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)

Wednesday 13 May

Combinatorics

Eventual sign coherence

The sign coherence of $c$-vectors is one of the fundamental theorems of cluster algebras with principal coefficients. In 2019, Gekhtman and Nakanishi posed the asymptotic sign coherence conjecture for arbitrary cluster algebras of geometric type, which says sign coherence should eventually hold in any sufficiently generic infinite mutation sequence. We prove that their conjecture holds almost always for skew-symmetric cluster algebras of arbitrary rank. That is, we prove that with probability $1$, the sequence of $c$-vectors obtained by random mutation of an arbitrary quiver eventually becomes sign-coherent. Our results also establish the conjecture in full generality for many families of quivers by studying a new class of brog quivers.
Read the full abstract

Combinatorics

A study on Type-2 isomorphic circulant graphs. Part 4: 960 triples of Type-2 isomorphic circulant graphs $C_{54}(R)$

This article is the $4^{th}$ part of a detailed study on Type-2 isomorphic circulant graphs having ten parts \cite{v2-1}-\cite{v2-10} and is a continuation of Part 3. Here, we obtain all 960 triples of Type-2 isomorphic circulant graphs of order 54 and each triple of isomorphic circulant graphs is of Type-2 isomorphic w.r.t. $m$ = 3.
Read the full abstract

Rings and Algebras

Commuting maps of inflated algebras

Commuting maps on a class of algebras called inflated algebras are investigated. In particular, we can prove that every commuting map $\theta$ on such an algebra is of the form $\theta(x)=c x+\mu(x)$, where $c$ belongs to the base field $K$ of characteristic not 2, and $\mu$ is a central-valued linear map.
Read the full abstract

Numerical Analysis

A refined CJ–SS–RR method with a reliable removal approach of spurious Ritz values for the Hermitian eigenvalue problem

Under the hypothesis that the deviations of the desired eigenvectors of the matrix $A$ from the underlying subspace tend to zero, the Ritz vectors may not converge and have poor or little accuracy. This phenomenon is not unusual and particularly occurs when the associated Ritz values are close, which is independent of the eigenvalue distribution of $A$. For the (block) SS–RR methods, there are possibly more Ritz values that converge to the same desired eigenvalue(s) counting multiplicity in the region of interest, meaning that some of the Ritz values must be spurious and the corresponding residual norms of the Ritz pairs may not be small. Consequently, the (block) SS–RR methods including the CJ–SS–RR method cannot base on the corresponding residual norms to effectively identify if the Ritz values in the region are genuine or spurious. This paper proposes refined SS–RR, abbreviated as SS–RRR, methods based on the refined Rayleigh–Ritz projection that compute the eigenpairs of large matrices with the eigenvalues located in the given region. We present a new approach to accurately implement the RRR methods more efficiently than ever before for a general subspace.Exploiting the unconditional convergence of the refined Ritz vectors when the subspace is sufficiently accurate, we propose a tune-free removal approach to effectively remove spurious Ritz values with a rigorous theory supported, and develop a restarted CJ–SS–RRR algorithm. Numerical experiments show that the restarted CJ–SS–RRR algorithm is more efficient and effective than the restarted CJ–SS–RR algorithm.
Read the full abstract

Numerical Analysis

Walk on spheres and Array-RQMC

We use Array-RQMC sampling in a walk on spheres (WoS) algorithm for Dirichlet boundary value problems. On a collection of problems, we find that Array-RQMC-WoS reduces the Monte Carlo MSE or variance by factors ranging from $71$-fold to $3087$-fold at $n=2^{17}$ trajectories. The variance is known to be $o(1/n)$ but attains empirical rates between $n^{-1.4}$ and $n^{-1.8}$ in our examples. A simpler RQMC-WoS algorithm studied in Ho and Owen (2026) has more theoretical support but only reduced variance by 1.8 to 10.7-fold on the same set of examples. In order to explain this improvement, we introduce a column-wise mean dimension of the RQMC error based on Sobol' indices. It matches the usual mean dimension for Monte Carlo and the mean dimension of a dual lattice error for randomized lattices. We find for a gasket example from Crane et al. (2025) that the mean dimension of Array-RQMC-WoS errors is much higher than an analogous Array-MC-WoS algorithm has. v2 replaced v1's QMCPy lattice with Korobov lattices from LatNet Builder, but left the old abstract in the meta-data v3 corrects the v2 abstract in this meta data
Read the full abstract

