Peer Revenue Past citations are no guarantee of future returns

Papers

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

Mathematics

math Watch Mathematics (sign in to use)

Tuesday 28 April

Combinatorics

Eigenvalues of $\boldsymbol{L_\alpha}-$matrices under graph operations

Let $G$ be a simple graph, $A(G)$ its adjacency matrix, and $D(G)$ its diagonal degree matrix. In 2022, \citeauthor{Wang2020} (\cite{Wang2020}) defined the family of matrices $L_\alpha$ as the convex linear combination: \[ L_\alpha(G) = \alpha D(G) + (\alpha - 1)A(G), \] where $\alpha \in [0,1]$. The study of the spectrum of this family of matrices may provide a unified framework for analyzing the spectra of the adjacency, degree, and Laplacian matrices ($D(G) - A(G)$). In this work, we investigate the spectrum of $L_\alpha$ under graph operations and within specific families of graphs.
Read the full abstract

Rings and Algebras

Undecidability problems for semifree DG algebras

We prove that the stable tame isomorphism, quasi-isomorphism, and derived Morita equivalence problems for semifree noncommutative differential graded algebras (DGAs) are all undecidable. This resolves half of Problem 5.16 from the K3 Problem List in Low-Dimensional Topology. We present two solutions, both obtained (essentially autonomously) by Gemini Deep Think / Aletheia.
Read the full abstract

Group Theory

The Inverse Monoid of Partial Inner Automorphisms of a Semigroup

We introduce the inverse monoid of inner partial automorphisms of a semigroup – a tool that associates to every semigroup an inverse semigroup. When the semigroup is a group, this inverse semigroup is isomorphic to the group of inner automorphisms with a zero adjoined. We then describe this structure for completely simple semigroups, the full transformation monoid, and the endomorphism monoid of a finite $G$-set when $G$ is a finite abelian group. The paper ends with some open problems.
Read the full abstract

All authors João Araújo, Wolfram Bentz, Michael Kinyon, Janusz Konieczny, António Malheiro, Valentin Mercier

Number Theory

The cardinality of a set containing the pairwise sums of a fixed number of integers

Revisiting a $50$-year-old estimate of Choi, Erdős and Szemerédi, we show that if $A \subseteq \{1, 2, \ldots, 2n\}$ satisfies $|A| \ge n + 1.2 \cdot 10^8$, then there exist five distinct integers whose pairwise sums are all contained in $A$. In order to guarantee pairwise sums of three or four integers instead, we show that one can replace the constant $1.2 \cdot 10^8$ by $1$ or $3$ respectively, which are both optimal.
Read the full abstract

Numerical Analysis

State-Dependent Lyapunov Analysis of Rank-1 Matrix Factorization

We develop a state-dependent Lyapunov framework for gradient descent on rank-1 matrix factorization. A parameterized quadratic certificate $I(\delta;\,\cdot)$ generates strictly nested sublevel sets. Their ordering assigns each point a state $\delta$, while a boundary-inward property makes this state monotone along gradient-descent trajectories. Together with internal chain transitivity, this geometry identifies the limiting dynamics even when all relevant stationary points are unstable. For scalar and rank-1 factorization, it yields convergence to a global minimizer below the stability threshold and to a balanced period-$2$ orbit in a post-critical interval, for almost every initialization in an explicit region. We formalize the mechanism through structural and dynamical axioms. Within the origin-centered, exchange-symmetric quadratic class, the axioms determine the ordered level-set geometry uniquely up to a monotone relabeling of the state, recovering the scalar geometry identified by Liang–Montúfar. Additional analytic and numerical examples suggest broader applicability.
Read the full abstract

