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)

Thursday 7 May

Numerical Analysis

Low-rank kernel methods for American option pricing

We propose a scalable and theoretically grounded low-rank conditional expectation model for recursive Monte Carlo optimal stopping problems, in particular American option pricing. Our method reformulates the estimation of continuation values as a learning problem in a reproducing kernel Hilbert space, in which the conditional expectation is represented as a linear operator acting on future payoffs. This perspective yields an offline-online decomposition: the operator is learned once from simulated data and subsequently reused across all exercise dates, eliminating the need to recompute regression models at each step of the backward recursion. We establish convergence guarantees and derive bounds quantifying the approximation errors across exercise dates. Numerical experiments demonstrate the speed and accuracy of the proposed approach relative to extant methods.
Read the full abstract

Optimization and Control · Cross-listed from cs.NE

CoupleEvo: Evolving Heuristics for Coupled Optimization Problems Using Large Language Models

Many real-world optimization problems consist of multiple tightly coupled subproblems whose solutions must be coordinated to achieve high overall performance. However, existing large language model driven automated heuristic design approaches are limited to single-problem settings. In this paper, we propose CoupleEvo. CoupleEvo proposes three evolutionary coordination strategies to evolve heuristics for coupled optimization problems: the sequential strategy evolves heuristics for one subproblem after the other; the iterative strategy alternates the evolution of heuristics for different subproblems over successive generations; and the integrated strategy evolves heuristics for all problems simultaneously. The approach is evaluated on two representative coupled optimization problems. Experimental results show that decomposition-based strategies (sequential and iterative) provide more stable convergence and higher solution quality, while the integrated evolution strategy suffers from increased search complexity and variability. These findings highlight the importance of coordinating evolutionary search across interdependent subproblems and demonstrate the potential of LLM-driven heuristic design for complex coupled optimization problems. The code is available: https://github.com/tb-git-kit-research/CoupleEvo.
Read the full abstract

All authors Thomas Bömer, Bastian Amberg, Max Disselnmeyer, Anne Meyer

Numerical Analysis

Nonlinear RMM-GKS for Large-Scale Dynamic and Streaming Inverse Problems with Uncertain Forward Operators

Many practical imaging systems suffer from uncertainty in acquisition geometry – such as projection angles in computed tomography or sensor positions in photoacoustic tomography – leading to nonlinear inverse problems that require joint estimation of both the image and the forward model parameters. Standard approaches that assume a known linear forward operator fail to account for these uncertainties, resulting in significant reconstruction artifacts. We propose a nonlinear recycled majorization-minimization generalized Krylov subspace (NL-RMM-GKS) framework for large-scale inverse problems with uncertain forward operators. The method extends MM-GKS to nonlinear settings by combining majorization-minimization for nonsmooth regularization with Krylov subspace projection and subspace recycling, ensuring bounded memory usage. Two complementary formulations are developed: an alternating minimization approach that alternates between image updates and Gauss-Newton parameter estimation, and a variable projection approach that eliminates the image variable and optimizes directly over the parameters using inexact inner solves. We further introduce streaming variants that process data sequentially, enabling reconstruction from large or dynamically acquired datasets without storing the full operator. For dynamic problems, we incorporate two temporal regularization strategies – optical flow and anisotropic total variation – as plug-in choices within the framework. We carry out rigorous numerical experiments in fan-beam computed tomography and photoacoustic tomography to demonstrate that our proposed framework achieves high-quality reconstructions with bounded memory requirements, making it suitable for large-scale dynamic imaging problems.
Read the full abstract

All authors Toluwani Okunola, Mirjeta Pasha, Misha E. Kilmer, James G. Nagy, Eric de Sturler

Numerical Analysis

A penalty-free bulk–surface CutFEM stabilized by lattice Green's function extensions