Information Theory · Cross-listed from physics.class-ph

Natural frequency estimation using complex-frequency excitations

Complex frequency excitations, oscillating signals whose amplitude decreases exponentially in time, have recently been demonstrated to significantly increase the effective quality factor of mechanical resonators. In this work, we investigate the accuracy of natural frequency estimation in mechanical systems under noise using such excitations. The analysis is performed on an underdamped linear time-invariant single-degree-of-freedom spring-mass-damper system. We employ tools from information theory, namely Fisher information, to systematically quantify the sensitivity of complex-frequency excitation to measurement noise. Explicit closed-form expressions are derived relating Fisher information to excitation and system parameters under both Gaussian white and colored noise. The theoretical predictions are verified through Monte Carlo numerical simulations. The results indicate that appropriate selection of excitation parameters can significantly enhance the Fisher information, leading to improved estimation accuracy under complex-frequency excitations compared with conventional harmonic excitations. Experimental results demonstrate the advantages of complex-frequency excitation in terms of both accuracy and robustness of natural-frequency estimation. These findings establish a foundation for the development of high-performance sensors and novel nondestructive evaluation methods.
Read the full abstract

Combinatorics

A short proof of Mathar's 2021 recurrence conjecture for the Lehmer-Comtet diagonal A045406

For OEIS sequence A045406, the column-2 diagonal of the Lehmer-Comtet triangle A008296, R. J. Mathar contributed in September 2021 the conjectured order-2 P-recursive recurrence \[ a(n) + (2n-7)\,a(n-1) + (n-4)^{2}\,a(n-2) \;=\; 0,\qquad n \ge 5. \] We give a short proof. Detlefs's harmonic-number closed form $a(n) = (-1)^n (2 H_{n-3} - 3)(n-3)!$ for $n \ge 3$ collapses the left-hand side, after factoring out $(-1)^n (n-5)! (n-4)$, to a polynomial identity in $n$ with coefficient $H_{n-4}$. The $H_{n-4}$-coefficient simplifies to $(n-3) - (2n-7) + (n-4) = 0$ (using $H_{n-3} = H_{n-4} + 1/(n-3)$ and $H_{n-5} = H_{n-4} - 1/(n-4)$); the constant remainder is $0$ for the same reason. The supplementary archive contains a SymPy script verifying both pieces symbolically, the e.g.f.\ expansion against the harmonic closed form, and Mathar's recurrence numerically for $n = 5, \ldots, 5000$.
Read the full abstract

2605.12839 · 13 May · Duplicates arXiv:2605.11351 (same author, one day earlier). Both prove one theorem: with v(m)=m!(2H_m-3), A045406(n)=(-1)^n v(n-3) and A001711(n)=v(n+3)/4, so both stated recurrences are the single identity v(m)-(2m-1)v(m-1)+(m-1)^2 v(m-2)=0; the printed terms agree up to the factor 4 after alignment. Withdrawing the later paper; 2605.11351 stands

Dynamical Systems

Pseudo-Anosov flows and the geometry of Anosov-like group actions

We show that the action on its orbit space induced by a pseudo-Anosov flow on a closed $3$-manifold (and more general Anosov-like actions) can be seen as an isometric action on a Gromov-hyperbolic space. When the flow is not $\R$-covered, we show that this action admits elements that are weakly properly discontinuous and deduce that elements of $\pi_1(M)$ that do not represent a periodic orbit of the flow are generic for any word metric coming from a finite generating set. We also give a number of other geometric group-theoretic results for Anosov-like group actions on bifoliated planes.
Read the full abstract

