Peer Revenue Play credits. Real papers.

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Mathematics

math Watch Mathematics (sign in to use)

Thursday 9 April

Geometric Topology

Kirby diagrams for an infinite family of exotic $\mathbb{R}^4$'s

Eli, Hom, and Lidman showed that the manifolds produced by attaching the simplest positive Casson handle $CH^+$ to a slice disc complement of the ribbon knot $T_{2,n}\#T_{2,-n}$ for $n\ge3$ and odd, and removing the boundary, form a countably infinite family of exotic $\mathbb{R}^4$'s. They provided a Kirby diagram for the case $n=3$. In this short note, we extend this for $n\ge3$ and odd, and provide Kirby diagrams for two such families of exotic $\mathbb{R}^4$'s, which are then shown to be equivalent. We then generalise these diagrams to a family of exotic $\mathbb{R}^4$'s built using ribbon disc complements of the pretzel knots $P(n,-n,2k)$.
Read the full abstract

Analysis of PDEs

Relaxation dynamics of the continuum Kuramoto model with non-integrable kernels

We study the asymptotic behavior of the continuum Kuramoto model with a fractional Laplacian-type kernel. For this, we construct global weak solutions via a two-parameter regularization procedure using a kernel truncation with fractional dissipation. Using a priori uniform estimates derived in fractional Sobolev spaces, we employ compactness arguments to construct global weak solutions to the singular continuum Kuramoto model. Furthermore, we also establish an exponential relaxation toward the initial phase average in $L^2$-norm under suitable assumptions on initial data and system parameters. These findings provide a rigorous characterization of the existence of solutions and the emergent dynamics of Kuramoto ensembles under physically important strongly singular interactions, including power-law singular kernels and Coulomb-type kernels.
Read the full abstract

All authors Li Chen, Seung-Yeal Ha, Xinyu Wang, Valeriia Zhidkova

Numerical Analysis · Cross-listed from stat.ML

On the Unique Recovery of Transport Maps and Vector Fields from Finite Measure-Valued Data

We establish guarantees for the unique recovery of vector fields and transport maps from finite measure-valued data, yielding new insights into generative models, data-driven dynamical systems, and PDE inverse problems. In particular, we provide general conditions under which a diffeomorphism can be uniquely identified from its pushforward action on finitely many densities, i.e., when the data $\{(\rho_j,f_\#\rho_j)\}_{j=1}^m$ uniquely determines $f$. As a corollary, we introduce a new metric which compares diffeomorphisms by measuring the discrepancy between finitely many pushforward densities in the space of probability measures. We also prove analogous results in an infinitesimal setting, where derivatives of the densities along a smooth vector field are observed, i.e., when $\{(\rho_j,\text{div} (\rho_j v))\}_{j=1}^m$ uniquely determines $v$. Our analysis makes use of the Whitney and Takens embedding theorems, which provide estimates on the required number of densities $m$, depending only on the intrinsic dimension of the problem. We additionally interpret our results through the lens of Perron–Frobenius and Koopman operators and demonstrate how our techniques lead to new guarantees for the well-posedness of certain PDE inverse problems related to continuity, advection, Fokker–Planck, and advection-diffusion-reaction equations. Finally, we present illustrative numerical experiments demonstrating the unique identification of transport maps from finitely many pushforward densities, and of vector fields from finitely many weighted divergence observations.
Read the full abstract

Optimization and Control

Parameter-Free Non-Ergodic Extragradient Algorithms for Solving Monotone Variational Inequalities

Monotone variational inequalities (VIs) provide a unifying framework for convex minimization, equilibrium computation, and convex-concave saddle-point problems. Extragradient-type methods are among the most effective first-order algorithms for such problems, but their performance hinges critically on stepsize selection. While most existing theory focuses on ergodic averages of the iterates, practical performance is often driven by the significantly stronger behavior of the last iterate. Moreover, available last-iterate guarantees typically rely on fixed stepsizes chosen using problem-specific global smoothness information, which is often difficult to estimate accurately and may not even be applicable. In this paper, we develop parameter-free extragradient methods with non-asymptotic last-iterate guarantees for constrained monotone VIs. For globally Lipschitz operators, our algorithm achieves an $o(1/\sqrt{T})$ last-iterate rate. We then extend the framework to locally Lipschitz operators via backtracking line search and obtain the same rate while preserving parameter-freeness, thereby making parameter-free last-iterate methods applicable to important problem classes for which global smoothness is unrealistic. Our numerical experiments on bilinear matrix games, LASSO, minimax group fairness, and state-of-the-art maximum entropy sampling relaxations demonstrate wide applicability of our results as well as strong last-iterate performance and significant improvements over existing methods.
Read the full abstract

