Pith. sign in

cs.CG

Computational Geometry

Roughly includes material in ACM Subject Classes I.3.5 and F.2.2.

Papers reviewed in the last 7 days lead, then the papers readers actually read. Ranking is not a quality score.

sort pith recommended most recent

Structured model yields uniform power for persistence diagram testing

By translating latent template separation into population mean signal, the framework enables two-sample tests with matching minimax rates.

· “Statistical Inference for Persistence Diagrams via Landmark Embeddings: Minimax Theory and Finite Approximation”

open re-runnable review →
Figure from the paper

A fractal curve makes 2-Wasserstein comparisons up to 2000x faster

The SK-Wasserstein distance sorts persistence-diagram codes instead of solving planar matchings, staying faithful to W2 in tests.

· “Sierpi\'nski--Knopp Wasserstein Distance for Persistence Diagrams and Applications to 2-Wasserstein Approximation”

open re-runnable review →
Figure from the paper

Online hyperplane search value is computable in any dimension

N-COMP theorem proves the minimax competitive ratio C*_D can be approximated to any precision, despite an infinite-dimensional Bellman…

· “Bellman--Shoreline Search in Arbitrary Dimension: Exponential Vector Oscillators, Active Memory, Precession, and Effective Computability”

open re-runnable review →
Figure from the paper

Lunar EMST expected cost scales as constant times √n

New generalization of minimum spanning trees to colored points proves asymptotic growth, extending a classic 1959 result.

· “Lunar Generalizations of the Euclidean Minimum Spanning Tree in the Plane and their Expected Costs”

open re-runnable review →
Figure from the paper

Interaction of point clouds reduces to one forced integer curve

The overlap's Euler characteristic is the unique symmetric, stable interaction profile, computable in near-optimal time.

· “The Intersection Euler Characteristic Profile: Euler Calculus and Stability for Topological Interaction of Ball Unions”

open re-runnable review →

Pseudoinverse metric exceeds 1-WL and improves neuron typing

A closed-form matrix inverse gives a training-free graph signature that beats standard message passing on 3D neuron data.

· “Tropical Algebraic Geometry for Neuronal Representations: An Arakelov-Green Measure Based Descriptor for Graph Learning”

open re-runnable review →
Figure from the paper

browse all of cs.CG → full archive · search · sub-categories