Probability · Cross-listed from cond-mat.stat-mech
Tensor-Network Population Annealing
Takumi Oshima, Yuma Ichikawa and Koji Hukushima
We propose a hybrid sampling method, tensor-network population annealing (TNPA), which combines tensor-network (TN) initialization with population annealing (PA). We apply this method to the two-dimensional Edwards-Anderson Ising
spin glass. The approach is motivated by the limitations of existing methods: TN-based samplers can become numerically unstable in frustrated
spin systems at low temperatures, whereas conventional PA requires a long annealing schedule when started from the high-temperature limit. In TNPA, TN contractions are used only within a reliable temperature range to generate initial configurations that are close to the
equilibrium distribution. The subsequent low-temperature equilibration is then carried out by PA. To stabilize the initialization process, we introduce a diagnostic based on the effective sample size that adaptively selects the initialization temperature. The proposed framework provides a practical and physically motivated route to low-temperature sampling by combining the complementary strengths of TN and PA.
Read the full abstract ▾
Optimization and Control
Proximal Nonlinear Conjugate Gradient Methods for Composite Optimization
Shodai Hamana and Yasushi Narushima
The nonlinear conjugate gradient methods are known to be an effective approach for standard unconstrained optimization problems especially for large-scale problems.This paper proposes a proximal nonlinear conjugate gradient method, which extends the nonlinear conjugate gradient methods to composite objective functions, namely, the sum of a smooth nonconvex function and a nonsmooth convex function, and its extension to the case where the nonsmooth function is weakly convex. The proposed method uses the forward-backward residual which is defined by using the proximal mapping instead of the gradient and determines the search direction based on the three-term Hestenes-Stiefel (HS) formula. We establish the global convergence of the proposed
algorithm under standard assumptions for both convex and weakly convex nonsmooth functions, and analyze its convergence rate in terms of stationarity.In addition, when the smooth term is strongly convex, we prove that the generated sequence converges to the optimal solution and establish its convergence rate. Finally, numerical experiments show that the proposed method is stable and achieves better performance than existing methods in both convex and nonconvex settings.
Read the full abstract ▾
Differential Geometry
Faces of invariant convex sets in representations of nontrivial copolarity
Yi Shi
Let $(V, G)$ be an orthogonal representation of a compact Lie group $G$ with nontrivial copolarity, and $\Sigma$ a fat section of $(V, G)$. If $E$ is a $G$-invariant compact convex set in $V$, then $P=E\cap\Sigma$ is a convex set in $\Sigma$. We prove that up to conjugacy the face structure of $E$ is completely determined by that of $P$ and that a face of $E$ is exposed if and only if the corresponding face of $P$ is exposed. Our result generalizes the result proved by Leonardo Biliotti, Alessandro Ghigi and Peter Heinzner in the case where $(V, G)$ is a polar representation.
Read the full abstract ▾
Optimization and Control
Oscillating solutions to the mean-field Langevin descent-ascent flow
Jean-Christophe Mourrat and Loucas Pillaud-Vivien
We present a counterexample to the statement of convergence of the mean-field Langevin descent-ascent flow on $\mathbb{R}^2$. We consider payoff functions that are shaped as a double well in each coordinate, and for which the deterministic dynamics admits a limit cycle. When the coupling between the two coordinates is sufficiently strong and the entropic regularization sufficiently small, we show that the mean-field dynamics remains close to this cyclic behavior, and in particular, does not converge.
Read the full abstract ▾
Analysis of PDEs
Rigidity of codimension-1 isometric immersions in complete manifolds
Mert Baştuğ
We establish an asymptotic
rigidity result for isometric immersions of codimension-1. Specifically, we consider a sequence of immersions from a compact $d$-dimensional
manifold into a complete $(d+1)$-dimensional
manifold whose elastic energies vanish asymptotically, where the elastic energy quantifies both stretching and bending. We show that such a sequence admits a subsequence converging to an isometric immersion. This extends a result of Alpern, Kupferman, and Maor to the case of complete target
manifolds, where the lack of compactness introduces additional analytical difficulties. The proof is based on an approach using local quantitative
rigidity estimates, obtained via a reduction to the Euclidean setting. This method avoids the use of Young measures and provides a flexible framework that may be of independent interest.