All authors Thomas Barthelmé, Kathryn Mann, Neige Paulet, Abdul Zalloum

Geometric Topology

Underlying Stokes and de Rham structures for Arnold-type invariants

We introduce a framework on dual complexes for studying Arnold-type invariants of immersed curves and immersed surfaces via local finite-difference structures associated with Alexander numberings. For generic immersed plane curves and generic immersed surfaces, we define locally normalized maps $d^k \phi$ on dual skeleta and show that suitable evaluations recover the Arnold-type invariants $St_{(1)}$ and $St_{(2)}$. In particular, we establish normalized discrete Stokes-type compatibilities between adjacent dual skeleta and derive corresponding Shumakovitch-type identities for curves and surfaces. The normalization coefficients are determined by finite-difference factorial structures together with multiplicities of local configurations. We further interpret the iterated-integral-type structures appearing in Shumakovitch-type identities through finite-difference structures and highest-degree local Stokes compatibilities on dual complexes. We also reinterpret the slice formula for $St_{(2)}$ and $St_{(1)}$ as a compatibility relation between slicing and local operations on the dual complex. These results provide a unified framework in which global Arnold-type invariants arise as distribution-type evaluations of local data on dual complexes. The framework further clarifies the distinction between untwisted local closures and globally twisted structures such as the original Arnold invariant $St$, and suggests the existence of higher-degree local operations associated with the same dual-complex structure.
Read the full abstract

Tuesday 12 May

Rings and Algebras

Transposed Poisson structure on the Witt-type algebra $\mathcal{W}(a,-1)$: Derivations, Automorphisms, and Rota–Baxter operators

In this paper, we provide a comprehensive study of the structural properties of the transposed Poisson algebra $\mathcal{W}(a,-1)$. We classify several types of linear maps, including derivations, local derivations, quasi-derivations, and $\delta$-derivations, showing that non-trivial $\delta$-derivations exist only for $\delta=1$ and $\delta=\frac{1}{2}$. Furthermore, we describe the groups of automorphisms, local automorphisms, 2-local automorphisms, and quasi-automorphisms. We also investigate Rota–Baxter operators of weight $1$ on $\mathcal{W}(a,-1)$. Specifically, we classify operators that are homogeneous with respect to both the standard $\mathbf{Z}$-grading and a $\mathbf{Z}_2$-grading, establishing a rigidity result for the latter case. Finally, we classify all $\mathbf{W}$-compatible Novikov–Poisson structures, demonstrating that the associative product on the Witt algebra is universally compatible with its known Novikov structures.
Read the full abstract

Information Theory · Cross-listed from cs.LG

SynIB: Informational Bottleneck for Maximizing Synergy in Multimodal Learning

A central objective in multimodal learning is to capture synergy: task-relevant information that arises only from the joint use of multiple modalities, and is not available from any single modality alone. While most approaches operate at the architectural level through larger or more complex fusion models, we propose a complementary axis: shaping the training objective itself. Standard training often emphasizes unimodal or redundant information, falling short on examples that require cross-modal reasoning. We formalize multimodal synergy through information theory and introduce the Synergistic Information Bottleneck (SynIB), a scalable objective that targets synergy directly. To prioritize learning synergy, SynIB motivates the model to predict accurately from all modalities while penalizing confidence when information from any modality is withheld. Alongside the standard task loss, the model runs forward passes with one modality masked at a time and is penalized for remaining confident, which would indicate reliance on unimodal cues rather than cross-modal interactions. We validate SynIB in two regimes. On synthetic XOR tasks where the ground-truth synergy is known by construction, standard training fails to recover it while SynIB does. On five real-world benchmarks, including three MultiBench affective tasks, Hateful Memes with CLIP-ViT and DeBERTa backbones, and a controllable irony extension of CREMA-D we introduce, SynIB improves accuracy on synergy-dependent examples by up to 7.8% and overall accuracy by up to 3.8%.
Read the full abstract

All authors Konstantinos Kontras, Teodora Gagaleska, Thomas Strypsteen, Christos Chatzichristos, Matthew Blaschko, Maarten De Vos, Paul Pu Liang