Analysis of PDEs

Supercritical Schrödinger equations involving integro-differential operators and vanishing potentials

This paper is devoted to the study of the existence of positive and bounded solutions for a Schrödinger type equation defined on the entire Euclidean space, involving a general integro-differential operator. We consider the case where the potential is nonnegative and vanishes at infinity with a nonlinearity exhibiting critical or supercritical growth in the Sobolev sense. To overcome the lack of compactness and the difficulties imposed by the general structure of the nonlinearity, we employ variational methods combined with a penalization technique. Unlike the classical fractional Laplacian framework, where specific regularity results, decay estimates, and the $s$-harmonic extension are available, our approach relies on a weak Maximum Principle combined with the construction of a supersolution based on the truncated fundamental solution of the fractional Laplacian to control the asymptotic behavior of the solutions. We prove that, for sufficiently small perturbation parameters and under suitable decay conditions on the potential, the equation admits a nontrivial solution.
Read the full abstract

Wednesday 8 April

History and Overview of Mathematics

A Gas-Driven Algorithm for Variants of the Moving Sofa Problem

This paper presents a numerical algorithm based on the idea of dynamical systems for solving the moving sofa problem. By introducing a physical model driven by gas pressure, we transform the geometric optimization problem into a dynamical system. Numerical experiments show that the method effectively approximates the known Gerver's sofa result, and provides numerical estimates of the possible maximum area for corridors with different angles. In particular, we observe an intersection of the two motion patterns at $43.327\ldots^\circ$, where the locally maximal sofa area is $1.8674\ldots$, and the dominant pattern switches on either side of this critical angle.
Read the full abstract

Optimization and Control · Cross-listed from eess.SY

A Unified Control Theory Derivation of Discrete-Time Linear Ensemble Kalman Filters

The ensemble Kalman filter (EnKF) has become a standard methodology for state estimation in high-dimensional systems, yet its various stochastic and deterministic formulations often appear conceptually disconnected. In this paper, a unified derivation framework for EnKF algorithms are established by leveraging the classical duality between estimation and optimal control, which is the key concept in deriving Kalman filter. By recasting the minimum variance estimation problem into second order moment for the ensembles, we demonstrate that seemingly distinct EnKF variants – both with or without perturbed observation – can be systematically classified. Specifically, the duality based framework reveals that the operational differences among these variety of EnKF algorithms reduce to a specific choice of hyperparameters. Ultimately, this perspective not only covers existing EnKF variants but also provides a systematic foundation for designing novel hybrid filters using control theory approach.
Read the full abstract

General Mathematics

Scale-valued sets: a minimal framework for generalized set models

Many generalized set models have the same basic form: they assign a value to each object, and the main difference lies in the kind of values that are allowed. This paper studies that common form through scale-valued sets (SV-sets), defined as maps $U\times E\to\Sigma$, where $U$ is a universe, $E$ is a parameter set, and $\Sigma$ is a bounded De Morgan lattice. With a suitable choice of scale, SV-sets include ordinary sets, fuzzy sets, soft sets, bounded multisets, intuitionistic fuzzy sets, $L$-fuzzy sets, and Type-2 fuzzy sets. We study the basic structure of SV-sets. The relation between SV-sets and lattice-valued interval soft sets is also discussed. For complete chains, the SV setting gives a natural topological construction, and for groups, it gives an algebraic structure through SV-subgroups. The applications show how graded suitability and supporting evidence can be kept together in a single model, whereas one-coordinate reductions lose information.
Read the full abstract

Spectral Theory

Typical hyperbolic surfaces have a spectral gap greater than $2/9 - \epsilon$

In this article, we prove that typical hyperbolic surfaces, sampled with the Weil-Petersson probability measure, have a spectral gap at least $2/9 - \epsilon$. This is an intermediate result on the way to our proof of the optimal spectral gap $1/4 - \epsilon$, building on the results of the first part of this series. A significant part of the proof is an explicit inclusion-exclusion argument to exclude tangles at the level of precision $1/g$.
Read the full abstract

Dynamical Systems

Generalized Kolmogorov systems with applications to astrophysics and biology

We consider generalized Kolmogorov system. We prove the existence of heteroclinic trajectory. We apply the results to astrophysical models for self-gravitating particles and predator-prey systems.