Read the full abstract ▾
Statistics Theory
Empirical interpretation of the Pitman efficiency
Tadeusz Inglot
We study an empirical interpretation of the Pitman efficiency in testing for uniformity in the two-parametric family of the beta distributions. We show that for contamination models the Pitman efficiency approximates relative efficiency very well.
Optimization and Control
Polyconvexity with Moments and Sums of Squares
Giovanni Fantuzzi, Didier Henrion, Martin Kružík et al. (+2 more)
A function of a
matrix is polyconvex when it can be expressed as a convex function of the
matrix minors. Polyconvexity is a regularity condition ensuring existence of minimizers in nonlinear elasticity and, more broadly, in vectorial problems of the calculus of variations, when minimizing integral gradient functionals. The polyconvex envelope of a function is the largest polyconvex lower bound. Yet deciding whether a given energy is polyconvex, or computing the polyconvex envelope, are generally difficult problems. This paper focuses on
polynomial matrix functions. We propose (i) tractable convex-optimization based sufficient conditions to certify polyconvexity via sum-of-squares (SOS) technology, and (ii) a principled numerical method to compute the polyconvex envelope pointwise, based on the moment-SOS hierarchy from
polynomial optimization.
Read the full abstract ▾
All authors Giovanni Fantuzzi, Didier Henrion, Martin Kružík, Ajay Murali, Stephan Weis
Numerical Analysis
A regularized truncated finite element method for degenerate parabolic stochastic PDE on non-compact graph
Jianbo Cui, Mihály Kovács and Derui Sheng
We study the numerical approximation of a class of degenerate
parabolic stochastic partial differential equations on non-compact metric graphs, which naturally arise in the asymptotic analysis of
Hamiltonian flows under small noise perturbations. The numerical discretization of these equations faces several challenges, including the non-compactness of the graph, the degeneracy of the differential operator near vertices, and the non-
symmetry of the associated bilinear form. To address these issues, we propose a multi-step numerical strategy combining graph truncation, localized coefficient regularization, and finite element spatial discretization. By incorporating localization techniques, tightness arguments, and resolvent estimates, we establish the strong convergence of the proposed scheme in a weighted $L^2$-space. Our results provide a systematic methodology that is potentially extensible to more general non-compact graphs and degenerate operators.
Read the full abstract ▾
Spectral Theory
Some universal inequalities for Dirichlet eigenvalues of the Laplacian on a Euclidean convex domain
Kei Funano
We establish two universal inequalities for Dirichlet
eigenvalues of the
Laplacian on a Euclidean convex domain.
Optimization and Control
Nesterov Acceleration with Operator Decomposition
Jaewook Lee, Ernest K. Ryu and Chulhee Yun
We propose Nesterov acceleration with Operator Decomposition (NOD), which extends Nesterov's accelerated gradient descent (NAG) from smooth strongly convex optimization to the broader setting of strongly monotone, Lipschitz operators. The key insight is to decompose the operator into cyclically monotone and monotone components, with the Asplund decomposition providing the tightest such representation, and to have the
algorithm utilize the decomposed oracles. NOD and its analysis subsume the classical theory of Nesterov acceleration and yield an iteration complexity for finding an $\epsilon$-accurate solution of \[ \Theta\left(\sqrt{\frac{L_{\phi}}{\mu} + \frac{L_{\mathbb{S}}^2}{\mu^2}} \,\log \frac{1}{\epsilon}\right), \] where $\mu$ is the strong monotonicity parameter, $L_\phi$ is the Lipschitz constant of the cyclically monotone component, and $L_{\mathbb{S}}$ is the Lipschitz constant of the (possibly acyclic) monotone remainder.
Read the full abstract ▾
Combinatorics
Geordie Williamson, Oded Yacobi and Paul Zinn-Justin
We present a new method for constructing Hadamard
matrices that combines
transformer neural networks with local search in the PatternBoost framework. Our approach is designed for extremely sparse combinatorial search problems and is particularly effective for Hadamard
matrices of Goethals–Seidel type, where
Fourier methods permit fast scoring and optimisation. For orders between 100 and 200, it produces large numbers of inequivalent Hadamard
matrices, and for larger orders, it succeeds where local search from random initialisation fails. The largest example found by our method has order 252. In addition to these new constructions, our experiments reveal that the
transformer can discover and exploit useful hidden
symmetry in the search space.
