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)

Friday 8 May

Algebraic Geometry

Equisingularity in families of double point curves

In this paper, we provide a systematic comparison between the equisingularity of a 1-parameter unfolding $F = (f_t, t)$ of a finitely determined map germ $f: (\mathbb{C}^2, 0) \to (\mathbb{C}^3, 0)$ and the equisingularity of its associated families of double point curves: $D(F)$, $F(D(F))$, $D^2(F)$, and $D^2(F)/S_2$. We also construct explicit counterexamples to several natural questions concerning the equisingularity of these loci. As a key application, we introduce new families of complete intersection curves – referred to as Henry–type families – which are topologically trivial but fail to satisfy the Whitney equisingularity conditions. Finally, we generalize classical formulas for double point curves, originally established for map germs from $(\mathbb{C}^2, 0)$ to $(\mathbb{C}^3, 0)$, to the higher-dimensional setting of map germs from $(\mathbb{C}^n, 0)$ to $(\mathbb{C}^{2n-1}, 0)$ for $n \geq 3$, thereby endowing the associated curves with a convenient analytic structure.
Read the full abstract

Probability · Cross-listed from cs.LG

Scaling Limits of Long-Context Transformers

We study the long-context limit of softmax self-attention with a fixed query and a random context of $n$ i.i.d. keys on the sphere with the intrinsic dimension of the key distribution held fixed, viewing the inverse temperature $\beta_n$ as the scaling parameter that decides whether attention degenerates into uniform averaging or collapses onto the single closest key. We show that the critical scale at which selectivity emerges is determined by the local exponent of the distance-to-query distribution near zero rather than by global features of the context, and scales like $\beta_n^\ast \asymp n^{2/(d-1)}$ for uniform keys on $\mathbb{S}^{d-1}$. Furthermore, we characterize the limiting laws of the ordered attention weights and of the attention output across all regimes of $\beta_n$: a subcritical regime in which the output reduces to a local average around $q$ with explicit deterministic bias and Gaussian fluctuations; a critical regime in which a finite collection of nearest keys retains macroscopic mass without single-key collapse; and a supercritical regime in which all mass concentrates on the closest key. Of notable interest is the subcritical case with identity value matrix where the attention map approximately implements a backward heat equation.
Read the full abstract

All authors Giuseppe Bruno, Shi Chen, Zhengjiang Lin, Yury Polyanskiy, Philippe Rigollet

Information Theory

Non-binary LDPC codes for Data Storage

In modern data storage systems, non-binary LDPC codes for recovering from disk failures are increasingly considered strong competitors to MDS codes such as Reed-Solomon codes. Since disk failures can be modeled as erasures, we analyze non-binary LDPC codes over a $q$-ary field in the $q$-ary erasure channel, relative to MDS codes. Our focus is on non-binary LDPC codes whose parity-check matrix is obtained by replacing the non-zero entries of a binary base matrix by elements of a $q$-ary finite field. For such LDPC codes, we introduce the notion of ultimate distance, which upper-bounds their minimum distance. We derive a random-coding bound on the number of non-correctable erasure patterns for the Gallager ensemble of regular non-binary LDPC codes under maximum-likelihood decoding. An algorithm for finding the ultimate distance is presented. A low-complexity algorithm for searching for the minimum distance of the non-binary LDPC code is proposed. Finally, we construct examples of non-binary LDPC codes achieving the ultimate distance.
Read the full abstract

All authors Irina Bocharova, Boris Kudryashov, Henk D.L. Hollmann, Vitaly Skachek

Information Theory

On Reducing Decoding Complexity of Successive-Cancellation List Flip Decoding of Polar Codes