Numerical Analysis

Euler Scheme for Stochastic Functional Differential Equations Driven by Fractional Brownian Motion via Fractional Calculus Techniques

We study a stochastic functional differential equation (SFDE) with memory driven by a fractional Brownian motion (fBm) with Hurst parameter H>1/2. An Euler-type numerical scheme is proposed and analyzed under suitable regularity conditions on the drift and diffusion coefficients using tools from fractional calculus. We prove the convergence of the scheme and derive the corresponding rate in terms of the discretization step. Numerical simulations illustrate the theoretical results and confirm the accuracy of the proposed method.
Read the full abstract

All authors Alexander Abreu, Héctor Araya, Lisandro Fermin, Johanna Garzón, Soledad Torres

Numerical Analysis · Cross-listed from quant-ph

Neural Quantum Spectral Operator Learning for Solving Partial Differential Equations

Partial differential equations (PDEs) are central to modeling physical and engineering systems, but repeatedly solving parametric PDEs remains computationally expensive. Operator learning enables fast surrogate inference, yet typically requires large input-output paired datasets generated by costly high-fidelity PDE solvers. Unsupervised operator learning frameworks alleviate data dependency but remain hindered by computational bottlenecks. To address this, we propose Neural Variational Quantum Linear Solver (NVQLS), the first hybrid quantum-classical operator learning framework leveraging the Legendre–Galerkin weak formulation. We critically resolve the sign ambiguity in VQLS energy minimization, preventing erroneous solution representations. Additionally, we introduce a neural embedding, a novel encoding scheme to map varying forcings and PDE coefficients into parameterized quantum circuit representations. These structural innovations provide theoretical computational complexity advantages under efficient state preparation schemes, while achieving superior accuracy compared to a representative classical baseline. Validations on 1D and 2D parametric PDEs under diverse boundary conditions demonstrate NVQLS's capability to simultaneously process varying inputs, offering a scalable unsupervised approach to quantum-enhanced operator learning.
Read the full abstract

All authors Chanyoung Kim, Myeonghwan Seong, Yujin Kim, Daniel K. Park, Youngjoon Hong

Combinatorics · Cross-listed from cs.CL

TriVAL: A Tri-Validation Framework for Faithful Automatic Optimization Modeling

Optimization modeling serves as the pivotal bridge between natural-language problem descriptions and optimization solvers, and remains a cornerstone for bringing operations research (OR) into real-world decision making. Recent advances in large language models (LLMs) have driven significant progress in automatic optimization modeling. However, existing methods still lack explicit validation during the modeling process, allowing errors introduced in earlier stages to carry through the pipeline and ultimately reduce final modeling accuracy. To address this challenge, we introduce TriVAL, a tri-validation framework that performs explicit validation at three stages of automatic optimization modeling: semantic specification, mathematical formulation, and code generation. At each stage, TriVAL follows a construct-validate-revise loop that assesses the current result against stage-specific criteria and revises it when needed. This design helps identify and correct errors before they accumulate across stages, helping preserve faithfulness throughout the modeling process. To evaluate automatic optimization modeling on more challenging combinatorial problems, we further introduce NL4COP, a benchmark of 150 instances across 50 diverse problem types with more complex decision logic, more tightly coupled constraints, and more demanding modeling requirements than existing benchmarks. Experiments on NL4COP and established benchmarks show that TriVAL consistently outperforms state-ofthe-art methods, with the largest gains on the most challenging problems.
Read the full abstract

All authors Ziyang Fang, JinXi Wang, Jinghui Zhong, Yew-Soon Ong

Combinatorics

Half-flips are 5-avoidable

A word contains a half-flip if it contains non-empty factors $uv$ and $vu$ where $|u|=|v|$. Fici reports a non-constructive proof of the existence of an infinite word over a finite alphabet avoiding half-flips and asks for the size of the smallest alphabet over which half-flips may be avoided. Currie and Rampersad have proposed a pure morphic word over 8 letters and a morphic word over 5 letters and conjecture that they avoid half-flips. We present a pure morphic word over 5 letters that avoids half-flips. We also show that half-flips with $|u|\ge2$ are 3-avoidable and that half-flips with $|u|\ge4$ are 2-avoidable.
Read the full abstract