Read the full abstract ▾
Numerical Analysis
A discontinuous Galerkin method with fractal elements
Sergio Gómez, David Hewett and Andrea Moiola
We formulate, analyse, and implement a discontinuous Galerkin finite element method (DG-FEM) for the approximation of the solution of an
elliptic boundary value problem in a domain with fractal boundary. We consider the case of the Poisson equation in the Koch snowflake domain with zero Dirichlet boundary conditions, but our methodology can be generalised to other cases. Rather than first approximating the snowflake domain by a polygonal "prefractal" and then applying a standard DG-FEM on the prefractal, we define a DG-FEM on the snowflake itself, using a geometry-conforming mesh (a fractal tiling) consisting of fractal elements, each similar to the original snowflake. Fluxes across inter-element boundaries, which are fractal curves, are represented in a weak way by integrals over element subdomains. We show how, for local
polynomial basis functions, these integrals can be evaluated exactly using the similarity of the elements. We prove well-posedness and quasi-optimality of the method, and provide a partial convergence analysis. We present numerical results for piecewise linear and piecewise quadratic basis functions, which demonstrate the effectiveness of the method. We also apply our method to the related Dirichlet
eigenvalue problem.
Read the full abstract ▾
Analysis of PDEs
Quantitative propagation of chaos for particle systems with bounded kernels and multiplicative noise
Ning Jiang and Rongli Mo
We prove the quantitative propagation of chaos for
stochastic particle systems with interaction in both the drift and the
diffusion coefficients, provided the drift kernel is bounded and free of Lipschitz or smoothness assumptions. Our proof is based on the relative
entropy framework of Jabin and Wang \cite{JW2018}, and applies and extends their work on the exponential laws of large numbers. We extend one of their exponential laws of large numbers from the drift to the
diffusion kernel to handle the error term arising from multiplicative noise in the
entropy evolution equation. Proving this extension relies on a dynamic combinatorial analysis.
Read the full abstract ▾
Spectral Theory
Optimizing Riesz means of Robin Laplace operators on cuboids in a semiclassical limit
Matthias Baur and Simon Larson
We study asymptotic shape optimization for Riesz means of Robin
Laplacian eigenvalues among cuboids of fixed measure. Our focus is the regime where the Robin parameter is proportional to the square root of the spectral parameter defining the Riesz means. Here, a transition emerges based on the precise ratio between the two parameters: as the spectral parameter tends to infinity, sequences of maximizers shift from converging to the unit cube to lacking convergent subsequences entirely. Key tools include two-term spectral asymptotics and uniform inequalities for the Riesz means. Notably, the transition point governing the behavior of optimizers may differ from the point at which the second asymptotic term changes sign. This shows that heuristics based solely on asymptotics for a fixed domain fail to accurately predict the asymptotic behavior of maximizers.
Read the full abstract ▾
Mathematical Physics · Cross-listed from cond-mat.str-el
Parent Hamiltonian construction of generalized Calogero-Sutherland models
Hari Borutta, Andreas Feuerpfeil and Yasir Iqbal
The Calogero-Sutherland model is a paradigmatic integrable system describing one-dimensional non-relativistic particles with inverse-square-type interactions. At interaction strength $\lambda=2$, the CSM exhibits a deep connection to anyon physics, featuring the Laughlin-Jastrow
polynomial as its exact ground state. Motivated by this structure, we develop a general reverse-engineering construction of positive semidefinite continuum parent
Hamiltonians for trial states admitting a rational conformal field theory description with central charge $c<1$. By leveraging the null-vector structure of the underlying primary fields and the associated Belavin-Polyakov-Zamolodchikov equations, we derive corresponding many-body annihilation operators. We then apply this construction explicitly to the Moore-Read and $k=3$ Read-Rezayi states-relating to Ising and Fibonacci anyons, respectively-obtaining continuum
Hamiltonians for which these Jack-
polynomial states, viewed as one-dimensional
chiral spin liquid wave functions, are exact zero modes. We emphasize, however, that our construction does not by itself establish ground-state uniqueness or determine the nature of the excitation spectrum.