The recently proposed SCLF decoding algorithm for polar codes improves the error-correcting performance of state-of-the-art SCL decoding. However, it comes at the cost of a higher complexity. In this paper, partitioned polar codes tailored for the proposed PSCLF decoding algorithm are used to reduce the complexity of SCLF. Indeed, compared to SCLF, PSCLF allows early termination and is able to restart by skipping part of the decoding tree traversed sequentially. In order to maximize the coding gain, design of partitions tailored to PSCLF is proposed. In this extended paper, dynamic flip metric is used, as well as the possibility to flip multiple times during SCL. An analysis on the impact of this strategy on the early-termination or the CRC collisions encountered in PSCLF is carried out. Error-correction performance of multiple code rates and multiple partition strategies are shown. With the baseline algorithm SCL with $L=2$, degradation of $0.05$ dB is shown with respect to SCL-64, using $\omega=3$ flip per trial with $T_{max}=300$ trials. Numerical results show that the proposed PSCLF algorithm has an error-correction performance gain of up to 0.1 dB with respect to SCLF with same decoding parameters. This work is also compared with existing techniques to reduce the complexity of the SCLF decoding algorithm. The proposed algorithm reduces the complexity up to 77 % at the frame-error rate of $0.01$ with respect to SCLF and is able to reduce more the decoding complexity of SCLF embedding as well a restart mechanism. The average execution time of PSCLF matches the latency of SCL at $\text{FER}=4\cdot10^{-3}$ and lower.
Read the full abstract

Optimization and Control

Local Nonconvexity Indices for \(C^{1,1}\) Functions via Generalized Hessians

Davydov, Moldavskaya, and Zitikis introduced local indices for quantifying the lack of convexity of a \(C^2\) function by measuring the nuclear-norm distance of its Hessian from the cone of positive semidefinite matrices. This paper develops a local analogue for functions of class \(C^{1,1}\). At a point \(x\), the classical Hessian is replaced by the Clarke-type generalized Hessian set \(\Hess(h;x)\), defined as the closed convex hull of limiting Hessians at nearby twice differentiability points. Evaluating the same spectral functional over \(\Hess(h;x)\) gives an interval-valued local nonconvexity index whose lower and upper endpoints represent, respectively, the least and greatest visible second-order nonconvexity at \(x\). The construction reduces to the original smooth index when \(h\in C^2\), vanishes for convex \(C^{1,1}\) functions, is invariant under orthogonal changes of variables, satisfies a subadditivity inequality for the upper endpoint under sums, and is upper semicontinuous in its upper endpoint. We also relate the upper endpoint to a pointwise weak-convexity curvature modulus and give explicit \(C^{1,1}\setminus C^2\) examples. The paper is deliberately local in scope: it proposes a scalar diagnostic extracted from generalized Hessian sets, not a replacement for the richer second-order variational theory of nonsmooth convexity.
Read the full abstract

Optimization and Control

A Unified Lyapunov-IQC Framework for Uniform Stability of Smooth Quadratic First-Order Accelerated Optimizers

We develop a unified Lyapunov-integral quadratic constraint (IQC) framework for establishing uniform stability of first-order accelerated optimization algorithms in the $\beta$-smooth and $\gamma$-strongly convex regime. Classical analyses of uniform stability, such as the work of Hardt, Recht, and Singer for stochastic gradient descent (SGD), rely on direct coupling arguments and case-by-case control of iterate differences under random sampling. Extending such arguments to accelerated methods, such as Nesterov Accelerated Gradient (NAG), is complicated by the presence of higher-order state dynamics induced by momentum. We first extend this classical approach with the use of Lyapunov functions to provide a uniform stability bound for smooth quadratic NAG, and supplement this result with small-scale numerical experiments. We then extend this framework by modeling first-order accelerated optimizers as Lur'e-type feedback interconnections between a linear dynamical system and a (non-linear) gradient operator. $\beta$-Smoothness and $\gamma$-strong convexity are encoded a sector IQC inequality. Under this representation, uniform stability is certified via the existence of a quadratic Lyapunov function satisfying a finite-dimensional linear matrix inequality (LMI) in the form of a feasibility problem, which can be solved via semi-definite programming (SDP). We instantiate this framework for NAG and show how classical uniform stability bounds can be recovered via this framework. These results underscore a structural connection between optimization dynamics and robust control theory, providing a modular methodology for reliable and reproducible numerical certification of uniform stability and generalization behavior of first-order methods via convex optimization tools that is adaptable to increasingly complex optimization algorithms.
Read the full abstract