General Mathematics

Gertsch quotient living in the "poor man's adele ring" $\mathcal{A}$: Kurepa-Bell-Wilson congruence

Wilson's theorem is notably related to left factorials, expressed as $K_p \equiv \mathbf{Bell}_{p-1} - 1 \pmod p$, for prime $p\geq3$. This study examines a Kurepa-Bell-Wilson congruence (KBW), $\frac{K_p + 1}{p}\equiv \frac{ \mathbf{Bell}_{p-1}}{p}+ W_p \pmod{p}$, and demonstrates that it naturally generates the non-zero "Gertsch quotient ($\mathbb{G}_p$)," which, for larger primes modulo $p$ resides in the poor man's adele ring $\mathcal{A}$ .
Read the full abstract

Information Theory · Cross-listed from cs.LG

Joint Interference Detection and Identification via Adversarial Multi-task Learning

Precise interference detection and identification are crucial for enhancing the survivability of communication systems in non-cooperative wireless environments. While deep learning (DL) has advanced this field, existing single-task learning (STL) approaches neglect inherent task correlations. Furthermore, emerging multi-task learning (MTL) methods often lack a theoretical foundation for quantifying and modeling task relationships. To bridge this gap, we establish a theoretically grounded MTL framework for joint interference detection, modulation identification, and interference identification. First, we derive an upper bound for the weighted expected loss in MTL frameworks. This bound explicitly connects MTL performance to task similarity, quantified by the Wasserstein distance and learnable task relation coefficients. Guided by this theory, we present the adversarial multi-task interference detection and identification network (AMTIDIN), which integrates adversarial training to minimize distributional discrepancies across tasks and uses adaptive coefficients to model task correlations dynamically. Crucially, we conducted a quantitative analysis of task similarity to reveal intrinsic task relationships, specifically that modulation identification and interference identification share a substantial feature overlap distinct from interference detection. Experiments demonstrate that AMTIDIN outperforms its independently trained single-task counterparts and MTL baselines under the evaluated conditions of limited training data, short signal lengths, and low signal-to-noise ratios (SNRs)
Read the full abstract

All authors H. Xu, L. Hu, B. He, S. Wang

Numerical Analysis

Universal, sample-optimal algorithms for recovery of anisotropic functions from i.i.d. samples

A key problem in approximation theory is the recovery of high-dimensional functions from samples. In many cases, the functions of interest exhibit anisotropic smoothness, and, in many practical settings, the nature of this anisotropy may be unknown a priori. Therefore, an important question involves the development of universal algorithms, namely, algorithms that simultaneously achieve optimal or near-optimal rates of convergence across a range of different anisotropic smoothness classes. In this work, we consider universal approximation of periodic functions that belong to anisotropic Sobolev spaces and anisotropic dominating mixed smoothness Sobolev spaces. Our first result is the construction of a universal algorithm. This recasts function recovery as a sparse recovery problem for Fourier coefficients and then exploits compressed sensing to yield the desired approximation rates. Note that this algorithm is nonadaptive, as it does not seek to learn the anisotropic smoothness of the target function. We then demonstrate optimality of this algorithm up to a dimension-independent polylogarithmic factor. We do this by presenting a lower bound for the adaptive $m$-width for the unit balls of such function classes. Finally, we demonstrate the necessity of nonlinear algorithms. We show that universal linear algorithms can achieve rates that are at best suboptimal by a dimension-dependent polylogarithmic factor. In other words, they suffer from a curse of dimensionality in the rate – a phenomenon which justifies the necessity of nonlinear algorithms for universal recovery.
Read the full abstract

Mathematical Physics · Cross-listed from hep-th

Vacuum-induced current density from a magnetic flux threading a cosmic dispiration in $(D+1)$-dimensional spacetime