We introduce a penalty-free cut finite element method for surface elliptic problems coupled to a harmonic bulk field on a Cartesian grid. Instead of adding a stabilization term, the method restricts the active finite element space by a discrete bulk harmonic extension represented with the lattice Green's function, together with a local extrapolation near the interface. The resulting method is a symmetric Galerkin scheme posed on a reduced space embedded in the standard active-mesh finite element space. Under stated geometric, regularity, and approximation assumptions, we establish optimal $\mathcal{O}(h)$ and $\mathcal{O}(h^2)$ surface convergence rates, as well as robust, cut-independent algebraic conditioning. Furthermore, a density formulation based on lattice layer potentials is shown to act as an operator preconditioner; the single-layer parametrization yields an $\mathcal{O}(1)$ algebraically well-conditioned system without introducing any tunable parameters. Two-dimensional experiments on circular, deformed, and smooth nonconvex interfaces confirm the predicted convergence and robustness under changes in the cut position. Finally, a three-dimensional torus experiment demonstrates the identical construction using a seven-point bulk stencil and trilinear surface traces, exhibiting the expected optimal error rates.
Read the full abstract

Optimization and Control

FAB: A First-Order AB-based Gradient Algorithm for Distributed Bilevel Optimization over Time-Varying Directed Graphs

Distributed optimization over time-varying directed graphs has shown promising performance in addressing challenges posed by complex communication constraints in real-world scenarios. In many practical settings, however, the direct application of distributed optimization algorithms encounters additional difficulties, most notably hyperparameter tuning, which our empirical observations suggest can be effectively mitigated by integrating bilevel optimization. Motivated by these findings, we study distributed bilevel optimization over time-varying directed networks, a problem that remains largely unexplored due to the compounded challenges arising from consensus bias in dynamic unbalanced communication and the nested optimization structure. In this work, we propose a fully first-order distributed gradient-based algorithm that integrates the Push-Pull (also known as AB) communication strategy with a value function-based penalty method and establish its non-asymptotic convergence properties. Notably, a simplified variant of our analysis framework for nonconvex single-level distributed optimization establishes a convergence rate for the Push-Pull algorithm, thereby resolving an open question concerning its convergence over time-varying directed graphs. Empirical evaluations across diverse tasks, including hyperparameter tuning, data hyper-cleaning, and reinforcement learning, validate the effectiveness and efficiency of the proposed algorithm.
Read the full abstract

All authors Yaoshuai Ma, Xiao Wang, Wei Yao, Jin Zhang

Number Theory

$\delta$-Badly approximable numbers and ubiquitously losing sets

We consider a natural filtration $\boldsymbol{\operatorname{Bad}}(\delta) \subset \boldsymbol{\operatorname{Bad}}(\delta')$ for $\delta \geq \delta'>0$ on the set of badly approximable numbers to complement the filtration of the well approximable numbers by the $\tau$-well approximable numbers. We show that the set $\boldsymbol{\operatorname{Bad}}(\delta)$ is a $(1/3, 18 \delta)$-winning set and give a lower bound on its Hausdorff dimension. We introduce the notion of $(\alpha, \beta)$-$\textit{ubiquitously losing sets}$ to the theory of Schmidt games, give an upper bound on the Hausdorff dimension of an $(\alpha, \beta)$-ubiquitously losing set that is strictly less than full Hausdorff dimension, show that $\boldsymbol{\operatorname{Bad}}(\delta)$ is a $(1/2, 18/\delta)$-ubiquitously losing set, and give an upper bound on the Hausdorff dimension of $\boldsymbol{\operatorname{Bad}}(\delta)$ that is strictly less than one. Combined with a finite intersection property and a bilipschitz transfer property, we obtain results for finite intersections of translates of $\boldsymbol{\operatorname{Bad}}(\delta)$.
Read the full abstract

Numerical Analysis · Cross-listed from cs.LG

INEUS: Iterative Neural Solver for High-Dimensional PIDEs

In this paper, we introduce INEUS, a meshfree iterative neural solver for partial integro-differential equations (PIDEs). The method replaces the explicit evaluation of nonlocal jump integrals with single-jump sampling and reformulates PIDE solving as a sequence of recursive regression problems. Like Physics-Informed Neural Networks (PINNs), INEUS learns global solutions over the entire space-time domain, yet it offers a more efficient treatment of nonlocal terms and avoids the computationally expensive differentiation of full PIDE residuals. These features make INEUS particularly well suited for high-dimensional PDEs and PIDEs. Supported by a contraction-based convergence proof for linear PIDEs, our numerical experiments show that INEUS delivers accurate and scalable solutions for various high-dimensional linear and nonlinear examples.
Read the full abstract