Differential Geometry

Minimal surfaces with closed curvature lines

We investigate complete non-orientable minimal surfaces of finite total curvature in $\mathbb{R}^3$ such that their ends are foliated by closed lines of curvature. This condition on the ends is necessary if they have a piece inside some Euclidean ball that is free boundary. It turns out this is a rigid situation, and we are able to show, among further obstructions, that there are no such surfaces with one end.
Read the full abstract

Statistics Theory · Cross-listed from stat.ML

Sinkhorn Treatment Effects: A Causal Optimal Transport Measure

We introduce the Sinkhorn treatment effect, an entropic optimal transport measure of divergence between counterfactual distributions. Unlike classical quantities such as the average treatment effect, this measure captures differences across entire distributions. We analyze this divergence as a statistical functional and show it can be written as a smooth transformation of counterfactual mean embeddings with an appropriate kernel. This characterization allows us to establish first-order pathwise differentiability in general, and second-order pathwise differentiability under the null hypothesis of equal counterfactual distributions. Leveraging this smoothness, we construct debiased estimators and use them to obtain asymptotically valid tests for distributional treatment effects with a fixed entropic regularization parameter. Because the power of the test depends on this unknown parameter, we further propose an aggregated test that combines evidence across a grid of regularization choices. Experiments on simulated and image data demonstrate the practical advantages of our estimator and testing procedure.
Read the full abstract

History and Overview of Mathematics

Learning from Ramanujan: Elementary Approaches to Profound Ideas

We revisit several entries from Ramanujan's notebooks which follow from more elementary arguments than a first glance may suggest. Our goal is to demystify these results through more accessible proofs, while also shining some light on the web of interconnections within the notebooks and demonstrating the continuing relevance of Ramanujan's methods. Classical and modern tools, such as multisection, telescoping sums, partial fraction decomposition and Fourier analysis, are employed to reprove and extend identities originally presented without explanation. These contributions try not only to enrich our understanding of Ramanujan's intuition but also to offer new avenues for exploration in number theory, special functions and mathematical analysis.
Read the full abstract

Numerical Analysis

Randomized quasi-Monte Carlo for walk on spheres

We investigate the use of randomized quasi-Monte Carlo (RQMC) in walk on spheres algorithms to solve boundary value problems for functions with Dirichlet boundary conditions in $\mathbb{R}^d$. For harmonic functions with $d=2$, the integrands of interest are periodic indicator functions over regions $\Theta$ in the torus $\mathbb{T}^k$. We give conditions for $\partial\Theta$ to have $k-1$ dimensional Minkowski content which allows us to use results of He and Wang (2015). The RQMC estimates involve multiple values of $k$. We see sampling variances decreasing with the number $n$ of sample points at slightly better than Monte Carlo rates. The median variance rate in $4$ RQMC methods over $5$ worked examples, including some with $d=3$ and some with nonzero source functions, was slightly better than $O(n^{-1.1})$. The variance reduction factors ranged from $1.8$ to $10.7$ at $n=2^{17}$. None of the four RQMC methods dominated the others. Changes: cite Liu (2025), describe the lattices that were used, describe usage of AI
Read the full abstract

Mathematical Physics

A mathematical study of periodic band inversion

We give a mathematical analysis of the periodic band inversion phenomenon observed by Tan–Devakul for an electron in a two-dimensional periodic potential coupled to a circularly polarized photon cavity mode. In the strong-coupling limit, we derive an effective Bloch Hamiltonian and prove convergence of the low-lying bands. For a cosine potential, we explain the periodic closing and reopening of the first spectral gap, prove the existence and generic persistence of Dirac cones at the gap-closing points, and compute the Chern numbers associated to isolated band clusters. We also show that higher isolated band clusters cannot persist in the small-coupling regime. Finally, we resolve an apparent sign discrepancy between Berry curvature computations and Chern numbers by tracking the descent from the covering space to the Brillouin torus.
Read the full abstract

Numerical Analysis · Cross-listed from cs.LG

Transformers Can Implement Preconditioned Richardson Iteration for In-Context Gaussian Kernel Regression