We investigate the vacuum-induced current density for a charged scalar field in a $(D+1)$-dimensional cosmic dispiration spacetime threaded by a magnetic flux. This background combines a cosmic string and a screw dislocation, yielding a nontrivial helical geometry. By constructing the normalized mode functions of the Klein–Gordon equation, we evaluate the Wightman function and obtain the vacuum expectation value of the current density. We show that, in addition to the azimuthal component describing a persistent current around the defect, a nonvanishing axial component is induced as a direct consequence of the helical structure of the spacetime. Both components are periodic functions of the magnetic flux, depending only on its fractional part, reflecting the Aharonov–Bohm nature of the effect. Closed expressions are obtained for both massive and massless fields in arbitrary dimensions. We demonstrate that the screw dislocation parameter plays a crucial role in the behavior of the induced currents, leading to the regularization of the axial component at the origin and controlling its magnitude. The asymptotic behavior of both components is analyzed in detail. Our results reduce to known expressions in the absence of the screw dislocation, providing a consistency check. In particular, we examine the physically relevant $(3+1)$-dimensional case, where numerical analysis reveals nontrivial features arising from the interplay between topology and gauge effects.
Read the full abstract

Probability

Limiting Root Distribution of Random Log-concave Polynomials

Log-concave polynomials arise naturally in combinatorics, geometry, and probability. We introduce a natural family of random log-concave polynomials and study their limiting root distributions. Our starting point is the uniform model, in which the coefficient vector $(a_0,\ldots,a_n)$ is chosen uniformly from $[0,1]^{n+1}$ subject to the log-concavity constraint. More generally, for $\alpha\ge0$, we define \[ P_{n,\alpha}(z)=\sum_{i=0}^n a_i^{n^\alpha}z^i. \] Thus, $\alpha=0$ corresponds to the uniform model. For each fixed $0\le\alpha<1$, we prove that the empirical root measure converges almost surely to the uniform probability measure on the unit circle. For the critical case $\alpha=1$, which we call the {\emph beta model}, the empirical root measure instead converges in probability to a rotationally invariant probability measure that is absolutely continuous with respect to planar Lebesgue measure.
Read the full abstract

Optimization and Control · Cross-listed from cs.RO

Safe Large-Scale Robust Nonlinear MPC in Milliseconds via Reachability-Constrained System Level Synthesis on the GPU

We present GPU-SLS, a GPU-parallelized framework for safe, robust nonlinear model predictive control (MPC) that scales to high-dimensional uncertain robotic systems and long planning horizons. Our method jointly optimizes an inequality-constrained, dynamically-feasible nominal trajectory, a tracking controller, and a closed-loop reachable set under disturbance, all in real-time. To efficiently compute nominal trajectories, we develop a sequential quadratic programming procedure with a novel GPU-accelerated quadratic program (QP) solver that uses parallel associative scans and adaptive caching within an alternating direction method of multipliers (ADMM) framework. The same GPU QP backend is used to optimize robust tracking controllers and closed-loop reachable sets via system level synthesis (SLS), enabling reachability-constrained control in both fixed- and receding-horizon settings. We achieve substantial performance gains, reducing nominal trajectory solve times by 97.7% relative to state-of-the-art CPU solvers and 71.8% compared to GPU solvers, while accelerating SLS-based control and reachability by 237x. Despite large problem scales, our method achieves 100% empirical safety, unlike high-dimensional learning-based reachability baselines. We validate our approach on complex nonlinear systems, including whole-body quadrupeds (61D) and humanoids (75D), synthesizing robust control policies online on the GPU in 20 milliseconds on average and scaling to problems with 2 x 10^5 decision variables and 8 x 10^4 constraints. The implementation of our method is available at https://github.com/Jeff300fang/gpu_sls.
Read the full abstract

Combinatorics

On the connected Turán number of Berge paths and Berge cycles

Given a graph $F$, a Berge copy of $F$ (Berge-$F$ for short) is a hypergraph obtained by enlarging the edges arbitrarily. Győri, Salia and Zamora determined the maximum number of hyperedges in a connected $r$-uniform hypergraph on $n$ vertices containing no Berge path of length $k-1$ for all $k\geq 2r+14$ and sufficiently large $n$, and asked for the minimum $k_0$ such that this extremal number holds for all $k\geq k_0$. In this paper, we prove that the extremal number holds for all $k\geq 2r+2$ and fails for $k\le 2r+1$, thereby completely resolving the problem posed by Győri, Salia and Zamora. Moreover, we improve the result of Füredi, Kostochka and Luo, who determined the maximum number of hyperedges in a $2$-connected $n$-vertex $r$-uniform hypergraph containing no Berge cycle of length at least $k$ for all $k\geq 4r$ and sufficiently large $n$, by showing that this extremal number holds for all $k\geq 2r+2$ and fails for $k\le 2r+1$. Our approach reduces Berge-Turán problems to classical extremal graph theory problems, and applies recent work of Ai, Lei, Ning and Shi concerning the feasibility of graph parameters and the Kelmans operation.
Read the full abstract