Statistics Theory

Adaptive Confidence Intervals in Efron's Gaussian Two-Groups Model

Robust uncertainty quantification is increasingly important in modern data analysis and is often formalized under Huber's model, which allows an $\varepsilon$-fraction of arbitrary corruptions. In many experimental sciences, however, the measurement protocol is well controlled, and contamination is more plausibly introduced upstream. Motivated by this noise-oblivious nature of adversaries, we study confidence intervals for the null location parameter $\theta$ in Efron's Gaussian two-groups model, where an unknown fraction $\varepsilon$ of observations have arbitrarily shifted means, but all samples share the same law of additive Gaussian measurement noise with variance $\sigma^2$. We characterize the minimax-optimal length among confidence intervals with a prescribed coverage level uniformly over the unknown contamination proportion and all noise-oblivious adversaries. Although prior work has shown that the minimax point estimation rate of theta does not deteriorate when $\varepsilon$ becomes unknown, our results reveal that, with a given $\sigma^2$, the minimax-optimal length of confidence intervals that are adaptive to unknown $\varepsilon$ is of order $\sigma (n^{-1/4}+\varepsilon^{1/2}/\max\{1, \log(en \varepsilon^2)\}^{1/2})$, which is polynomially worse than the optimal length when $\varepsilon$ is known. When the variance $\sigma^2$ is also unknown, we show a further degradation: no adaptive confidence interval can be shorter than $\Omega(\sigma n^{-1/8})$. Algorithmically, we introduce a Fourier-based certification procedure built on Carathéodory's positive-semidefiniteness constraints. By scanning candidate points and accepting those whose residual characteristic function is certifiably consistent with a Gaussian location mixture, our algorithm attains the minimax lower bound in the known-variance setting and is computable in polynomial time.
Read the full abstract

Combinatorics

Subgroups of Finite Fields As Cap Sets

We show the subgroup of 20 nonzero fourth powers in the finite field of order 81 is a cap set. Similarly, the subgroup of 9 nonzero seventh powers in the field of order 64 is a cap set. These are the cases related to the card games of SET and EvenQuads, and both are known to be maximal cap sets. A corollary is that the cosets of these subgroups form a partition by maximal caps of the multiplicative groups of their respective fields. We identify certain multiplicative subgroups of fields of orders 243 and 729 as cap sets, and show in general that the subgroup of $(2^n-1)$-th powers is a cap set in the field of order $2^{2n}$.
Read the full abstract

Differential Geometry

Positive mass theorem for initial data sets with arbitrary ends

We showed a positive energy theorem for asymptotically flat initial data sets with the concept of spectral PSC by He-Shi-Yu, Bi-Hao-He-Shi-Zhu and Brendle-Wang; and the Jang equation in Schoen-Yau, Eichmair and Jang. Then, we proved a quantitative shielding theorem concerning the causal property of the energy-momentum vector of an asymptotically hyperbolic manifold. As a result, we established the positive mass theorem for complete asymptotically hyperbolic manifolds satisfying the dominant energy condition. As corollaries, we also obtained corresponding results for manifolds with asymptotically locally hyperbolic ends with a certain symmetry.
Read the full abstract

Statistics Theory

Asymptotics of ultra-high-dimensional generalized spiked sample covariance matrix

This paper investigates the asymptotics of eigenstructure of sample covariance matrix under the spiked covariance matrix model in ultra-high-dimensional settings, where the dimensionality can grow much faster than the sample size with $ p \asymp n^{\alpha} $, $ \alpha > 1 $. We establish the first-order convergence limits of eigenvalue locations and eigenvector projections of properly scaled sample covariance matrix. Our results are extensions of \cite{bloemendal16,ding21}.
Read the full abstract