In this paper, we study in-context kernel ridge regression (KRR) with Gaussian kernels and show, both theoretically and empirically, that a standard softmax-attention transformer can approximate the KRR predictor during its forward pass. Under bounded-data assumptions, we construct a single-head transformer whose forward pass approximately implements preconditioned Richardson iteration on the associated kernel system. The construction uses $O(\log(1/\epsilon))$ blocks and MLP width $O(\sqrt{N/\epsilon})$ to achieve $\epsilon$-accurate prediction for prompts of length $N$. Our construction reveals a functional decomposition within the transformer architecture: softmax attention produces a row-normalized Gaussian-kernel operator needed for cross-token interactions, while MLP layers act locally to approximate the intra-token scalar arithmetic required by the update. Empirically, we train GPT-2-style transformers on Gaussian-process regression tasks and observe that they progressively align with the exact Gaussian KRR estimator across depth in terms of both prediction error and induced weights, with ablations further supporting this trend. Comparisons with classical KRR solvers also show that deeper layers align with later solver iterates. Together, we empirically demonstrate that the pretrained transformers exhibit progressive refinement toward exact Gaussian KRR across depth, and theoretically establish inexact preconditioned Richardson iteration as a concrete mechanism for approximating this predictor within an explicitly constructed softmax-attention transformer.
Read the full abstract

All authors Mingsong Yan, Dongyang Li, Charles Kulick, Sui Tang

Optimization and Control

Robust Learning Meets Quasar-Convex Optimization: Inexact High-Order Proximal-Point Methods

Robust learning aims to maintain model performance under noise, corruption, and distributional shifts, which are prevalent in modern machine learning applications. This work shows that examples of robust learning problems can be formulated as (strongly) quasar-convex optimization problems, which admit a benign landscape with no saddle points. We then propose HiPPA, an inexact high-order proximal-point method that employs a model-value gap to control the inexactness of subproblem solutions. Notably, we prove global convergence of HiPPA to global minima and establish that it attains a (local) linear or superlinear convergence rate, depending on the regularization order and inexactness control. Our numerical experiments on robust feature-alignment distillation indicate strong empirical performance of HiPPA and results consistent with our theoretical findings.
Read the full abstract

Functional Analysis

Characterization of weights for the variable fractional maximal operator and weighted inequalities for variable fractional rough operators

We characterize the class of weights related to the boundedness of variable fractional maximal operator $M_{\beta(\cdot),r(\cdot)}$ on variable Lebesgue spaces. This extend previously known results, including those corresponding to the fractional operator $M_{\beta(\cdot),1}$. In addition, we introduce a class of kernels $K$ satisfying a new variable Hörmander-type condition $H_{\beta(\cdot),r(\cdot)}$. For the fractional operator $T_{\beta(\cdot)}$ given by a kernel in $H_{\beta(\cdot),r(\cdot)}$, we prove a Coifman-Fefferman inequality and weighted inequalities in variable Lebesgue space. Finally, we provide examples of kernels in this variable Hörmander class.
Read the full abstract

Statistics Theory

Asymptotics for likelihood ratio tests of boundary points with singular information and unidentifiable nuisance parameters

We establish the asymptotic distribution of likelihood ratio tests when nuisance parameters are unidentifiable under the null hypothesis, parameters of interest lie on the boundary, and the information matrix of the identifiable parameters may be singular. Mixture models, change point models and genetic linkage analysis exhibit all three features simultaneously. Under the null hypothesis the limit is the supremum of a chi-bar-squared process, whose marginal distributions are mixtures of chi-squared distributions with weights depending on the nuisance parameter; under local alternatives it is the supremum of a noncentral chi-bar-squared process. Unlike prior work on singular information, where singularity stems from the parameter of interest and changes the limit's form, here it occurs only at isolated values of the nuisance parameter, and under additional conditions the limit keeps its nonsingular form. Classical boundary results and application-specific results for mixtures, change points and linkage analysis are recovered and extended.
Read the full abstract