Information Theory · Cross-listed from cs.AI

NOVA: Fundamental Limits of Knowledge Discovery Through AI

Can AI systems discover new knowledge through iterative self-improvement, and at what cost? We introduce NOVA, which models the “generate, verify, accumulate, retrain” loop as an adaptive sampling process over a knowledge space. We give sufficient conditions for accumulated genuine knowledge to cover a finite domain and show how violations produce contamination, forgetting, exploration failure, and acceptance failure. We then analyze how adaptive generation arises from recursive retraining. In an explicit distribution-level model where accepted artifacts influence the next generator, we identify a recursive-feedback phase transition. Unanchored feedback can lock generation onto early accepted artifacts and leave initially reachable valid artifacts undiscovered with positive probability. Anchoring updates to a persistent base distribution prevents unbounded distortion and guarantees continued exposure. Under imperfect verification, we identify a contamination trap: as easy knowledge is exhausted, even small false-positive rates can admit invalid artifacts faster than genuine discoveries. We show that Good–Turing estimation is a local batch-diversity diagnostic, not an estimator of the historically undiscovered valid mass governing long-term progress. Under a Zipf tail with exponent $\alpha>1$, the cumulative generation cost of obtaining $D$ distinct genuine discoveries satisfies $R_{\rm cum}(D)=\Theta(c_{\rm gen}D^\alpha)$. When the valid base distribution has such a tail, anchored retraining preserves the exposure needed for this scaling law. Finally, we show how human guidance, generation, and verification can redirect or expand discovery when autonomous sampling stalls because of repetition, vanishing exposure, or unreliable verification.
Read the full abstract

2605.15219 · 12 May · Added an explicit recursive retraining model showing how accepted outputs reshape future generation. New results characterize when repeated retraining suppresses undiscovered artifacts and when mixing updates with a fixed base distribution preserves exposure. Corrected the Zipf discovery-cost proof and expanded the analysis. Main results and implications remain unchanged

Dynamical Systems

Characterization of stability radii for robustly asymptotically stable dissipative Hamiltonian differential-algebraic systems

We study linear time-invariant dissipative Hamiltonian differential-algebraic systems. We characterize when the systems are robustly asymptotically stable and derive exact conditions and bounds when this property is lost under structure-preserving perturbations.
Read the full abstract

All authors Peter Benner, Volker Mehrmann, Anshul Prajapati, Punit Sharma

Number Theory

Local certification of residual squareclasses in $\mathbb Q(\sqrt{2},\sqrt{pq},\sqrt{ps})$: one-bit, affine, and finite-choice Hilbert-symbol frameworks

Recent works of El Hamam described explicit fundamental systems of units for several families of multiquadratic fields of degrees 8 and 16. In the degree-8 field $L^+ = \mathbb{Q}(\sqrt{2}, \sqrt{pq}, \sqrt{ps}),$ the corrected classification still leaves a residual binary indeterminacy: one must decide which of two explicitly constructed squareclasses gives the final unit generator. In this paper, we make this remaining bit explicit. First, we give an explicit local criterion deciding the parameter $\mu \in \{1, \epsilon_{pq}\}$ left open in recent literature. The criterion is first expressed in terms of Hilbert symbols at a single finite place, and is then sharpened to a residue criterion at a chosen split auxiliary rational prime. Second, we show that the standard residue datum $D(p,q,s) = \left( p \bmod 8,\,\, q \bmod 8,\,\, s \bmod 8,\,\, \biggl(\dfrac{q}{p}\biggr),\,\, \biggl(\dfrac{s}{p}\biggr),\,\, \biggl(\dfrac{q}{s}\biggr) \right)$ does not determine the final generator: we compute explicit triples with the same $D(p,q,s)$ but opposite values of the residual bit. Third, we place the one-bit problem inside a hierarchy of local-certification results in $K^\times/K^{\times2}$: besides the linear residual-choice statement, we prove an affine local-certification theorem for residual-choice cosets and a finite-test-set separation theorem for arbitrary finite candidate families.
Read the full abstract