Information Theory · Cross-listed from quant-ph

Exponential quantum advantage in processing massive classical data

Broadly applicable quantum advantage, particularly in classical data processing and machine learning, has been a fundamental open problem. In this work, we prove that a small quantum computer of polylogarithmic size can perform large-scale classification and dimension reduction on massive classical data by processing samples on the fly, whereas any classical machine achieving the same prediction performance requires exponentially larger size. Furthermore, classical machines that are exponentially larger yet below the required size need superpolynomially more samples and time. We provide evidence for these quantum advantages in real-world applications, including single-cell RNA sequencing and movie review sentiment analysis, demonstrating four to six orders of magnitude reduction in size with fewer than 60 logical qubits. These quantum advantages are enabled by quantum oracle sketching, an algorithm for accessing the classical world in quantum superposition using only random classical data samples. Combined with classical shadows, our algorithm circumvents the data loading and readout bottleneck to construct succinct classical models from massive classical data, a task provably impossible for any classical machine that is not exponentially larger than the quantum machine. These quantum advantages persist even when classical machines are granted unlimited time or if BPP = BQP, and rely only on the correctness of quantum mechanics. Together, our results establish machine learning on classical data as a broad and natural domain of quantum advantage and a fundamental test of quantum mechanics at the complexity frontier.
Read the full abstract

All authors Haimeng Zhao, Alexander Zlokapa, Hartmut Neven, Ryan Babbush, John Preskill, Jarrod R. McClean, Hsin-Yuan Huang

Group Theory

On the separability of some Green biset functors

We show that the Green biset functor $R_{\mathbb{C}}$ of complex characters over $\mathbb{Z}$, is not separable, i.e. it is not projective as a bimodule over itself. Also, we show that $RB_G$, the Burnside biset functor shifted by a finite group $G$, over a commutative ring $R$, is separable if and only if $|G|$ is invertible in $R$. Finally, to address the question of the relation between functors and their evaluations, we show that the Burnside $R$-algebra $RB(G)$ is separable if and only if $|G|$ is invertible in $R$.
Read the full abstract

Complex Variables

New local characterizations of the weighted energy class $\mathcal{E}_{\chi,\mathrm{loc}}(\Omega)$

Let \(\Omega\subset\mathbb{C}^n\) be a hyperconvex domain and let \(\chi:\mathbb{R}^-\to\mathbb{R}^+\) be a decreasing function. This note studies the local weighted energy class \(\mathcal{E}_{\chi,\mathrm{loc}}(\Omega)\) introduced in \cite{HHQ13}. We establish two main results on local membership in this class. First, we prove a new local boundedness property for the weighted Monge–Ampère energy: if \(u\in\mathrm{PSH}^-(\Omega)\) admits suitable local majorants in \(\mathcal{E}_{\chi,\mathrm{loc}}\) near the boundary of every relatively compact hyperconvex subdomain \(D\Subset\Omega\), then the weighted energy \(\int_K \chi(u)(dd^c u)^n\) remains locally finite for every compact set \(K\subset D\). This gives the first explicit local control of the energy functional and is new even in the unweighted setting. Second, we obtain a substantial improvement concerning the local control of the Monge–Ampère measure. We show that if, in addition to the boundary condition, \((dd^c u)^n\) is locally dominated by \((dd^c w)^n\) for some \(w\in\mathcal{E}_{\chi,\mathrm{loc}}(D)\) inside \(D\), then \(u\in\mathcal{E}_{\chi,\mathrm{loc}}(D)\). This domination condition is strictly weaker than the previous requirement of local finiteness of the weighted energy, thereby significantly enlarging the class of admissible functions. Our results extend and refine the local theory developed in \cite{Q24,Q25} and provide a more flexible framework for plurisubharmonic functions with possible singularities on compact subsets.
Read the full abstract

Logic

On Petr Novikov's problem of ordered systems of uniform sets

We prove that every ordinal $\alpha<\omega_2$ is the order type of a certain system of uniform Borel sets in the sense of a well-ordering relation defined by Petr Novikov. This result gives a positive answer to a problem posed by Nicolas Luzin in 1935.