Read the full abstract ▾
Information Theory
Exact Outage Probability and Ergodic Capacity Analysis of NOMA in Rayleigh Fading Channels
Arafat Al-Dweik, Alok Kumar Shukla and Sami Muhaidat
This work derives the exact outage probability (OP) and ergodic capacity (EC) for the near user (NU) in the widely adopted two-user downlink non-orthogonal multiple access (NOMA) over fading channels. By noting that the noise and fading become dependent after successive interference cancellation (SIC), the exact analysis is derived by considering the joint probability density functions (PDFs) of the post-SIC noise and fading, which are typically considered to be independent and modeled using the same PDFs before the SIC. The derived exact PDFs are used to evaluate the impact of residual interference accurately. The derived interference and noise PDFs are used to derive an exact closed-form formula for NU outage and a single-integral expression for EC. Moreover, a closed-form, accurate expression is derived for the EC. Unlike existing work, the derived formulae are parameter-free, leading to more accurate performance evaluation of such systems. Monte Carlo simulation results validate the derived analysis and demonstrate that legacy Gaussian/residual-factor models can significantly misestimate outage and EC at low-to-moderate signal-to-noise ratios (SNRs) and under unbalanced power allocation. Moreover, the obtained results show that the widely considered residual interference factor, which is bounded by [0, 1], is not sufficient to capture the actual impact of residual interference due to a SIC failure, and it cannot be treated as an independent variable because it depends on the power allocation, SNR, and outage threshold. In addition to the fading-noise dependence, for two-dimensional modulations, the real and imaginary components of the noise become dependent as well.
Read the full abstract ▾
Numerical Analysis
An Adaptive Log-Laguerre Spectral Method for the Radial Dirac Equation: Resolving Asymptotic Decay and Core Singularities in Atomic Calculations
Sheng Chen, Sihong Shao and Shuai Wu
The high-precision solution of the radial Dirac equation is fundamental to relativistic
quantum chemistry, essential for reliable pseudopotential generation and all-electron electronic structure methods. Capturing both the non-
polynomial singularities at the origin and the state-dependent asymptotic decay on semi-infinite domains presents a significant computational challenge. In this work, we propose the Adaptive Log-Laguerre Spectral Method (ALLSM), a novel coupled spectral-element solver that seamlessly integrates three advanced mathematical methodologies into a unified framework. Specifically, Generalized Log-Orthogonal Functions (GLOFs) are deployed in the near-core region to intrinsically approximate complex $r^s$ singular behaviors without requiring prior knowledge of the exact analytical exponent $s$. Concurrently, an adaptive Laguerre spectral method is employed to dynamically capture diverse exponential tails on $[0, \infty)$, avoiding artificial domain truncation. To structurally guarantee spectral purity across this bipartite basis, the framework rigorously incorporates the Inverse Dirac Operator Method (IDOM), effectively eliminating variational collapse and spurious states. Validated across diverse physical regimes, including Coulomb, finite-
nucleus, and screened potentials, the proposed solver restores exponential convergence and consistently achieves relative accuracies of $10^{-10}$. This work provides a robust, pollution-free computational kernel for atomic structure calculations, establishing a highly reliable numerical standard for complex molecular simulations.
Read the full abstract ▾
Information Theory
Capacity-Region-Achieving Sparse Regression Codes for MIMO Multiple-Access Channels
Hao Yan, Lei Liu, Yuhao Liu et al. (+2 more)
This paper proposes a coding framework for capacity-region-achieving sparse regression (SR) codes over
MIMO multiple-access channels (
MIMO-MAC), where a single SR code is used for each user at the transmitter. With random semi-
unitary dictionary
matrices applied for encoding, multiple-access OAMP (MA-OAMP) enables reliable parallel interference cancellation (PIC) at the receiver. Theoretically, an optimal coding principle with the MA-OAMP receiver, which achieves the sum capacity and, in combination with time sharing, achieves the entire capacity region, is established as the guiding principle for designing capacity-region-achieving codes. Accordingly, a coding scheme for capacity-region-achieving SR codes is proposed via proper power allocation over the position-modulated signals.