Algebraic Geometry

Components of strata of k-differentials and their orbit closures

We obtain a complete classification of components of strata of holomorphic and meromorphic k-differentials. We show that, when genus is at least two and outside of explicit exceptions when k < 4, there is one primitive nonhyperelliptic component unless k is odd and all singularities have even order, in which case there are two distinguished by their Arf invariant. The exceptions include new sporadic components of strata of cubic differentials. Our work provides a new proof of earlier results of Kontsevich-Zorich, Boissy, Lanneau, and Chen-Gendron when k = 1, 2. The proofs are almost purely algebraic, relying on the multiscale compactification of Bainbridge-Chen-Gendron-Grushevsky-Moller. This answers a question of Chen-Yu. We also show that for any component of a stratum of finite area $k$-differentials of positive genus, the smallest GL(2,R)-orbit closure containing all of its holonomy covers is as big as possible. This answers the positive genus analogue of a "long term goal" of Mirzakhani-Wright. The result also holds in genus zero strata that contain surfaces with Euclidean cylinders, thus addressing infinitely many cases of the original question of Mirzakhani-Wright.
Read the full abstract

Optimization and Control · Cross-listed from eess.SY

Co-Learning Port-Hamiltonian Systems and Optimal Energy-Shaping Control

We develop a physics-informed learning framework for energy-shaping control of port-Hamiltonian (pH) systems from trajectory data. The proposed approach co-learns a pH system model and an optimal energy-balancing passivity-based controller (EB-PBC) through alternating optimization with policy-aware data collection. At each iteration, the system model is refined using trajectory data collected under the current control policy, and the controller is re-optimized on the updated model. Both components are parameterized by neural networks that embed the pH dynamics and EB-PBC structure, ensuring interpretability in terms of energy interactions. The learned controller renders the closed-loop system inherently passive and provably stable, and exploits passive plant dynamics without canceling the natural potential. A dissipation regularization enforces strict energy decay during training, thereby enhancing robustness to sim-to-real gaps. The proposed framework is validated on state-regulation and swing-up tasks for planar and torsional pendulum systems.
Read the full abstract

Differential Geometry

A note on four dimensional Shrinking Gradient Ricci Solitons with Constant Scalar Curvature

Let $(M^4, g, f)$ be a four-dimensional complete noncompact gradient shrinking Ricci soliton with the equation $Ric+\nabla^2f= \frac{1}{2}g$. If its scalar curvature is $1$, Cheng-Zhou \cite{Cheng-Zhou} proved that it is a finite quotient of $\mathbb{R}^2\times \mathbb{S}^2$. In this note we present an alternative proof by analyzing the asymptotic geometry at infinity.
Read the full abstract

Representation Theory · Cross-listed from hep-th

Quantum Flat Connections, KZ equations, and Integrability

N=2 supersymmetric Yang-Mills theories are described in terms of a Hitchin system over a Riemann surface C. Focusing on strongly coupled Argyres-Douglas theories, we show that the corresponding flat bundle over C can be quantized such that the resulting quantum flat connection is integrable. For $sl_2$, the quantum connection takes values in $gl_2$(A) where A is an associative algebra which we explicitly describe for the cases of Painlevé I, II and IV. Moreover, we find that the quantum connection is equivalent to irregular versions of Knizhnik-Zamolodchikov (KZ) connections. Utilizing a suitable gauge transformation, one can show that the corresponding KZ equations give rise to BPZ equations.
Read the full abstract

Combinatorics

A Schur-positivity classification for complete multipartite graphs

A graph is Schur-positive if its chromatic symmetric function expands non-negatively in the Schur basis. We determine a full Schur-positivity classification for complete multipartite graphs by showing that a complete multipartite graph $K_\lambda$ is Schur-positive if and only if either $\lambda_i\in \{1,2\}$ for all $i$ or $\lambda=(3,2^\beta)$ for some $\beta\ge 1$. These results extend earlier classifications for complete bipartite and complete tripartite graphs to full generality. Our proofs combine structural arguments ruling out most cases, with a combinatorial analysis of Schur coefficients for the remaining family $K_{(3,2^\beta)}$ via special rim hook $G$-tabloids. Along the way, we establish a simpler formula for Schur coefficients of incomparability graphs, which we then apply to compute the coefficients of interest in terms of non-increasing sequences.
Read the full abstract