Combinatorics

Unimodality of $q$-Fibonomial coefficients for small cases

Bergeron–Ceballos–Küstner introduced the $q$-Fibonomial coefficients \( \qfibonom{m+n}{n}\), gave a combinatorial interpretation of the $q$-Fibonomial coefficients via a weighted path-domino tiling model, and conjectured that these polynomials are unimodal. We prove the conjecture for $n\leq3$. For the $n=2$ case, we give a combinatorial proof of both unimodality and symmetry by defining a nearly symmetric saturated chain decomposition on the set of tilings. For all three cases, we give an algebraic proof. Finally, for the $n=3$ case, we establish a more general unimodality result for certain products of $q$-analogs and propose several related conjectures.
Read the full abstract

All authors Brendan B. Connelly, Ezekiel Ito, Thomas C. Martinez, Olha Shevchenko, Kacey Yang

Combinatorics

The Distribution of the Deepest Leaves in Binary Trees

We study the extreme local structure of plane binary trees through the limiting distribution of the number of leaves at maximum depth. For every $m\geq1$, we determine the limiting probability $\kappa^{[m]}$ that the deepest level contains $2m$ leaves. The resulting sequence is encoded by a generating function $K(y)$, obtained from a bivariate refinement of the Catalan iteration. We show that, at the critical point of this iteration, a parabolic recurrence governs the marked generating functions. Using continuous iteration theory and a Fatou coordinate, we transform this recurrence into the functional equation $K((1+y/4)^2)=K(y)+4$, which describes the limiting law. Its singular behavior yields the exponential tail $\kappa^{[m]}\sim4^{-m+1}$. Numerical evaluation gives, e.g., $\kappa^{[1]}\approx0.7009$, $\kappa^{[2]}\approx0.2226$ and an average number of deepest leaves equal to $\hat{\kappa}=4\kappa^{[1]}\approx2.8037$.
Read the full abstract

Optimization and Control

The Method of Ellipcenters for strongly convex minimization

This work is about ME, the Method of Ellipcenters. ME was recently introduced by these very authors as a first order accelerated scheme for unconstrained minimization. Its iterates are all centers of ellipses carefully designed to somehow capture ill-conditioning of the underlying optimization problem. In the first article on ME, we were able to prove that it converges with linear rate when the objective function is quadratic and strongly convex, while here we derive convergence for any differentiable strongly convex objective. This investigation was inspired by the great performance of ME in quadratic minimization against steepest descent with exact line search, FISTA, Barzilai-Borwein and Conjugate Gradient. The experiments we carry out now, make ME even more attractive from the numerical point of view. On top of that, the theory seems promising for quite more general settings.
Read the full abstract

All authors Roger Behling, Ramyro Correa, Eduarda Ferreira, Vincent Guigues

Optimization and Control

Accuracy and Relationships of Quadratic Models in Derivative-free Optimization

We study three quadratic models in model-based derivative-free optimization: the minimum norm (MN), minimum Frobenius norm (MFN), and quadratic generalized simplex derivative (QS) models. Despite their widespread use, their approximation accuracy and relationships have not been systematically explored. We establish fully linear error bounds for all three models, removing the uniformly bounded model Hessian assumption required in existing MN analyses and deriving the first such results for the QS model. We further analyze Hessian approximation accuracy via directional error bounds, showing that all three models achieve fully quadratic accuracy along sample directions under a mild condition on the sample set. This reveals a form of directional fully quadratic accuracy not captured by existing theory. Finally, we characterize the relationships among these models, identifying conditions under which they coincide and clarifying their structural connections.
Read the full abstract