Optimization and Control · Cross-listed from cs.LG

PACE: Prune-And-Compress Ensemble Models

Ensemble models achieve state-of-the-art performance on prediction tasks, but usually require aggregating a large number of weak learners. This can hinder deployment, interpretability, and downstream tasks such as robustness verification. Remedies to this issue fall into two main camps: pruning, which discards redundant learners, and compression, which generates new ones from scratch. We introduce PACE, a framework that interleaves these paradigms in a two-phase strategy. First, new learners are actively generated via a theoretically grounded procedure to enhance the diversity of the initial ensemble. When no more relevant learners can be found, a second phase of pruning is performed on this enriched ensemble. During both operations, PACE allows fine control on the faithfulness to the original ensemble. Experiments show that our method outperforms prior pruning and compression methods while offering principled control of faithfulness guarantees.
Read the full abstract

All authors Fabian Akkerman, Julien Ferry, Théo Guyard, Thibaut Vidal

Classical Analysis and ODEs

Convexity of the embedding parameter sets of some analytic function spaces

In this note, we study the geometric structure of the parameter sets governing continuous embeddings between weighted Bergman-Orlicz spaces. First, for a fixed pair of growth functions, we show that the set of admissible weight exponents $(\alpha, \beta)$ is convex, provided the growth functions satisfy specific log-convexity and log-concavity conditions of the inverses. Second, we consider the dual problem where the weight exponents are fixed. We prove that the collection of growth function pairs that yield such an embedding is log-convex under a natural interpolation of their inverses. We then obtain interpolated embeddings between Bergman-Orlicz spaces.
Read the full abstract

Information Theory

Fluid Antenna Systems Enabling 6G HRLLC With Port Switching Delay

Fluid antenna systems (FAS) exploit antenna position reconfigurability to unlock massive spatial diversity within compact form factors, making them a promising enabler for 6G user terminals (UTs). However, practical port switching incurs latency and signaling overhead, which can be particularly detrimental to hyper-reliable low-latency communications (HRLLC) under finite blocklength operation. This paper investigates FASenabled HRLLC by explicitly capturing the coupled effects of spatial correlation, port switching delay, and finite blocklength coding. We derive exact closed-form expressions for the average block error rate (BLER) and average achievable rate over spatially correlated fading channels. The resulting analysis reveals a fundamental design trade-off: increasing the number of ports improves diversity but linearly reduces the effective blocklength, thereby intensifying finite-blocklength penalties. A key theoretical contribution is a rigorous proof that reliability, achievable rate, and energy efficiency are strictly unimodal in the port dimension, ensuring a unique optimal port configuration. Furthermore, we characterize an explicit switching-delay threshold that separates regimes where FAS yields net gains over fixed-position antenna (FPA) systems. Numerical results validate the analysis and show that substantial HRLLC performance gains are achievable when the switching latency remains below the derived bound.
Read the full abstract

All authors Xusheng Zhu, Kai-Kit Wong, Hao Xu, Chenguang Rao, Hyundong Shin

Geometric Topology

Geodesic simplices of pseudo-hyperbolic space

We give a cohomological interpretation of the geodesic simplices of the pseudo-hyperbolic space of signature $(p,q)$ and formulate a necessary and sufficient condition for such a simplex to have finite volume. As a corollary, we obtain that every ideal geodesic polytope in the pseudo-hyperbolic space of signature $(2,2)$ has finite volume.
Read the full abstract

Combinatorics

On Ramsey goodness of $K_{2,n}$ versus cycles

A graph $G$ is called $H$-good if $R(G,H)=(|G|-1)(\chi(H)-1)+\sigma(H)$, where $\sigma(H)$ denotes the size of the smallest color class in a $\chi(H)$-coloring of $H$. In Ramsey theory, it is an interesting problem to study whether a graph $G$ is $H$-good or not. In this article, we study the Ramsey goodness of the pair $(K_{2,n},C_m)$, which naturally lies between the classical star-cycle and book-cycle problems. We prove that \begin{equation*} R(K_{2,n},C_{\{m,m+1\}})=m+1. \end{equation*} for all $m\ge 2n+1$, and consequently establish that \begin{equation*} R(K_{2,n},C_{m})=m+1. \end{equation*} for all $m\ge 3n+4$. This proves that $C_m$ is $K_{2,n}$-good in this range and improves a particular case of a result on the Ramsey goodness by Pokrovskiy and Sudakov. Further, we provide a construction of a graph that disproves the $C_{m}$-goodness of $K_{2,n}$ for all even $m$ satisfying $n\geq m+2$.
Read the full abstract

Algebraic Geometry

An explicit formula for the Artin invariant of smooth K3 hypersurfaces

We characterize the Artin invariant of a smooth K3 hypersurface in terms of quasi-$F$-splitting. As an application, we obtain an explicit formula for this invariant.