Rings and Algebras

Exterior-Model Spinors in Split Rank: Exact Levi Images and Square-Determinant Obstructions

Let $K$ be a field with $2 \in K^\times$, and let $H_W$ denote the standard hyperbolic form on $W \oplus W^*$. We study the exterior spinor model $S = \bigwedge V(W)$ together with the spin-to-orthogonal map for this split form, keeping the chosen hyperbolic presentation explicit. The main results determine the field-sensitive part of the split Levi image. In positive split rank the kernel of $\mathrm{Spin}(V,Q) \to SO(V,Q)$ is $\{\pm 1\}$; therefore the exterior spinor action descends to the orthogonal image only projectively. For the split line the image of $\mathrm{Spin}(H_K) \to SO(H_K)$ is precisely the square-scaling subgroup. In arbitrary split rank we construct explicit Clifford representatives for hyperbolic transvections and chosen-line square scalings, prove the weight-2 torus conjugation law, and show that any split Levi lift acts on $\bigwedge V(W)$ as a scalar multiple of the natural exterior action. If $\det(g) \in u^2$, the transported Levi element $\hat{g} = (g, g^{-\top})$ admits an explicit even unitary Clifford lift acting as $u^{-1} \bigwedge(g)$ on $S$. In finite split rank at least three, if $H_W \in \mathrm{im}(\mathrm{Spin}(H_W) \to SO(H_W))$, then $g_{H_W} \in \det(g) \cdot K^{\times 2}$. Equivalently, the spin image meets the split Levi subgroup exactly in its square-determinant subgroup. This recovers, by direct Clifford calculation, the determinant-modulo-squares spinor-norm criterion on the split Levi.
Read the full abstract

Analysis of PDEs

Reconstruction of the Support of an Inhomogeneity for the Fractional Helmholtz Equation

We consider the inverse scattering problem for inhomogeneous media of compact support governed by the fractional s-Helmholtz equation, with $0<s<1$, in dimensions $d=1,2,3$. In particular, we study the determination of the support of the inhomogeneity from the far-field pattern of the scattered field generated by plane waves for all incident directions at a fixed frequency. The far-field pattern is defined as the principal term in the asymptotic expansion of the scattered field at infinity. It is shown in \cite{zilberberg2026limiting} that, up to a multiplicative constant, this coincides with the far-field pattern corresponding to the classical Helmholtz equation with the same inhomogeneity. Our approach is based on the development of the factorization method, which not only leads to an efficient and easily implementable reconstruction algorithm, but also provides a uniqueness result for determining the support of an admissible set of inhomogeneities. A fundamental ingredient in the analysis is a new transmission eigenvalue problem, whose eigenvalues must be excluded. Therefore, we prove that they are discrete with no finite accumulation points. We present numerical examples in dimension $d=2$ for both the direct and inverse problems.
Read the full abstract

Optimization and Control

Reinforcement Learning for Public Safety Power Shutoffs Under Decision-Dependent Uncertainty and Nonlinear Wildfire Ignition Models

Power grid infrastructure is an increasingly significant source of wildfire ignitions and poses severe risks to communities in fire-prone regions. Public Safety Power Shutoffs (PSPS) have emerged as a critical operational tool for utilities to mitigate this risk by proactively de-energizing portions of the grid under high-threat conditions. These shutoffs, however, impose costs on affected communities, and it is therefore essential that PSPS decisions be informed by realistic models of wildfire ignition risk. Current Mixed Integer Programming based methods require restrictive structural assumptions about the probability models for line failures caused by power line ignitions. While these simplifications yield tractable solutions, the resulting models may differ significantly from the true underlying dynamics. In this paper, we propose a reinforcement learning framework based on Proximal Policy Optimization that learns to adjust the topology of a distribution system by interacting directly with a simulator that accommodates any line failure probability model without imposing such restrictions. We test our methodology on 54-bus and 138-bus distribution systems and demonstrate its ability to lower operational costs compared to existing methods while allowing only marginally increased compute times as network size grows.
Read the full abstract

