arXiv subjects

Alexander Baumgartner

Publications and source records attributed to Alexander Baumgartner.

7 recordsLinked to original sources

arXiv2026

Orbital Counting in Conjugacy Classes

In this article we consider a restricted orbital counting problem for the action of certain discrete groups on suitable spaces. In particular, we present asymptotics for counting those points in an orbit restricted to a single conjugacy class. A classical example would be cocompact actions of a discrete group acting isometrically on a simply connected manifold with pinched negative curvature. More generally, we obtain results for convex cocompact actions on $CAT(-1)$ spaces.

math.DS↗

arXiv2026

The Story is Not the Science: Execution-Grounded Evaluation of Mechanistic Interpretability Research

Reproducibility crises across sciences highlight the limitations of the paper-centric review system in assessing the rigor and reproducibility of research. AI agents that autonomously design and generate large volumes of research outputs exacerbate these challenges. In this work, we address the growing challenges of scalability and rigor by flipping the dynamic and developing AI agents as research evaluators. We propose the first execution-grounded evaluation framework that verifies research beyond narrative review by examining code and data alongside the paper. We use mechanistic interpretability research as a testbed, build standardized research output, and develop MechEvalAgent, an automated evaluation framework that assesses the coherence of the experimental process, the reproducibility of results, and the generalizability of findings. We show that our framework achieves above 80% agreement with human judges, identifies substantial methodological problems, and surfaces 51 additional issues that human reviewers miss. Our work demonstrates the potential of AI agents to transform research evaluation and pave the way for rigorous scientific practices.

cs.CY↗

arXiv2025

On the Ruelle-Mayer Transfer Operators for H\"older Continuous Functions

We consider a family of operators connected with the geodesic flow on the modular surface. We show certain spectral information is retained after expanding their domain to the space of $\alpha$-H\"older continuous functions on the unit interval. For example, the point spectra associated with the Maass cusp forms and non-trivial zeroes of the Riemann zeta function to the right of the critical line remain unchanged when the H\"older constant is $(1/2+\varepsilon)$ and $3/4$ respectively. We briefly consider a three-term functional equation introduced by Lewis in the H\"older setting and provide a partial classification of solutions in this setting.

math.NT↗

arXiv2025

Generalization of Variadic Structures with Binders: A Tool for Structural Code Comparison

This paper introduces a novel anti-unification algorithm for the generalization of variadic structures with binders, designed as a flexible tool for structural code comparison. By combining nominal techniques for handling variable binding with support for variadic expressions (common in abstract syntax trees and programming languages), the approach addresses key challenges such as overemphasis on bound variable names and difficulty handling insertions or deletions in code fragments. The algorithm distinguishes between atoms and two kinds of variables (term and hedge variables) to compute best generalizations that maximally preserve structural similarities while abstracting systematic differences. It also provides detailed information to reconstruct original expressions and quantify structural differences. This information can be useful in tasks like code clone detection, refactoring, and program analysis. By introducing a parametrizable rigidity function, the technique offers fine-grained control over similarity criteria and reduces nondeterminism, enabling flexible adaptation to practical scenarios where trivial similarities should be discounted. Although demonstrated primarily in the context of code similarity detection, this framework is broadly applicable wherever precise comparison of variadic and binder-rich representations is required.

cs.LO↗

arXiv2025

Nominal anti-unification

We study nominal anti-unification, which is concerned with computing least general generalizations for given terms-in-context. In general, the problem does not have a least general solution, but if the set of atoms permitted in generalizations is finite, then there exists a least general generalization which is unique modulo variable renaming and $\alpha$-equivalence. We present an algorithm that computes it. The algorithm relies on a subalgorithm that constructively decides equivariance between two terms-in-context. We prove soundness and completeness properties of both algorithms and analyze their complexity. Nominal anti-unification can be applied to problems were generalization of first-order terms is needed (inductive learning, clone detection, etc.), but bindings are involved.

cs.LO↗

arXiv2025

Generalization Problems with Atom-Variables in Languages with Binders and Equational Theories

Generalization problems in languages with binders involve computing the most common structure between expressions while respecting bound variable renaming and freshness constraints. These problems often lack a least general solution. However, leveraging nominal techniques, we previously demonstrated that a semantic approach with atom-variables enables the elimination of redundant solutions and allows for computing unique least general generalizations (LGGs). In this work, we extend this approach to handle associative (A), commutative (C), and associative-commutative (AC) equational theories. We present a sound and weak complete algorithm for solving equational generalization problems, which generates finite weak minimal complete sets of LGGs for each theory. A key challenge arises from solving equivariance problems while taking into account these equational theories, as identifying redundant generalizations requires recognizing when one expression (with binders) is a renaming of another while possibly considering permutations of sub-expressions. This unexpected interaction between renaming and equational reasoning made this particularly difficult, necessitating semantic tests within the equivariance algorithm. Given that these equational theories naturally induce exponentially large LGG sets due to subexpression permutations, future work could explore restricted theory fragments where the generalization problem remains unitary. In these fragments, LGGs can be computed efficiently in polynomial time, offering practical benefits for symbolic computation and automated reasoning tasks.

cs.LO↗

arXiv2024

Complex continued fractions, Kleinian and extremal theory for cusp excursions

For the each of the five Euclidean rings of complex quadratic integers, we consider a complex continued fraction algorithm with digits in the ring. We show for each algorithm that the maximal digit obeys a Fr\'echet distribution. We use this to find a limiting distribution for cusp excursions on Bianchi orbifolds associated with the aforementioned rings of quadratic integers.

math.DS↗