Combinatorics

Double-sum Rogers-Ramanujan type identities

As the $q$-analog of Chebyshev polynomials, $q$-Hermite polynomials form a cornerstone in the family of $q$-orthogonal polynomials, which play a fundamental role in quantum algebra and mathematical physics. Recently, Andrews obtained a series of Rogers-Ramanujan type identities by constructing Bailey pairs from Chebyshev polynomials. In this paper, by applying the expansion formula of Chebyshev polynomials in terms of $q$-Hermite polynomials and using the orthogonality relations, we derive a series of Rogers-Ramanujan type identities on double sums, which further generalized the known results due to Andrews, Shi, Sun and Yao.
Read the full abstract

Combinatorics

A $\mu$-distance for semidirected orchard phylogenetic networks

In evolutionary biology, phylogenetic networks are now widely used to represent the historical relationships between species and population, when this history includes reticulation events such as hybridization, gene flow and admixture between populations. Semidirected phylogenetic networks are appropriate models when the direction of some edges and the root position are not identifiable from data. Comparing semidirected networks is important in many applications. For rooted and directed networks, a $\mu$-representation was originally introduced to distinguish tree-child networks, and has since been extended in two different directions: to the larger class of orchard directed networks by adding an extra component that counts paths to reticulations; and to semidirected networks, through an edge-based variant. However, the latter does not provide a distance between semidirected and orchard networks. We introduce here a new edge-based $\mu$-representation capable of distinguishing distinct orchard binary semidirected networks. For this class, we provide a reconstruction algorithm and therefore obtain a true distance that is computable in polynomial time.
Read the full abstract

Mathematical Physics · Cross-listed from hep-th

Modular wedge localization, Majorana fields and the Tsirelson limit of the Bell-CHSH inequality

The massive Majorana field in $1+1$ dimension is employed to investigate the violation of the Bell-CHSH inequality in relativistic Quantum Field Theory. We give an explicit rapidity-space realization of the Summers-Werner modular-localization construction and reduce the vacuum Bell-CHSH correlator to a single spectral weight $h^2(\omega)$ for the modular operator. The resulting analytic families approach the Tsirelson bound in the vacuum state as their spectral weight concentrates near $\omega\approx0$, corresponding to the eigenvalue $\lambda^2 \approx 1$ of the modular operator.
Read the full abstract

All authors J. G.A. Caribé, M. S. Guimaraes, I. Roditi, S. P. Sorella

Combinatorics

Asymmetric bilateral Bailey pairs and Rogers-Ramanujan type identities

The theory of Bailey's transform provides a systematic method for deriving $q$-identities, the key factor of which is the Bailey pair. The concept of Bailey pair was first extended to bilateral version by Paule. In this paper, following Rogers' work on Fourier series, we derive two asymmetric bilateral Bailey pairs. By inserting them into the bilateral Bailey chains, we obtain several identities of Rogers-Ramanujan type, Andrews-Gordon type and also identities on false theta functions. Furthermore, based on the Bailey lattice due to Dousse, Jouhet and Konan, we get an asymmetric bilateral Bailey lemma which leads to identities on Appell-Lerch series. Moreover, by using the asymmetric bilateral Bailey lemmas due to Andrews and Warnaar, we get some identities on false theta functions and the generalized Hecke-type series.
Read the full abstract

