider the problem of homotopy-type reconstruction of compact shapes $X\subset\mathbb{R}^N$ that are $\mathrm{CAT}(\kappa)$ in the intrinsic length metric. The reconstructed spaces are in the form of Vietoris--Rips complexes computed from a compact sample $S$, Hausdorff--close to the unknown shape $X$. Instead of the Euclidean metric on the sample, our reconstruction technique leverages a path-based metric to compute these complexes. As naturally emerging in the framework of reconstruction, we also study the Gromov--Hausdorff topological stability and finiteness problem for general compact $\mathrm{CAT}(\kappa)$ spaces. Our techniques provide novel sampling conditions alternative to the existing and commonly used techniques using weak feature size and $\mu$--reach. In particular, we introduce a new parameter, called the {\em restricted distortion}, which is a generalization of the well-known global distortion of embedding. We show examples of Euclidean subspaces, for which the known parameters such as the reach, $\mu$--reach and weak features size vanish, whereas the restricted distortion is finite, making our reconstruction results applicable for such spaces.

Rafal Komendarczyk Sushovan Majhi Will Tran 27 pages, 1 figure, a few minor adjustments, Appendix shortened http://arxiv.org/abs/2502.10690v1 2025-02-15T06:36:58Z 2025-02-15T06:36:58Z

This paper presents a comparison between two versions of Bott Periodicity Theorems: one in topological K-theory and the other in stable homotopy groups of classical groups. It begins with an introduction to K-theory, discussing vector bundles and their role in understanding the algebraic and topological aspects of these spaces. Then the two versions of Bott periodicity, as well as the topological notions necessary to understand them, are further explored. The aim is to illustrate the connections and distinctions between these two theorems, deepening our understanding of their underlying mathematical structures such as topological K-theory and fiber bundles.

Ivan Z. Feng 18 pages, 1 figure http://arxiv.org/abs/2409.18871v2 2025-02-14T14:56:02Z 2024-09-27T16:06:28Z

For any geodesic metric space $X$, we give a complete cohomological characterisation of the hyperbolicity of $X$ in terms of vanishing of its second $\ell^{\infty}$-cohomology. We extend this result to the relative setting of $X$ with a collection of uniformly hyperbolic subgraphs. As an application, we give a cohomological characterisation of acylindrical hyperbolicity.

Francesco Milizia Nansen Petrosyan Alessandro Sisto Vladimir Vankov 27 pages, 4 figures. The introduction is expanded http://arxiv.org/abs/2502.10076v1 2025-02-14T10:55:15Z 2025-02-14T10:55:15Z

Temporal graphs effectively model dynamic systems by representing interactions as timestamped edges. However, analytical tools for temporal graphs are limited compared to static graphs. We propose a novel method for analyzing temporal graphs using Persistent Homology. Our approach leverages $\delta$-temporal motifs (recurrent subgraphs) to capture temporal dynamics %without aggregation . By evolving these motifs, we define the \textit{average filtration} and compute PH on the associated clique complex. This method captures both local and global temporal structures and is stable with respect to reference models. We demonstrate the applicability of our approach to the temporal graph classification task. Experiments verify the effectiveness of our approach, achieving over 92\% accuracy, with some cases reaching 100\%. Unlike existing methods that require node classes, our approach is node class free, offering flexibility for a wide range of temporal graph analysis.

Siddharth Pritam Rohit Roy Madhav Cherupilil Sajeev http://arxiv.org/abs/2502.01360v2 2025-02-14T10:28:36Z 2025-02-03T13:52:17Z

Previous research has proven that the set of maps implemented by neural networks with a ReLU activation function is identical to the set of piecewise linear continuous maps. Furthermore, such networks induce a hyperplane arrangement splitting the input domain into convex polyhedra $G_J$ over which the network $\Phi$ operates in an affine manner. In this work, we leverage these properties to define the equivalence class of inputs $\sim_\Phi$, which can be split into two sets related to the local rank of $\Phi_J$ and the intersections $\cap \text{Im}\Phi_{J_i}$. We refer to the latter as the overlap decomposition $O_\Phi$ and prove that if the intersections between each polyhedron and the input manifold are convex, the homology groups of neural representations are isomorphic to relative homology groups $H_k(\Phi(M)) \simeq H_k(M,O_\Phi)$. This lets us compute Betti numbers without the choice of an external metric. We develop methods to numerically compute the overlap decomposition through linear programming and a union-find algorithm. Using this framework, we perform several experiments on toy datasets showing that, compared to standard persistent homology, our relative homology-based computation of Betti numbers tracks purely topological rather than geometric features. Finally, we study the evolution of the overlap decomposition during training on various classification problems while varying network width and depth and discuss some shortcomings of our method.