Optimization and Control

Reachability-Based Design Optimization for Aircraft Maneuverability

This paper presents a method for incorporating control analysis into design optimization for highly-maneuverable aircraft. By studying reachable sets for aircraft dynamics, we ensure that the optimizer will take the aircraft's controlled capabilities into account. We compute reachable sets of linear dynamics for computational efficiency, and account for aircraft trim points to factor in asymmetric magnitude bounds on the input signals. We demonstrate the proposed method in design optimization of a blended-wing-body aircraft. Considering its longitudinal dynamics, we optimize the aircraft based on its reachable sets to yield improvements in its controlled performance. When designing a reference tracking controller, we find up to 57% faster settling time with approximately 77% less control effort for the optimized model's nonlinear dynamics.
Read the full abstract

All authors Steven Nguyen, Nicholas Orndorff, Jorge Cortés, Boris Kramer

Algebraic Geometry

Generalizations and minimalistic refinements of the t-birational Section Conjecture

In this note we give generalizations and prove 'minimalistic' refinements of the t-birational Section Conjecture (t-BSC), cf. [Be], by doing both: First, by extending the class of base fields over which the t-BSC holds, and second, by proving refinements of the t-BSC which involve much less, that is minimalistic, Galois theoretical information.
Read the full abstract

Logic

Elimination results for tame fields with finite residue fields

Building on work of Kuhlmann and Lisinski, we study the theory of the Hahn series field $\mathbb{F}_{q}(\!(\mathbb{Q})\!)$, over a finite field $\mathbb{F}_{q}$, equipped with the $t$-adic valuation, in a language of valued fields. We prove that every formula is equivalent to a formula $\exists y\colon f(x_{1},\ldots,x_{n},y)=0$, for a polynomial $f\in\mathbb{Z}[x_{1},\ldots,x_{n},y]$.
Read the full abstract

Combinatorics

The shape of a random numerical semigroup

We study statistical properties of random numerical semigroups of a given genus. We analyze the graph of a typical numerical semigroup, understood as a function from $\mathbb{N}$ to $\mathbb{N}$. If $S$ is a numerical semigroup of genus $g$, this leads us to consider the collection of points $\left(\frac{k-1}{g-1},\frac{a_k(S)}{g} \right)$ where $1 \le k \le g$ and $a_k(S)$ denotes the $k$th smallest nonzero element of $S$. We show that as $g \rightarrow \infty$, this set of points typically becomes closer to a union of two line segments. We prove analogous results for numerical semigroups ordered by Frobenius number.
Read the full abstract

Mathematical Physics

The dynamical algebra of the generic superintegrable model on the two-sphere

The rank two Jacobi algebra $\mathfrak{J}_2$ is identified as the dynamical algebra of the generic quadratic superintegrable model on the two-sphere. The physical representation of this algebra is obtained from its embedding in $\mathfrak{su}(1,1)^{\otimes 3}$. The exact solution of the model is derived algebraically from this representation. The wavefunctions are found to be expressed in terms of two-variable Jacobi polynomials whose characterization is a by-product of the algebraic treatment of the model.
Read the full abstract

All authors Nicolas Crampé, Quentin Labriet, Lucia Morey, Satoshi Tsujimoto, Luc Vinet, Alexei Zhedanov

Numerical Analysis

An Asymptotic-Preserving Dual Formulation Finite-Volume Method for the Thermal Rotating Shallow Water Equations