Differential Geometry

Complex methods in the asymptotics of Möbius energy integrals of helix curves

The Möbius energy of a curve is a topic of interest to physical knot theorists, harmonic analysts, and geometric analysts. In particular, there are many open questions regarding the gradient flow and critical points. The Gateaux derivative indicates the variation is dependent on curvature and torsion, and so to better understand the gradient, we investigate the family of helix curves, where the ratio of torsion to curvature is a constant proportional to the pitch. We fix the radius of the helix, and study the coiling in both directions: as the helix unravels to a straight line, and as it coils infinitely tight. Specifically, we study the arclength-rescaled Möbius energy density, which is a naturally tractable quantity under the Möbius energy's chord-arc comparison of inverse-square laws. The asymptotics of the uncoiling helix, corresponding to an energy decay, can be proven with a short chain of estimates. However, the asymptotics of the helix as it coils infinitely tight, blowing up the energy, is a much more involved calculation. Our strategy for calculating the asymptotics, initially reminiscent of the work by Kim-Kusner, begins with a meromorphic extension of the integrand. However, proving the asymptotic equivalence requires a fundamentally distinct strategy because our integrand has infinitely many poles. keywords: Möbius energy, helix, complex asymptotics, knot energies, physical knot theory, coiling, curves
Read the full abstract

Combinatorics

The unbreakable quasi-graphic matroids

A matroid M is unbreakable if it is connected and M/F is connected for every flat F of M . Oxley and Pfeil characterized the unbreakable graphic matroids, and Fife, Mayhew, Oxley, and Semple characterized the graphs underlying 3-connected unbreakable frame matroids. We extend the latter result by giving a complete characterization of the 3-connected unbreakable quasi-graphic matroids. As a special case we obtain a characterization of the 3-connected lifted-graphic matroids.
Read the full abstract

Information Theory · Cross-listed from stat.CO

Multi-Marginal Couplings for Metropolis-Hastings

Convergence diagnosis for Markov chain Monte Carlo is a matter of fundamental importance in computational statistics: it determines the resources allocated to a particular sampling problem and influences the practitioner's view of the quality of estimates obtained from a Markov chain. Motivated by this, we contribute to the emerging class of coupling-based convergence diagnostic algorithms. Concretely, we study coupling multiple Metropolis-Hastings chains using multi-marginal coupling. We introduce a natural objective for this setting and establish lower and upper bounds by drawing connections to list-level distribution coupling and distributed pairwise-matching problems. This analysis ultimately leads to a shared-randomness Poisson Monte Carlo construction for coupling multiple Markov chains. In this process, we avoid a key dimension-dependent bottleneck in the runtime complexity of classical Poisson Monte Carlo by developing an adaptive rule for updating the point process, yielding significant gains in high-dimensional settings. Experiments on grand couplings of Markov chains show that our methods improve coalescence rates across dimensions, reducing meeting times by up to 50% compared with existing baselines.
Read the full abstract

All authors Buu Phan, Gergely Flamich, Ashish Khisti, Shahab Asoodeh

Numerical Analysis

Fast and Stable Gradient Approximation for Quadratic Forms of Hermitian Matrix Functions

Objectives involving quadratic forms $u^*f(A(\theta))u$ for Hermitian $A$ arise widely in scientific computing and machine learning. Lanczos efficiently approximates these quantities, but differentiating through its recurrence can introduce numerical instability and costly reorthogonalization. Building on existing low-rank Krylov approximations of Fréchet derivatives, we develop a matrix-free gradient implementation that reuses the forward Lanczos quantities without differentiating through the recurrence. We analyze its discrepancy from the exact gradient of the finite Lanczos approximation, deriving an exact residual-coupled error identity and a bound involving the best degree-$(m-1)$ uniform polynomial approximation error of $f'$, multiplied by a basis-sensitivity factor. The discrepancy vanishes when the Krylov subspace is invariant. Experiments demonstrate accurate gradients and runtime improvements over differentiated-Lanczos and modified-Arnoldi baselines. It remains numerically stable without reorthogonalization in all tested settings, including cases where adjoint-based Lanczos differentiation becomes unstable.
Read the full abstract

Information Theory

Resolution Information: Limits of Ambiguity Resolution for Generative Communication