Read the full abstract ▾
All authors Hao Yan, Lei Liu, Yuhao Liu, Burak Çakmak, Giuseppe Caire
Mathematical Physics · Cross-listed from quant-ph
Saikat Sur
We derive an upper bound on the maximum balanced bipartite
entanglement entropy of ground states of many-body
Hamiltonians defined on a graph, agnostic to any particular model, that possesses a nontrivial
automorphism group. We show that the
entropy is bounded by the logarithm of a weighted sum of multiplicities of irreducible representations of the bipartition-preserving
automorphism subgroup. This bound complements the known degeneracy-based bound, with neither universally dominating the other. For the complete graph $K_n$, the new bound yields an exponential improvement from linear to logarithmic scaling in the system size, consistent with the exact value of the
entropy.
Read the full abstract ▾
Information Theory
Robust Rate-Splitting Design for Mixed Dual-Polarized Integrated Satellite-Terrestrial Networks Under Polarization Mismatch
Jaehyup Seong, Juhwan Lee, Jungwoo Lee et al. (+2 more)
Dual-polarized transmission offers a promising approach to improve spectral efficiency in multiantenna networks by reusing frequency and time resources across orthogonal polarization domains. Building upon this advantage, this paper investigates interference management in mixed dual-polarized integrated satellite-terrestrial networks (MDP-ISTN), comprising a circularly polarized (CP) satellite sub-network and a linearly polarized (LP) terrestrial sub-network. To this end, we employ rate-splitting multiple access (RSMA), which enables flexible non-orthogonal transmission through partial interference decoding and partial interference treating-as-noise. Specifically, to jointly mitigate both inter-network interference between the CP low Earth orbit (LEO) satellite and LP terrestrial sub-networks as well as intra-network interference within each sub-network, we propose an MDP-RSMA framework that incorporates inter-network rate-splitting (RS) with a super-common message together with intra-network RS. Moreover, we account for practical challenges in MDP-ISTN, including polarization mismatch, channel depolarization, and imperfect channel state information at the transmitter. To maximize the minimum user rate among all satellite and terrestrial users, we formulate a robust precoder optimization problem and develop a weighted minimum mean square error (WMMSE)-based
algorithm tailored to the proposed MDP-RSMA. Numerical results demonstrate that the proposed scheme significantly improves the minimum user rate over several baseline schemes across diverse MDP-ISTN scenarios.
Read the full abstract ▾
All authors Jaehyup Seong, Juhwan Lee, Jungwoo Lee, Sean Kwon, Wonjae Shin
Representation Theory
Algorithms on the Pyasetskii involution on local Langlands parameters of classical groups
Alexander Hazeltine and Chi-Heng Lo
We give an
algorithm to compute the Pyasetskii involution for $\mathrm{Sp}_{2n}$, $\mathrm{SO}_{2n+1}$ and $\mathrm{O}_{2n}$. The
algorithm is a combination of Moeglin-Waldspurger's
algorithm for the Pyasetskii involution for $\mathrm{GL}_n$ ([MW86]) and Lanard-M${\'i}$nguez's
algorithm for the Aubert-Zelevinsky involution of bad parity representations for classical groups ([LM25]). In particular, we give a
geometric interpretation of the bad parity case of Lanard-M${\'i}$nguez's
algorithm.
Read the full abstract ▾
Optimization and Control
A Spectral-based ISS small-gain theorem for boundary control systems with infinite couplings
Yassine El Gantouh, Jun Zheng, Guchuan Zhu et al. (+1 more)
We study the input-to-state stability (ISS) of boundary control systems allowing for infinitely many boundary couplings. Using
semigroup perturbation theory and the theory of positive linear operators on
Banach lattices, we derive a spectral small-gain condition ensuring exponential ISS. We further investigate linear Boltzmann-type equations on an infinite network of intersecting circles, incorporating delays, scattering, and disturbances acting at the junction. For this class of systems, we prove that a spectral small-gain condition on the transmission operator
matrix guarantees exponential ISS with respect to disturbances propagating through the network. Moreover, we derive explicit ISS estimates for {certain} classes of dynamical processes. Finally, we demonstrate the practical applicability of our results by considering two important classes of time-delayed transmission conditions.