We propose a new second-order asymptotic-preserving (AP) dual formulation finite-volume (DF-FV) method for the thermal rotating shallow water (TRSW) equations. The TRSW system models geophysical flows characterized by horizontal temperature/density variations, exhibiting multi-scale dynamics due to the coexistence of fast rotational waves and slower advective processes. To efficiently address challenges associated with the multiscale nature of the TRSW system, we follow the DF-FV framework and develop a DF-FV method, in which both the conservative and nonconservative (primitive) forms of the equations are simultaneously solved, allowing the method to exploit the complementary strengths of each representation across different flow regimes. The primitive formulation is better suited for preserving the correct asymptotic behavior in nearly thermal quasi-geostrophic (TQG) regimes characterized by a low Rossby number, while the conservative formulation is essential for robust shock capturing in high-Rossby-number regimes, in which nonconservative discretizations may fail to converge to physically relevant weak solutions.
Read the full abstract

All authors Alina Chertock, Alexander Kurganov, Lorenzo Micalizzi, Nan Zhang

Rings and Algebras

On simple transposed Poisson algebras

We develop a structure theory for transposed Poisson algebras over fields of characteristic different from two. In particular, we prove that every finite-dimensional transposed Poisson algebra over an algebraically closed field decomposes as the direct sum of a unital ideal and a nilpotent ideal. As a consequence, we obtain restrictions on simple transposed Poisson algebras and use them to classify the simple finite-dimensional transposed Poisson algebras over an algebraically closed field of characteristic $p>3$. Precisely, we show that every such algebra has as underlying Lie algebra a Zassenhaus algebra $\mathcal{W}(1;n)$ and is isomorphic to one of the algebras of the family $\mathcal{W}_n(q)$ arising from a mutation of a natural associative commutative structure on $\mathcal{W}(1;n)$. We then study the corresponding isomorphism problem for the family $\mathcal{W}_n(q)$ and determine the irreducible finite-dimensional representations of these simple transposed Poisson algebras $\mathcal{W}_n(q)$ in the unital case. We conclude with some applications to Jordan superalgebras, weak-Leibniz algebras and quasi-Poisson algebras.
Read the full abstract

Information Theory

Multi-TRP Assisted UAV Detection in 3GPP 5G-Advanced ISAC Network

ISAC is currently being standardized within the 3GPP New Radio (NR) to enable cellular infrastructure to perform sensing using existing communication waveforms. While standardization is progressing, practical deployment may be limited by scenario-dependent observability constraints. For example, in UMa-AV scenarios, sensing with a single TRP can be affected by restricted angular coverage, partial blockage, and limited field of view, which may degrade detection reliability in three-dimensional UAV environments. For this reason, multi-TRP solutions have been suggested to improve spatial diversity and sensing robustness. In this paper, we present a system-level investigation of multi-TRP assisted monostatic sensing for UAV detection under standardized 3GPP UMa-AV channel assumptions and Release 19 evaluation parameters. We propose a spatial diversity fusion framework and evaluate the achievable performance of a 3GPP network by combining the measurements obtained independently at different TRP. Extensive evaluations demonstrate that multi-TRP assistance improves target observability, reduces spurious detections, and tightens localization error distributions at the cost of additional sensing overhead due to the need for multiple TRPs to periodically allocate radio resources for sensing measurements. In the evaluated scenario, results show that a voting threshold of two assisting TRPs achieves an optimal trade-off between miss detection probability and false alarm suppression, meeting 3GPP performance objectives. Furthermore, we quantify the sensing overhead and show that proper system design, tuned to the application requirements, can substantially reduce its impact: for example, extending the sensing refresh interval beyond the 128 ms coherent processing interval to 1 s reduces the effective overhead from 29 % to approximately 3.7 %, enabling more scalable network deployment.
Read the full abstract

All authors Neeraj Varshney, Steve Blandino, Jian Wang, Anuraag Bodi, Camillo Gentile, Nada Golmie

Numerical Analysis

A New Asymptotic-Preserving Dual Formulation Finite-Volume Method for the Compressible Euler Equations