In generative communication, the transmitter sends a compact generative description, such as model parameters or a latent representation, rather than raw data. The receiver uses this description to form a posterior belief over the underlying state and to resolve semantic ambiguity: which interpretation, decision, or action is supported by the received representation? Inspired by Shannon's geometric view of communication as uncertainty resolution, we introduce resolution information as the minimum information update, measured in nats, required to move the receiver's posterior belief into a low-ambiguity semantic region. Our work yields three main results. First, when the receiver can form any posterior belief, corresponding to the ideal unconstrained case, resolution information reduces to a binary divergence that depends only on each region's prior probability. In this case, the shape of the regions is irrelevant. Under repeated sampling, ambiguity decays exponentially with an exponent equal to the resolution information, giving it an operational meaning as an ambiguity exponent. Second, when the generative representation constrains the posterior family, as in practice, geometry becomes operational and can create irreducible ambiguity floors: half-spaces remain resolvable, whereas polytope-type regions can exhibit residual ambiguity that no amount of additional information can remove. These results reveal a fundamental departure from classical channel coding. In Shannon theory, codes can be designed so that decoding regions separate messages and error probability vanishes below capacity. In generative communication, the model itself induces a constrained posterior geometry that may prevent asymptotic ambiguity resolution. The resulting limit is not on rate, but on resolvability itself.
Read the full abstract

Information Theory

Quantum Precoded Polar Codes

We introduce a new family of CSS codes obtained from rate-1 precoded polar codes, which harnesses the precoding benefits obtained for classical short blocklength polar codes. We optimize the rate profile and precoder of these codes with a genetic algorithm, and present codes of dimension $ [\![256, 2 ]\!] $ and $ [\![512, 2]\!] $ that have logical error rates similar to the $ [\![1201, 1, 25 ]\!] $ surface code over the depolarizing channel.
Read the full abstract

Combinatorics

On groups with D-finite cogrowth series

The cogrowth series of a group with respect to a finite generating set is an important combinatorial quantity that seems very difficult to compute exactly, as evidenced by the scarcity of known examples. In this paper, we give a particular infinite family of presentations for which the cogrowth series can be determined as the constant term of an algebraic function, which shows that it is D-finite and, with more work, not algebraic. Our proof exploits the fact that for a particular choice of subgroup, the corresponding Schreier graph has finite tree width, and by considering paths in the cosets and the Schreier graph separately, we are able to construct a system of generating functions which count paths. We find the asymptotics of this system to conclude that the groups have D-finite but non-algebraic cogrowth series. We also apply our method to some additional examples which have some similarities with the infinite family above, and again show they have D-finite but non-algebraic cogrowth series. These examples lend some support to the conjecture that if a group has an algebraic cogrowth series, then it must be virtually-free, and adds to the small collection of known examples of groups having D-finite cogrowth series for at least one finite generating set.
Read the full abstract

Dynamical Systems · Cross-listed from cs.LG

Identifying the nonlinear string dynamics with port-Hamiltonian neural networks

Hybrid machine learning combines physical knowledge with data-driven models to enhance interpretability and performance. In this context, Port-Hamiltonian Systems (PHS), which generalize Hamiltonian mechanics to describe open, non-autonomous dynamical systems, have been successfully integrated with neural networks under the name Port-Hamiltonian Neural Networks (PHNNs). While the ability of PHNNs to identify Hamiltonian ordinary differential equation (ODE) systems has already been demonstrated, their application to learning Hamiltonian partial differential equation (PDE) systems remains largely unexplored. This limitation restricts their use in musical acoustics, where instruments are typically modeled as distributed parameter systems governed by PDEs. In this work, we demonstrate how to learn the nonlinear string dynamics from data in a physically-consistent framework through a PHNN extension to PDEs. By constructing structured neural network architectures based on PHS, we can recover both the Hamiltonian governing the string and the dissipation affecting it. This approach outperforms baseline, non-physics-informed methods in terms of both accuracy and interpretability. Numerical experiments using synthetic data demonstrate the ability of the proposed PHNN model to identify and emulate the nonlinear dynamics of the system.
Read the full abstract