Mathematical Physics · Cross-listed from hep-th

Causal self-dual nonlinear electrodynamics from the Born-Infeld theory

Recently we have proposed a new auxiliary-field formulation for self-dual nonlinear electrodynamics (NLED) which makes use of two building blocks: (i) a seed self-dual theory $L(F_{\mu\nu};g)$, where $F_{\mu \nu}$ is the electromagnetic field strength and $g$ a duality-invariant coupling constant; and (ii) a scalar potential $W(\psi)$. Our formulation is based on the Lagrangian $ \mathfrak{L}(F_{\mu\nu};\psi) = L(F_{\mu\nu};\psi) + W(\psi)$, where $\psi$ is an auxiliary scalar field. Integrating out $\psi$, using its equation of motion, one obtains a $\mathsf{U}(1)$ duality-invariant NLED. Different self-dual NLEDs are derived by choosing different potentials $W(\psi)$. In the case that the seed Lagrangian defines the Born-Infeld theory, in this paper we demonstrate that the resulting models for self-dual NLED are causal and provide a general solution of the self-duality equation. We also elaborate on the procedure to relate our formulation to that developed by Russo and Townsend.
Read the full abstract

Information Theory

Locally Repairable Codes with Availability via Elliptic Function Fields

Locally repairable codes with availability have become essential components in modern large-scale distributed cloud storage systems and numerous other applications. In this paper, we focus on the construction of locally repairable codes with one or two recovering sets via elliptic function fields. Prior pioneering work by Li et al. (IEEE Trans. Inf. Theory, vol. 65, no. 1, 2019) and Ma and Xing (J. Comb. Theory Ser. A., vol. 193, 2023) employed maximal supersingular elliptic curves to obtain several optimal (classical) locally repairable codes. In contrast, we consider ordinary elliptic curves with many rational points. This approach yields several new families of \(q\)-ary optimal locally repairable codes with length \(O(q+2\sqrt{q})\) and flexible locality. Consequently, our work broadens the selection of curves available for the construction of optimal locally repairable codes. Furthermore, we present a general framework for constructing locally repairable codes with two recovering sets via automorphism groups of elliptic function fields. To realize this framework, we devise a novel construction for determining the functions \(e_i\) in the construction of locally repairable codes. By employing both supersingular and ordinary elliptic curves, we obtain several families of locally repairable codes with two recovering sets. In particular, we construct a family of \(q^2\)-ary locally repairable codes with two recovering sets, achieving length \(O(q^2+2q)\) and Singleton-defect \(O\!\left(\frac{2\ell}{q^2+2q-8\ell}\right)\), where \(\ell \mid\mid q + 2\) with \(4\ell < q\).
Read the full abstract

Representation Theory

A Microlocal Description of Aubert-Zelevinsky Duality on Unipotent $L$-Parameters

We give a microlocal description of the Aubert–Zelevinsky involution for all unipotent representations of all inner forms of simple adjoint unramified $p$-adic groups. Via the realization of enhanced $L$-parameters as perverse sheaves, we show that the involution corresponds to the composition of three operations on an endoscopic subgroup: Fourier transform, Chevalley involution and duality on local systems. When the group is not inner to an unramified triality form of $D_4$ we further show that one does not need to pass to an endoscopic subgroup. This was previously verified in certain special examples by several authors where only the contribution by Chevalley involution and Fourier transform was observed. Duality on local systems is invisible in those examples since only self-dual local systems appear. Motivated by categorical considerations, we provide a second formulation, involving complex conjugation from the compact form of the dual group, giving a covariant functor of perverse sheaves that agrees with Aubert–Zelevinsky duality on $L$-parameters and as involutions of graded Hecke algebras. This formulation holds without passing to an endoscopic subgroup and is valid for all inner forms of simple adjoint unramified groups. Finally, we prove the microlocal Hiraga conjecture for unipotent $A$-parameters of inner-to-split simple adjoint groups as a consequence of our results.
Read the full abstract

Optimization and Control

Stochastic Optimal Control for Jump Diffusion Models with Singular Drifts

We study a stochastic optimal control problem for jump-diffusion systems whose drift coefficient is piecewise Lipschitz continuous and exhibits threshold-induced discontinuities. Such dynamics naturally arise in applications with intervention policies triggered by safety levels, notably in insurance surplus management with dividend payments and capital injections. These features place the problem outside the scope of classical stochastic maximum principle (SMP) results, which rely on global smoothness assumptions. We establish both necessary and sufficient optimality conditions for this class of control problems. Our approach combines a Sobolev-type representation of the first variation process with smooth approximations and Ekeland's variational principle. As application, we study an optimal premium adjustment and reserve management policies for an insurance whose surplus is modelled by threshold-based dividend and capital injection policies.
Read the full abstract

All authors Edward Fuituh Kameh, Antoine-Marie Bogso, Olivier Menoukeu-Pamen, Felix Shu