Statistics Theory

Data Reuse and the Long Shadow of Error: Splitting, Subsampling, and Prospectively Managing Inferential Errors

When multiple investigators analyze a common dataset, the data reuse induces dependence across testing procedures, affecting the distribution of errors. Existing techniques of managing dependent tests require either cross-study coordination or post-hoc correction. These methods do not apply to the current practice of uncoordinated groups of researchers independently evaluating hypotheses on a shared dataset. We investigate the use of subsampling techniques implemented at the level of individual investigators to remedy dependence with minimal coordination. To this end, we establish the asymptotic joint normality of test statistics for the class of asymptotically linear test statistics, decomposing the covariance matrix as the product of a data overlap term and a test statistic association term. This decomposition shows that controlling data overlap is sufficient to control dependence, which we formalize through the notion of Expected Variance Ratio. This enables the closed form derivation of the variance of the joint rejection region under the global null as a function of pairwise correlations of test statistics. We adopt mean-variance portfolio theory to measure risk, defining the Expected Variance Ratio (EVR) as the ratio of the expected variance of the Type I error count to the independent baseline. We show that data splitting is asymptotically optimal among rules that ensure exact independence. We then use concentration inequalities to establish that subsampling techniques implementable by individual investigators can ensure an EVR close to $1$. Finally, we show that such subsampling techniques are able to simultaneously perform a number of tests while ensuring sufficient power and that the bounded EVR is $O\left(\frac{1}{r^2}\right)$ compared to data splitting's $O\left(\frac{1}{r}\right)$, where $r$ is the per-statistic fraction of data required.
Read the full abstract

All authors Reid Dale, Jordan Rodu, Maria E. Currie, Mike Baiocchi

Probability

Topology of Percolation Clusters: Central Limit Theorems beyond the Lattice

We prove central limit theorems (CLTs) for topological functionals of Bernoulli bond percolation on infinite graphs beyond the Euclidean lattice $\mathbb{Z}^{d}$. For quasi-transitive graphs of subexponential growth, we show that the number $K_{r}$ of open clusters intersecting the metric ball $B_{r}$ satisfies a CLT as $r\to\infty$. For amenable Cayley graphs, we prove a general CLT for stationary percolation functionals along Folner sequences under sequential stabilization and a finite-moment assumption, provided the group admits a left-orderable finite-index subgroup. This applies in particular to groups of polynomial growth. As an application, we obtain CLTs for Betti numbers of graph-generated random simplicial complexes, including clique and neighbor complexes. The proofs combine invariant edge orderings, martingale decompositions, and stabilization estimates for single-edge perturbations.
Read the full abstract

Numerical Analysis · Cross-listed from cs.CV

Mathematical Analysis of Image Matching Techniques

Image matching is a fundamental problem in Computer Vision with direct applications in robotics, remote sensing, and geospatial data analysis. We present an analytical and experimental evaluation of classical local feature-based image matching algorithms on satellite imagery, focusing on the Scale-Invariant Feature Transform (SIFT) and the Oriented FAST and Rotated BRIEF (ORB). Each method is evaluated through a common pipeline: keypoint detection, descriptor extraction, descriptor matching, and geometric verification via RANSAC with homography estimation. Matching quality is assessed using the Inlier Ratio - the fraction of correspondences consistent with the estimated homography. The study uses a manually constructed dataset of GPS-annotated satellite image tiles with intentional overlaps. We examine the impact of the number of extracted keypoints on the resulting Inlier Ratio.
Read the full abstract

Information Theory · Cross-listed from cs.LG

Learning is Forgetting: LLM Training As Lossy Compression

Despite the increasing prevalence of large language models (LLMs), we still have a limited understanding of how their representational spaces are structured. This limits our ability to interpret how and what they learn or relate them to learning in humans. We argue LLMs are best seen as an instance of lossy compression, where over training they learn by retaining only information in their training data relevant to their objective(s). We show pre-training results in models that are optimally compressed for next-sequence prediction, approaching the Information Bottleneck bound on compression. Across an array of open weights models, each compresses differently, likely due to differences in the data and training recipes used. However even across different families of LLMs the optimality of a model's compression, and the information present in it, can predict downstream performance on across a wide array of benchmarks, letting us directly link representational structure to actionable insights about model performance. In the general case the work presented here offers a unified Information-Theoretic framing for how these models learn that is deployable at scale.
Read the full abstract