All authors Karl Oskar Ekvall, Ola Hössjer, Matteo Bottai, J. M. Patrik Albin

Operator Algebras

Quantum hypergroups arising from ergodic coactions

Given a compact quantum group $\mathbb{G}$ and an ergodic action $L^\infty(\mathbb{X})\stackrel{\alpha}\curvearrowleft \mathbb{G}$ with algebraic core $\mathcal{O}(\mathbb{X})$, we show that the unital $*$-algebra $\mathcal{O}(\mathbb{X}\times_{\mathbb{G}} \bar{\mathbb{X}}):= \mathcal{O}(\mathbb{X})\square\overline{\mathcal{O}(\bar{\mathbb{X}})}$ carries the structure of an algebraic compact quantum hypergroup. This $*$-algebra admits two (generally distinct) $C^*$-algebra completions (`reduced' and `universal'), both carrying the structure of a $C^*$-algebraic compact quantum hypergroup. This provides a large class of new examples of (analytical) compact quantum hypergroups. We provide characterizations of coamenability for these compact quantum hypergroups, making use of the theory of equivariant correspondences.
Read the full abstract

Geometric Topology

Basepoints in Khovanov homology and nonorientable surfaces

We enhance the Khovanov TQFT using basepoint actions, over the field with two elements. Our enhanced Khovanov TQFT behaves similarly to gauge/Floer theoretic invariants of the double branched cover with opposite orientation: they both are invariant, in a certain sense, under taking the connected sum with the standard $\mathbb{RP}^{2}$ with Euler number -2, and they both vanish after taking the connected sum with the standard $\mathbb{RP}^{2}$ with Euler number 2. This invariance property answers a version of a question posed by Lipshitz and Sarkar. Furthermore, our construction establishes, as a special case, functoriality for the pointed Khovanov homology defined by Baldwin, Levine, and Sarkar.
Read the full abstract

Combinatorics

A short proof of Mathar's 2013 recurrence conjecture for the Laguerre sequence~A025166

For the OEIS sequence A025166, defined by $a(n) = -n!\,2^{n}\,L_{n}(1/2)$ where $L_{n}$ is the Laguerre polynomial of degree $n$, R.~J.~Mathar contributed in February 2013 the conjectured order-2 P-recursive recurrence \[ a(n) + (-4n+3)\, a(n-1) + 4(n-1)^{2}\, a(n-2) \;=\; 0, \qquad n \ge 2. \] We give a one-page proof. The exponential generating function $F(x) = -\exp\!\big(-x/(1-2x)\big)/(1-2x)$ satisfies the first-order linear ODE $(1-2x)^{2} F'(x) = (1-4x)\, F(x)$, and Mathar's recurrence then falls out by reading off the coefficient of $x^{n}/n!$. Both steps are short. The supplementary archive includes a SymPy script which checks the ODE identically and the recurrence numerically up to $n = 5000$.
Read the full abstract

Optimization and Control

Generalized Global Self-Optimizing Control for Chemical Processes: Part II Objective-Guided Controlled Variable Learning Approach

Self-optimizing control (SOC) aims to maintain near-optimal process operation by judiciously selecting controlled variables (CVs). In this series of work, the generalized global SOC (g2SOC) approach is proposed, which extends the concept of SOC to the whole operation space and uses general nonlinear functions to design CVs instead of linear combinations. In the first part of this series work, two numerical approaches for g2SOC are proposed: the optimization-based approach and the regression-based approach, based on a theoretical analysis of the existence of perfect self-optimizing CVs. The CVs designed by the former perform better, but are usually infeasible for large-scale problems. In this paper, we propose an algorithm called objective-guided controlled variable learning (OGCVL) that combines the advantages of both and has a better scalability. OGCVL is proposed for efficient CV design that seamlessly integrates symbolic and numerical computation techniques. Finally, the effectiveness of the OGCVL method is verified in two numerical examples. Both examples illustrate show that the OGCVL method is able to achieve good results while maintaining computational efficiency and is also feasible in large-scale problems.
Read the full abstract

All authors Chenchen Zhou, Hongxin Su, Xinhui Tang, Yi Cao, Shuang-Hua Yang, Lingjian Ye