Differential Geometry

Heat dispersion laws in smooth compact manifolds

Given a Lipschitz conductor $K$ in the smooth compact Riemannian $2\le n$-manifold $(M,g)$, such a half generic heat dispersion law $$ {\rm H^d}_{p,\varPhi,\varPsi}(K,M)=2^{-1} {\rm H^d}_{\Delta_p,\varPhi,\varPsi}(K,M) $$ is not only newly-established via Theorem 1.1 but also deeply-explored through not only Proposition 3.1 (a comparison law for the generic heat dispersion) but also Proposition 3.2 (a recycling law for the quasilinear Laplace-Robin eigenvalue).
Read the full abstract

Numerical Analysis · Cross-listed from stat.ML

Expressivity of Bi-Lipschitz Normalizing Flows: A Score-Based Diffusion Perspective

Many normalizing flow architectures impose regularity constraints, yet their distributional approximation properties are not fully characterized. We study the expressivity of bi-Lipschitz normalizing flows through the lens of score-based diffusion models. For the probability flow ODE of a variance-preserving diffusion, Lipschitz regularity of the score induces a flow of bi-Lipschitz diffeomorphic transport maps. This ODE bridge allows us to analyze the distributional approximation power of bi-Lipschitz normalizing flows and, conversely, derive deterministic convergence guarantees for diffusion-based transport. Our key idea is to use the probability flow ODE to link regularity of the score to regularity of the induced transport maps. We verify score regularity for broad target densities, including compactly supported densities, Gaussian convolutions of compactly supported measures and finite Gaussian mixtures. We obtain a universal distributional approximation result: Gaussian pullbacks induced by bi-Lipschitz variance-preserving transport maps are $L^1$-dense among all probability densities. For Gaussian convolution targets, we further obtain convergence in Kullback-Leibler divergence without early stopping.
Read the full abstract

Group Theory

The Normal Subgroup Theorem for lattices on two-dimensional Euclidean buildings

We prove the normal subgroup property for every group that acts properly and cocompactly on a two-dimensional Euclidean building: every normal subgroup has finite index or is contained in the finite kernel of the action. As a consequence, the non-residually finite lattices constructed by Titz Mite and the second author are virtually simple. They are the first known simple lattices on irreducible Euclidean buildings.
Read the full abstract

Algebraic Topology

On Weakly Contractible Non-Contractible Finite Topological Spaces of Ten Points

Cianci and Ottina proved that a homotopically trivial non-contractible finite $T_0$-space has at least nine points, and classified such spaces with exactly nine points. We complete the classification for ten points. The main tool is the notion of a naked pair: a relation $c<a$ between a minimal and a maximal point that does not factor through a middle point. We show that minimal homotopically trivial spaces with at most ten points have no naked pairs. This turns minimality into simple conditions on the incidences between the three levels of the space and, together with two lemmas of Cianci and Ottina and an Euler characteristic count, reduces the classification to short case analyses. No such space exists with one or two middle points; there are exactly six with three middle points, forming three types and their order-duals, and exactly four with four middle points. All ten spaces have height two, an antichain of middle points and collapsible order complexes, and the count is confirmed by an exhaustive computer enumeration. We also show that the absence of naked pairs persists up to seventeen points but not beyond, and that height three already occurs at eleven points.
Read the full abstract

Optimization and Control · Cross-listed from cs.LG

Grokking or Glitching? How Low-Precision Drives Slingshot Loss Spikes

Deep neural networks exhibit periodic loss spikes during unregularized long-term training, a phenomenon known as the "Slingshot Mechanism." Existing work usually attributes this to intrinsic optimization dynamics, but its triggering mechanism remains unclear. This paper proves that this phenomenon is a result of floating-point arithmetic precision limits. As training enters a high-confidence stage, the difference between the correct-class logit and the other logits may exceed the absorption-error threshold. Then during backpropagation, the gradient of the correct class is rounded exactly to zero, while the gradients of the incorrect classes remain nonzero. This breaks the zero-sum constraint of gradients across classes and introduces a systematic drift in the parameter update of the classifier layer. We prove that this drift forms a positive feedback loop with the feature, causing the global classifier mean and the global feature mean to grow exponentially. We call this mechanism Numerical Feature Inflation (NFI). This mechanism explains the rapid norm growth before a Slingshot spike, the subsequent reappearance of gradients, and the resulting loss spike. We further show that NFI is not equivalent to an observed loss spike: in more practical tasks, partial absorption may not produce visible spikes, but it can still break the zero-sum constraint and drive rapid growth of parameter norms. Our results reinterpret Slingshot as a numerical dynamic of finite-precision training, and provide a testable explanation for abnormal parameter growth and logit divergence in late-stage training.
Read the full abstract