Read the full abstract ▾
All authors Yassine El Gantouh, Jun Zheng, Guchuan Zhu, Dingshi Li
Combinatorics
On two conjectures for generalized off-diagonal Schur numbers
Yanyan Song and Yaping Mao
For an integer $t \geq 3$, let $\mathcal{L}(t)$ denote the linear equation $x_1 + x_2 + \cdots + x_{t-1} = x_t,$ where all variables are positive integers. For integers $k \geq 1$ and $t_0,t_1,\dots,t_{k-1} \geq 3$, the generalized Schur number $S(k;t_0,t_1,\dots,t_{k-1})$ is the least positive integer $N$ such that every $k$-coloring of $[1,N]$, for some $i \in \{0,1,\dots,k-1\}$, a solution to $\mathcal{L}(t_i)$ with all variables monochromatic in color $i$. In 2015, Ahmed and Schaal proposed a
conjecture: $S(3 ; 3, t, u)>3 t u-t u-u-1$ for $3=t<u$ and $3<t \leq u$. In this paper, we confirm this
conjecture. At the same paper, they also
conjecture that $S(3 ; s, t, u)=s t u-t u-u-1$ for $4 \leq s \leq t \leq u$. Motivated by the second
conjecture, we give a recursive lower bound of $S(r; k_0, k_1, \dots, k_{r-1})$ and upper bounds for $S(r; k_0, k_1, \dots, k_{r-1})$ and $S(r;k_0,\dots,k_{r-2},u)$ for all sufficiently large $u$.
Read the full abstract ▾
Representation Theory
Geometrization of the Schrödinger Model for the Minimal Representation of an Even Orthogonal Group: The de Rham Setting
Aaron Slipper
We construct and compare three $D$-module models for the minimal representation of the conformal group of an even-dimensional quadratic space. Let $V$ be a quadratic space over a field $\kappa$ of characteristic $0$, $C\subset V^*$ be the isotropic cone, $\Delta\in D_V$ be the associated Laplace–Beltrami operator, $G$ be the conformal group of $V$, and $D_C$ be the
algebra of Grothendieck differential operators on $C$. We prove that the category of finitely generated $D_C$-modules is equivalent both to a Kazhdan–Laumon glued category attached to the smooth locus $C^o$ and to a category of “harmonic” twisted $D$-modules on the projective conformal compactification $G/P \supset V$. The gluing is governed by the quadric
Fourier transform, while the harmonic model is built from a distinguished $G$-equivariant sheaf $H$ on $G/P$ extending the local quotient $D_V/D_V\Delta$. We prove a new
geometric interpretation of higher
symmetries of the
Laplacian as global sections of $H$, and use this connection to give a
geometric proof of the
theorem of Levasseur, Smith, and Stafford that $D_C$ is Noetherian despite the singularity of $C$. We also study, via a descent procedure we call “$F$-moment descent,” the algebraic geometry of the closure of the minimal nilpotent orbit of $G$, which is the quasiclassical analogue of the minimal representation. Finally, we analyze the filtered structure of $D_V/D_V\Delta$ as a right $D_C$-module, identifying its associated graded layers through a flat degeneration of an affine flag multicone whose special fiber is the Rees space of a natural ideal in $\kappa[\overline{O}_{min}]$.
Read the full abstract ▾
Mathematical Physics · Cross-listed from gr-qc
Open-channel quantization of JT gravity at finite cutoff
Ye Zhou
Direct canonical quantization of Jackiw–Teitelboim
gravity at finite cutoff leads to a critically singular Pöschl–Teller problem whose endpoint admits inequivalent self-adjoint domains. We solve the full one-parameter family, obtain its exact spectral data, and identify the extension parameter with a renormalized short-distance matching datum in geodesic-length space. Self-adjointness, positivity, and
unitary Brown–York
evolution leave a stable family, whereas a source-free local completion without an endpoint contact term selects the Friedrichs realization. With the Brown–York clock and a local cap prescription fixed, the disk amplitude distinguishes the domains. In the large-boundary limit the bulk operator becomes universally Liouville, while a cap tied to the receding endpoint can retain domain dependence. Analytic continuation also maps radial resonances to poles of the Brown–York resolvent. These results isolate the short-distance datum that distinguishes the reduced canonical realizations and that a microscopic completion would have to determine.
Read the full abstract ▾
Commutative Algebra
Infinitely many associated primes of local cohomology modules of ramified regular local rings
Linquan Ma
We construct examples of local
cohomology modules of ramified regular local rings with infinitely many associated
primes and infinite Bass numbers.