Optimization and Control · Cross-listed from cs.LG

Generalized Wasserstein Flow Matching: Nested Couplings for Point Cloud Generation

We formulate flow matching on probability measures over probability measures. Directly from the Wasserstein-on-Wasserstein (WoW) continuity equation, we show that admissible nested outer and inner couplings induce a valid metameasure flow and non-local velocity field. Validity requires neither optimal plans nor a smooth manifold. Practically, this opens up the design space of multisample flow matching: exact Wasserstein plans are distinguished by energy and straightness, while sliced and lazy linear plans yield scalable alternatives. Point-cloud experiments, including controlled Bures-Wasserstein and Euclidean minibatch baselines, demonstrate the resulting geometry-cost trade-off.
Read the full abstract

2605.08424 · 8 May · Keywords: flow matching, Wasserstein over Wasserstein, hierarchical optimal transport, multisample flow matching, minibatch optimal transport, sliced Wasserstein, Bures-Wasserstein, point cloud generation, generative models for distributions, measure-valued data, Wasserstein gradient flow, rectified flow, straight flows

Optimization and Control

Transcription-Induced Failure Modes in 6-DOF Rocket Landing Trajectory Optimization

Solving optimal control problems via large-scale NLP solvers depends on discretizing continuous dynamics. Yet, this transcription step hides critical vulnerabilities-most notably truncation error and invariant drift-that can drive solvers toward dynamically infeasible or suboptimal trajectories. To expose these hidden failures, we introduce a problem- and transcription-agnostic adversarial objective that leverages the structure of local truncation-error bounds to aggressively amplify such defects. When applied to a 6-DOF rocket-landing problem, we reveal a stark reliability gap: of fourteen transcription methods tested, only three satisfy rigorous validation criteria. These results also expose a striking performance inversion: even in the absence of classical stiffness, a fourth-order implicit scheme (GL2) matches the fidelity of a sixth-order explicit method (RK6). Using B-series expansions and symplectic Runge-Kutta theorems, we isolate the specific truncation errors and quaternion-invariant drift responsible for these failures. Crucially, these theoretical vulnerabilities dictate operational performance: in practical lateral-divert scenarios, the implicit GL2 consistently outperforms the explicit RK6 in both end-to-end solve speed and robustness.
Read the full abstract

All authors Prayag Sharma, Jonathan Y.M. Goh, Behçet Aç{\i}kmeşe, Franck Djeumou

Optimization and Control · Cross-listed from cs.LG

Central Limit Theorem for Two-Time-Scale Approximate Distributionally Robust RL

Designing model-free algorithms for distributionally robust reinforcement learning (DRRL) poses fundamental challenges. The robust Bellman operator is nonlinear in the transition kernel, which makes one-sample Bellman updates biased, while the adversarial optimization underlying robustness makes robust evaluation computationally demanding. To address these difficulties, we consider the natural small-ambiguity regime under Kullback–Leibler ambiguity sets and propose an approximate DRRL framework based on a first-order expansion of the relevant robust functional. This yields an approximate robust Bellman equation that removes the adversarial optimization while remaining first-order accurate in the ambiguity radius. To learn the fixed point of this approximate equation, we propose Mean-Variance Stochastic Approximation (MVSA), a model-free algorithm that uses only one-sample updates. This is achieved via a lifted stochastic approximation dynamics and a two-time-scale design. We then prove convergence and a central limit theorem for MVSA: its main iterate satisfies a central limit theorem at the canonical $n^{-1/2}$ scale, with explicitly characterized asymptotic covariances. Finally, we validate our theoretical findings with a numerical experiment.
Read the full abstract

Complex Variables

Structural aspects of extremal functions in the Krzyż conjecture

Extremal functions for the $n$th coefficient in the Krzyż conjecture are atomic singular inner functions with at most $n$ atoms. This paper gives a lower bound on the number of atoms $N$ of the form $N\geq cn$, marking progress toward proving the expected $N=n$. Furthermore, we prove new formulas for extremal functions using variational techniques. Using these results and several other methods, we establish new conditions on extremal functions which are equivalent to the Krzyż conjecture being true. We also characterize the possible analytic invariants of extremal functions.
Read the full abstract