The paper focuses on the development of numerical methods for the compressible Euler equations. It is well-known that if the Mach number is small, the system becomes stiff and hence explicit schemes suffer from severe time-step restrictions, making them inefficient or even impractical. Our objective is to develop an asymptotic preserving (AP) scheme that remains uniformly accurate and stable across all Mach numbers. Instead of the conservative hyperbolic flux splitting approach, which is widely used to design AP schemes, we consider a primitive (nonconservative) formulation and introduce a nonconservative hyperbolic splitting. The resulting system is discretized using a semi-implicit approach: the stiff part is handled semi-implicitly using second-order central differences, while the nonstiff part is treated explicitly using a second-order path-conservative central-upwind discretization. A key feature of our method is that the pressure at each time level is computed by solving a well-posed Poisson-type elliptic equation, thereby enforcing the AP property. Simultaneously, we evolve the conservative form of the system using a semi-discrete central-upwind (CU) scheme. At the end of each stage of the time discretization, we perform a special post-processing that selects the appropriate numerical solution depending on the Mach number. This guarantees that in low-Mach-number regimes, the solution is obtained by the AP nonconservative scheme, while in higher-Mach-number regimes, a sharp and physically relevant solution is computed by the conservative CU scheme. Numerical experiments confirm that the proposed AP scheme achieves the expected second order of accuracy and that the time-step constraint is independent of the Mach number, making it a robust and efficient alternative to conventional explicit methods.
Read the full abstract

All authors Alina Chertock, Smadar Karni, Alexander Kurganov, Lorenzo Micalizzi

Combinatorics

Counterexamples to an Extremal Conjecture for Random Cycle-Factors

Christoph, Draganić, Girão, Hurley, Michel, and Müyesser conjectured that, when $d\mid n$, the expected number of cycles in a uniformly random cycle-factor of a directed $d$-regular graph on $n$ vertices is uniquely maximised by the disjoint union of $n/d$ copies of the complete looped digraph $K_d^\circ$, with value $(n/d)H_d$ [FOCS 2025]. We disprove this conjecture in the strongest possible range. For every $d\ge 3$ and every multiple $n=kd$ with $k\ge 2$, we construct a directed $d$-regular graph on $n$ vertices whose uniformly random cycle-factor has expected cycle count strictly larger than $kH_d$. We also show that the conjectured extremal picture is correct in degree $d=2$, giving a sharp dichotomy between degree two and all higher degrees.
Read the full abstract

Classical Analysis and ODEs

Uncertainty Principle for distributions with Fourier transform in $L_{p,q}(\mathbb{R}^d)$

A version of the Uncertainty Principle says: There does not exist a non zero function in $L_p(\mathbb{R}^d)$ if its Fourier transform is supported by a set of finite $\alpha$-Hausdorff measure with $\alpha<2d/p$. This UP does not hold at the endpoint $\alpha=2d/p$. We find the sharp form of the UP in the limit case. We prove that there exists a non-zero function in the Lorentz space $L_{p,q}(\mathbb{R}^d)$ such that its Fourier transform is supported by a set of zero $(\frac{2d}{p},\beta)$-Netrusov–Hausdorff capacity if and only if $\beta>\frac{q}{2(q-1)}$.
Read the full abstract

Dynamical Systems

Spectral Selection in Symmetric Self-Attention Dynamics

We study self-attention dynamics on the unit sphere as an interacting particle system arising from an idealized Transformer-type update. Under a symmetry assumption on weight matrices given by $Q^\top K=V=V^\top$, the flow admits a gradient-flow structure and an exact reformulation in the eigenbasis of $V$, revealing a spectral mode-selection mechanism. We show that the dynamics exhibits two distinct asymptotic scenarios: homogeneous alignment toward the dominant eigendirection when one positive eigenvalue strictly dominates all others in modulus, and sign-split polarization toward the most negative eigendirection when $V$ is negative definite. In particular, we obtain local stability criteria for pure-mode equilibria and global selection results in both regimes. These results provide a rigorous finite-particle description of how the spectrum of the weight matrices organizes asymptotic patterns in a symmetric self-attention flow, and highlight how the symmetric setting renders the dynamics amenable to mathematical analysis.
Read the full abstract