Optimization and Control · Cross-listed from cs.LG
Continuous-time Online Learning via Mean-Field Neural Networks: Regret Analysis in Diffusion Environments
Erhan Bayraktar, Bingyan Han and Ziqing Zhang
We study continuous-time online learning where data are generated by a
diffusion process with unknown coefficients. The learner employs a two-layer neural network, continuously updating its parameters in a non-anticipative manner. The mean-field limit of the learning dynamics corresponds to a
stochastic Wasserstein gradient flow adapted to the data filtration. We establish regret bounds for both the mean-field limit and finite-particle system. Our analysis leverages the logarithmic Sobolev inequality, Polyak-Lojasiewicz condition, Malliavin calculus, and uniform-in-time propagation of chaos. Under displacement convexity, we obtain a constant static regret bound. In the general non-convex setting, we derive explicit linear regret bounds characterizing the effects of data variation, entropic exploration, and quadratic regularization. Finally, our simulations demonstrate the outperformance of the online approach and the impact of network width and regularization parameters.
Read the full abstract ▾
Optimization and Control · Cross-listed from cs.LG
Learning to Adapt: In-Context Learning Beyond Stationarity
Zhen Qin, Jiachen Jiang and Zhihui Zhu
Transformer models have become foundational across a wide range of scientific and engineering domains due to their strong empirical performance. A key capability underlying their success is in-context learning (ICL): when presented with a short
prompt from an unseen task,
transformers can perform per-
token and next-
token predictions without any parameter updates. Recent theoretical efforts have begun to uncover the mechanisms behind this phenomenon, particularly in supervised regression settings. However, these analyses predominantly assume stationary task distributions, which overlook a broad class of real-world scenarios where the target function varies over time. In this work, we bridge this gap by providing a theoretical analysis of ICL under non-stationary regression problems. We study how the gated linear
attention (GLA) mechanism adapts to evolving input-output relationships and rigorously characterize its advantages over standard linear
attention in this dynamic setting. To model non-stationarity, we adopt a first-order autoregressive process and show that GLA achieves lower training and testing errors by adaptively modulating the influence of past inputs – effectively implementing a learnable recency bias. Our theoretical findings are further supported by empirical results, which validate the benefits of gating mechanisms in non-stationary ICL tasks.
Read the full abstract ▾
Geometric Topology
On sections of Lefschetz fibrations and bundles over 2-complexes
Jonathan A. Hillman and Riccardo Pedrotti
We address the question of existence of sections of fibrations in two settings. First, we show that a bundle with base a finite 2-complex admits a section if and only if the inclusion of the fiber is $\pi_1$-injective and the associated short exact sequence of fundamental groups splits. Second, for Lefschetz fibrations over the disk we provide a complete algebraic criterion characterizing which loops in the boundary mapping
torus extend to continuous or smooth sections over the disk. Finally, we apply our results to achiral Lefschetz fibrations over the sphere obtained by doubling along the vertical boundary, and give a criterion ensuring the existence of at least two homologically distinct sections.
Read the full abstract ▾
Numerical Analysis
Convergence Analysis of the Hessian Discretisation Method for Fourth Order Semi-linear Elliptic Equations with General Source
Devika Shylaja
This paper presents a convergence analysis for the Hessian Discretisation Method (HDM) applied to fourth-order semilinear
elliptic equations involving a trilinear nonlinearity and general source, based on two complementary approaches. The HDM serves as a unified framework for the convergence analysis of various numerical schemes, including conforming and nonconforming finite element methods (ncFEMs) and gradient recovery (GR) based methods. Error estimates for the Adini ncFEM and GR methods are derived for the first time, which provide an explicit order of convergence. The analysis relies on four key HDM properties along with a suitable companion operator to establish convergence results. Moreover, a convergence analysis is developed within the HDM framework, which does not require additional regularity assumptions on the exact solution or the assumption that the exact solution is regular. The paper further discusses two significant applications: the Navier–Stokes equations in stream function–vorticity formulation and the von Kármán equations for plate bending. Numerical experiments are provided to demonstrate the performance of the GR method, Morley, and Adini ncFEMs.
Read the full abstract ▾