Statistics Theory

On Observation Time for Recovering Latent Hawkes Networks

Dynamics of interacting systems in engineering, society, and nature often evolve over latent networks that govern which entities can interact. We study the problem of inferring these networks from event-based observations, which arise naturally in finance, seismology, and neuroscience. While there is substantial algorithmic work addressing this important problem, theoretical results are scarce. In this paper we ask the following fundamental question: what is the minimum time that one must observe the dynamics in order to exactly recover the underlying network, as a function of the number $d$ of interacting entities? For a class of stationary Hawkes processes with sparse, weak interactions, we prove that an observation time of order $\log d$ is sufficient and necessary. For the upper bound we construct a two-stage estimator that uses clipped and binned event data for screening, followed by a least-squares refinement, and apply concentration bounds derived from the Poisson cluster representation. For the lower bound we combine Fano's inequality with Jacod's Girsanov formula for point processes on a suitable subclass of networks.
Read the full abstract

All authors Jonas Linkerhägner, Michele Bortolasi, Lorenzo Baldassari, Maarten V. de Hoop, Ivan Dokmanić

Algebraic Geometry

Equivariant Quotients of Derived Symplectic Spaces and Legendrian Intersection Theorem

The classical transversality lemma of contact geometry constructs a contact structure on a hypersurface transverse to a Liouville vector field using point-set topology and local flows. This paper translates the classical transversality lemma into the context of derived algebraic geometry and provides the derived Legendrian intersection theorem, along with various applications to moduli theory. In brief, we first prove that taking the quotient of a derived symplectic space descends the symplectic data to a contact structure, avoiding the transverse hypersurface requirement, such that the fundamental vector field of a weight 1 $\mathbb{G}_m$-action in the derived setting replaces the classical Liouville vector field. Second, we establish the derived Legendrian intersection theorem using base change, an $\infty$-categorical descent cube, and $\mathbb{G}_m$-equivariant lifts along the symplectification projection. As applications of the main results, we first examine the derived geometry of the discriminant loci of 1-jet bundles and show that these loci carry a $(-1)$-shifted contact structure. In addition, we show that our results apply to certain moduli problems, including projective Higgs bundles, $\ell$-adic local systems, and Lie 2-groups, and produce further examples of contact derived moduli stacks.
Read the full abstract

Dynamical Systems

Straight-line trajectories on the Mucube

The dynamics of straight line flows on compact half-translation surfaces (surfaces formed by gluing Euclidean polygons edge-to-edge via translations possibly composed with rotation by $\pi$) has been widely studied due to their connections to polygonal billiards and Teichmüller theory. However, much less is known when the underlying surface is non-compact or infinite type. In this paper, we consider the straight line flow of the Mucube – an infinite $\mathbb{Z}^3$-periodic half-translation square-tiled surface – first written about by Coxeter and Petrie and more recently studied by Athreya–Lee and Gutiérrez-Romo–Lee–Sánchez. We give a geometric description of the flow's periodic and drift orbits in terms of the Mucube's rigid symmetries, and we give a complete characterization of the set of directions in which the straight line flow is periodic on the Mucube – first in terms of a genus one quotient and second in terms of an infinitely generated subgroup of $\mathrm{SL}_2(\mathbb{Z})$. We use the latter characterization to obtain the Veech group (i.e. group of derivatives of affine diffeomorphisms) of the Mucube. Finally, we prove density of the sets of periodic and ergodic directions.
Read the full abstract

All authors Andre Oliveira, Felipe A. Ramírez, Chandrika Sadanand, Sunrose T. Shrestha

Mathematical Physics

Dispersive decay bounds for the SSH model on the half-line

We study the Schrödinger flow for the SSH model, a class of self-adjoint discrete dimer lattice Hamiltonians on the half-line. Using oscillatory integral techniques, we prove dispersive time-decay estimates, which quantify the spreading of energy throughout the lattice for a localized initial condition. Furthermore, we determine precise dependence of the constants in the decay rates on the parameters of the Hamiltonian. The analysis is complicated by the fact that as a consequence of the boundary condition, the expression for the propagator contains oscillatory integrals with nonintegrable singularities.
Read the full abstract