All authors Hanqing Liu, Jianjun Cao, Yuanze Li, Zijian Zhou

Mathematical Physics · Cross-listed from physics.plasm-ph

Hugoniot Relation for Multi-Temperature Euler Equations of Compressible Plasma Flows

Shock solutions for multi-temperature Euler equations are inherently ambiguous due to the loss of microscopic physical detail during model reduction and occurrence of non-conservative terms. This paper presents a detailed analytical study of shock structures in such models. We derive two distinct Hugoniot relations, each corresponding to a physically admissible shock solution: one for the general multi-temperature case and one for two-temperature plasma flows. Through classical analysis à la Courant–Friedrichs, we demonstrate that both satisfy admissibility conditions, revealing a fundamental non-uniqueness in shock structures. By relating these solutions to existing numerical schemes, the structure preserving and vanishing viscosity approaches, we provide physically justified references for constructing and evaluating discontinuous numerical approximations. In particular, we emphasize that the Hugoniot relation is not uniquely determined by the macroscopic PDEs alone, but must be supplied from external sources such as experiments or first-principles simulations. This insight demonstrates the essential role of microscopic physics in resolving shock ambiguity and contributes to the theoretical foundation for modeling discontinuous plasma flows.
Read the full abstract

Optimization and Control

Convergence guarantees for stochastic algorithms solving non-unique problems in metric spaces

We prove a general quantitative theorem on the asymptotic behavior of stochastic quasi-Fejér monotone sequences in a broad metric context. Concretely, our result explicitly constructs a rate of convergence for such process, both in mean and almost surely, under an abstract stochastic regularity assumption, derived from previous work of Kohlenbach, López-Acedo and Nicolae [Isr. J. Math. 232(1), pp. 261-297, 2019] on such notions in a deterministic context. Our notion of regularity extends and unifies many common conditions from the literature, such as generalized contractivity for self maps, weak sharp minima and error bounds for real-valued functions, uniform monotonicity and global metric subregularity for set-valued operators, related Polyak-Łojasiewicz or Kurdyka-Łojasiewicz conditions, as well as expected sharp growth as e.g. studied by Asi and Duchi [SIAM J. Optim. 29(3), pp. 2257-2290, 2019]. The rate is moreover highly uniform, depending only on very few data of the surrounding objects. We also discuss special cases which allow for the construction of fast rates in the form of linear non-asymptotic guarantees. We conclude by presenting three concrete methods from stochastic approximation where our results yield new rates of convergence, including the classical example of the stochastic proximal point method, a randomized variant of the Krasnoselskii-Mann scheme for solving stochastic fixed-point equations, and a Busemann subgradient method recently introduced by Goodwin, Lewis, López-Acedo and Nicolae [Math. Program., to appear], all of which make use of our metric generality by being formulated over complete geodesic metric spaces of nonpositive curvature.
Read the full abstract

Analysis of PDEs

Uniform small energy regularity for fractional geometric problems

We prove small energy regularity for a parabolic boundary reaction Ginzburg-Landau problem in the full range $s\in (0,1)$, answering a question posed by Hyder, Segatti, Sire and Wang. We also obtain a similar small energy regularity result for fractional harmonic maps to spheres. Both results are uniform as $s\to 1$.
Read the full abstract

Operator Algebras

The Quasi-linearity problem for Jordan-Banach algebras: a topological characterization

Let $\mathfrak{J}$ be a JB$^*$-algebra with no quotients isomorphic to any spin factor. Let $\mu$ be a local quasi-linear Jordan functional on $\mathfrak{J}_{sa}$. We show that $\mu$ is a linear functional on $\mathfrak{J}_{sa}$ if and only if the restriction of $\mu$ to the closed unit ball of $\mathfrak{J}_{sa}$ is uniformly weakly continuous.
Read the full abstract