All authors Henry C. Conklin, Tom Hosking, Tan Yi-Chern, Julian Gold, Jonathan D. Cohen, Thomas L. Griffiths, Max Bartolo, Seraphina Goldfarb-Tarrant

Probability · Cross-listed from stat.ME

Marginal Persistence and Dynamic Copula Dependence in Sovereign Rating Migration Counts: A Discrete Interval-Likelihood MAGMAR Analysis

This paper develops an observed-data likelihood for applying moving-aggregate modified autoregressive (MAGMAR) copula time-series models to discrete sovereign rating-migration counts with time-varying exposure. An annual count identifies a probability-integral-transform interval rather than a unique latent point, so the likelihood integrates the latent process over the complete sequence of count intervals. A guided sequential Monte Carlo implementation incorporates the parameter-dependent stationary marginal adjustment of MAGMAR directly inside this integration. In sovereign ratings from Fitch, Moody's, and S&P over 1963–2025, the migration counts are strongly overdispersed relative to a binomial margin and, under an exposure-conditioned static beta-binomial margin, MAGMAR improves substantially on MAG. The interpretation changes once the marginal mean is allowed to depend on the lagged migration rate: residual serial correlation in the count margin disappears, a reference dynamic-margin fit gives only a small MAGMAR improvement, and information criteria favor MAG. Numerical profiles, higher-precision refits, and parameter-recovery experiments show that marginal persistence and copula persistence are only weakly separated in this short discrete series. The results therefore support interval integration as the correct observation layer while showing that conclusions about latent dynamic dependence must be conditioned on the specification of the count margin.
Read the full abstract

Differential Geometry

The eta invariant of a circle bundle on a Fano manifold

We consider the spin-c Dirac operator on the unit circle bundle of a positive line bundle over a Fano manifold of even complex dimension. We compute the corresponding eta invariant in terms of Zhang's value of its adiabatic limit. This extends the earlier computation of the author from small to arbitrary values of the adiabatic parameter.
Read the full abstract

Analysis of PDEs

Ergodic Mean Field Games of Controls with State Constraints

In a mean field game of controls, players seek to minimize a cost that depends on the joint distribution of players' states and controls. We consider an ergodic problem for second-order mean field games of controls with state constraints, in which equilibria are characterized by solutions to a second-order MFGC system where the value function blows up at the boundary, the density of players vanishes at a commensurate rate, and the joint distribution of states and controls satisfies the appropriate fixed-point relation. We prove that such systems are well-posed in the case of monotone coupling and Hamiltonians with at most quadratic growth.
Read the full abstract

Combinatorics

Density of reliability roots of simple graphs in the unit disk

Brown and Colbourn (1992) showed that the complex roots of the reliability polynomial of connected multigraphs are dense in the unit disk and that the closure of the real roots is $[-1,0] \cup \{1\}$. We prove the simple graph analogues of both results, confirming a recent conjecture of Brown and McMullin. The proof uses the family of graphs $C_m[K_n]$ obtained by substituting each edge of a cycle $C_m$ with a complete graph $K_n$, and relies on the asymptotic behavior of the reliability and split reliability polynomials of $K_n$.
Read the full abstract

Analysis of PDEs

Partial regularity for $\mathscr{A}$-quasiconvex variational problems of linear growth

We prove that minimizers of variational integrals $$ \mathcal E(v)=\int_\Omega f(v)\quad\text{for }v\in\mathcal M(\Omega)\text{ such that } \mathscr{A} v=0, $$ are partially continuous provided that the integrands $f$ are strongly $\mathscr{A}$-quasiconvex in a suitable sense. We consider linear growth problems, linear PDE operators $\mathscr{A}$ of constant rank, and variations of the form $v+\varphi$ with $\mathscr{A}$-free $\varphi\in \mathrm{C}_{\mathrm{c}}^\infty(\Omega)$. Our analysis also covers the “potentials case” $$ \mathcal F(u)=\int_\Omega f( \mathscr{B} u)\quad\text{for } u\in\mathscr D'(\Omega)\text{ such that }\mathscr B u\in \mathcal M(\Omega), $$ where $\mathscr{B}$ is a different linear pde operator of constant rank. Both our main results extend to $x$-dependent integrands.
Read the full abstract