Differential Geometry

Canonical parameters on marginally trapped surfaces in the Minkowski 4-space

Marginally trapped surfaces are spacelike surfaces in the Minkowski space whose mean curvature vector is lightlike at each point. In general, the marginally trapped surfaces are determined by seven functions satisfying several conditions (differential equations). In the present paper, we introduce special principal parameters, called canonical, and prove that every marginally trapped surface of general type admits (at least locally) canonical principal parameters which allow us to reduce the number of functions. We prove a Fundamental existence and uniqueness theorem formulated in terms of canonical parameters, which states that every marginally trapped surface of general type is determined up to a motion by three smooth functions satisfying a system of partial differential equations. The special subclass of marginally trapped surfaces with parallel mean curvature vector field is described in terms of only one differential equation. Finally, the theory is enriched with examples of marginally trapped surfaces for which canonical parameters are determined.
Read the full abstract

Numerical Analysis

Kernel-based linear system identification using augmented Krylov subspaces

We propose a novel Krylov subspace method for estimating the finite impulse response (FIR) of a one-dimensional linear time-invariant systems. The method approximates the system's FIR using a kernel-based formulation combined with hyperparameter selection based on maximum likelihood estimation (MLE), which requires repeated evaluation of two terms: The data fit $\boldsymbol{y}^{\top} (\lambda \boldsymbol{I} + \boldsymbol{A})^{-1} \boldsymbol{y}$ and the model complexity $\log(\det (\lambda \boldsymbol{I} + \boldsymbol{A}))$, where $\boldsymbol{A}$ is a certain positive semidefinite matrix that admits fast matrix–vector products and $\lambda > 0$ is a regularization parameter. Instead of approximating these two quantities separately, we jointly approximate them using a single augmented Krylov subspace for $\boldsymbol{A}$. One major benefit of augmentation is that we obtain accelerated convergence when approximating the data fit quadratic form, through implicit preconditioning. Thanks to the shift invariance of Krylov subspaces, the extracted approximations can be used to evaluate the MLE objective for many values of $\lambda$ at little additional cost. We derive error bounds for the approximations, reflecting the benefits of augmentation demonstrated through multiple numerical experiments.
Read the full abstract

All authors Fabio Matti, Martin Skovgaard Andersen, Tianshi Chen, Daniel Kressner

Probability · Cross-listed from cs.LG

Convergence Analysis of Newton's Method for Neural Networks in the Overparameterized Limit

A convergence analysis is developed for the regularized Newton method for training neural networks (NNs) in the overparameterized limit. As the number of hidden units tends to infinity, the NN training dynamics converge in probability to the solution of a deterministic limit equation involving a "Newton neural tangent kernel" (NNTK). Explicit rates characterizing this convergence are provided and, in the infinite-width limit, we prove that the NN converges exponentially fast to the target data (i.e., a global minimizer with zero loss). We show that this convergence is uniform across the frequency spectrum, addressing the spectral bias inherent in gradient descent. The eigenvalues of the NTK for gradient descent accumulate at zero, leading to slow convergence for target data with high-frequency components. In contrast, the NNTK has uniformly lower bounded eigenvalues if the regularization parameter is selected appropriately, allowing Newton's method to converge more quickly for data with high-frequency components. Mathematical challenges that need to be addressed include the implicit parameter update of the Newton method with a potentially indefinite Hessian matrix and the fact that the dimension of this linear system of equations tends to infinity as the NN width grows. This substantially complicates deriving the training dynamics in the overparameterized limit as well as proving the convergence of the finite-width dynamics thereto. Our analysis identifies a scaling formula for selecting the regularization parameter, which we show can vanish at a suitable rate as the NN width becomes larger. In addition, we prove that, for sufficiently large numbers of hidden units, the regularized Hessian remains positive definite during training and the Newton updates for individual NN parameters converge to zero, demonstrating that the model behaves as a linearization around the initialization.
Read the full abstract