Kosio Beshkov http://arxiv.org/abs/2404.17639v3 2025-02-13T20:51:25Z 2024-04-26T18:00:32Z

We determine generalized symmetries for 4D theories engineered via type II strings on non-supersymmetric orbifold backgrounds $\mathbb{R}^{3,1} \times \mathbb{R}^6 / \Gamma$. Probe branes detect generalized symmetries via the adjacency matrix for fermionic degrees of freedom in an associated quiver gauge theory. In situations where the tachyons are sequestered away from the boundary $S^5 / \Gamma$, this exactly matches the result extracted from singular homology. In situations with an unsequestered tachyon which stretches out to the boundary, the presence of tachyonic pulses partitions up the space into several distinct sectors, and the net contribution again matches with the answer expected via quiver methods. For IIA backgrounds, the presence of a localized closed string tachyon leads to transitions in the spectrum of states, generalized symmetries, higher-group symmetries, as well as the level matrix of the associated symmetry topological field theory (SymTFT). For IIB backgrounds with a stack of spacetime filling probe D3-branes, the onset of a radiatively generated potential leads to similar considerations involving scale dependent transitions in the symmetries of the theory, including structures such as duality defects / interfaces.

Noah Braeger Vivek Chakrabhavi Jonathan J. Heckman Max Hübner v3: 48 pages + appendices, 21 figures, typos corrected, clarifications added http://arxiv.org/abs/2206.11802v3 2025-02-13T14:38:06Z 2022-06-23T16:16:36Z

In this paper, we classify the parameter ideals in $H^*(BA_4;\mathbb{F}_2)$ and in the Dickson algebra $H^*(BSO(3);\mathbb{F}_2)$ that are closed under Steenrod operations. Consequently, we obtain restrictions on the dimensions $n,m$ for which $A_4$ (and $SO(3)$) can act freely on $S^n\times S^m$.

Henrik Rüping Marc Stephan Ergun Yalcin 37 pages, 1 figure; v2: removed sections 8 and 9 to shorten the paper, added Theorem 1.4; v3: minor changes, final version J. Pure Appl. Algebra 229 (2025) 107899 http://arxiv.org/abs/2410.07570v2 2025-02-13T05:29:36Z 2024-10-10T03:11:46Z

We show the existence of $S^1\times C_p$-maps between certain representation spheres. As an application, we show that, in the family of abelian compact Lie groups, a group $G$ has the weak Borsuk-Ulam property (in the sense of Bartsch) if and only if $G$ is either a finite abelian $p$-group or a $k$-torus.

Ikumitsu Nagasaki http://arxiv.org/abs/2502.08847v1 2025-02-12T23:44:48Z 2025-02-12T23:44:48Z

A representation $\rho$ of a compact group $\mathbb{G}$ selects eigenvalues if there is a continuous circle-valued map on $\mathbb{G}$ assigning an eigenvalue of $\rho(g)$ to every $g\in \mathbb{G}$. For every compact connected $\mathbb{G}$, we characterize the irreducible $\mathbb{G}$-representations which select eigenvalues as precisely those annihilating the intersection $Z_0(\mathbb{G})\cap \mathbb{G}'$ of the connected center of $\mathbb{G}$ with its derived subgroup. The result applies more generally to finite-spectrum representations isotypic on $Z_0(\mathbb{G})$, and recovers as applications (noted in prior work) the existence of a continuous eigenvalue selector for the natural representation of $\mathrm{SU}(n)$ and the non-existence of such a selector for $\mathrm{U}(n)$.

Alexandru Chirvasitu 16 pages + references http://arxiv.org/abs/2408.05486v2 2025-02-12T17:29:28Z 2024-08-10T08:27:58Z

Topological deep learning (TDL) is a rapidly growing field that seeks to leverage topological structure in data and facilitate learning from data supported on topological objects, ranging from molecules to 3D shapes. Most TDL architectures can be unified under the framework of higher-order message-passing (HOMP), which generalizes graph message-passing to higher-order domains. In the first part of the paper, we explore HOMP's expressive power from a topological perspective, demonstrating the framework's inability to capture fundamental topological and metric invariants such as diameter, orientability, planarity, and homology. In addition, we demonstrate HOMP's limitations in fully leveraging lifting and pooling methods on graphs. To the best of our knowledge, this is the first work to study the expressivity of TDL from a \emph{topological} perspective. In the second part of the paper, we develop two new classes of architectures -- multi-cellular networks (MCN) and scalable MCN (SMCN) -- which draw inspiration from expressive GNNs. MCN can reach full expressivity, but scaling it to large data objects can be computationally expansive. Designed as a more scalable alternative, SMCN still mitigates many of HOMP's expressivity limitations. Finally, we create new benchmarks for evaluating models based on their ability to learn topological properties of complexes. We then evaluate SMCN on these benchmarks and on real-world graph datasets, demonstrating improvements over both HOMP baselines and expressive graph methods, highlighting the value of expressively leveraging topological information. Code and data are available at https://github.com/yoavgelberg/SMCN.

Yam Eitan Yoav Gelberg Guy Bar-Shalom Fabrizio Frasca Michael Bronstein Haggai Maron