Peer Revenue Past citations are no guarantee of future returns

Papers

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

Mathematics

math Watch Mathematics (sign in to use)

Thursday 7 May

Analysis of PDEs

Optimal regularity at the free boundary in one-dimensional first-order mean field games

We establish sharp regularity for the value function, the pressure, and the free boundary in one-dimensional first-order mean field games with power coupling and compactly supported density. Under a standard nondegeneracy assumption on the initial datum, the pressure \(p=m^\theta\) is Lipschitz continuous, the value function \(u\) is \(C^{1,1/2}\), and the two free boundary curves are smooth in time. If the initial pressure is smooth, then both \(p\) and \(u\) are smooth up to the free boundary from inside the positive phase. The proof works in Lagrangian coordinates and, through a singular change of variables, recasts the boundary degeneracy as a removable radial axis in effective dimension \(N=4+2/\theta\), allowing the application of recent estimates for even solutions to elliptic problems with degenerate weights.
Read the full abstract

Optimization and Control · Cross-listed from cs.LG

When Descent Is Too Stable: Event-Triggered Hamiltonian Learning to Optimize

Fixed-budget nonconvex optimization can fail not because local descent is unstable, but because it is too stable: after reaching a nearby stationary point, an optimizer may spend the remaining evaluations refining an uninformative local minimum. We formulate this failure mode as a control problem over optimizer dynamics, where the learner must decide when to descend, when to exploit a promising basin, and when stagnation should trigger movement elsewhere. We introduce SHAPE, a structured adaptive port-Hamiltonian task-family optimizer for event-triggered minima hunting under local information. Starting from gradient-descent dynamics, SHAPE lifts optimization to an augmented phase space $(q, p)$, where the primal state $q$ represents the candidate solution, the cotangent variable $p$ carries directional sensitivity, and a controller $u$ provides processed information from current gradient oracle. Within each stage, a learned Hamiltonian vector field induces structured local descent; across stages, a fixed event clock in the implementation updates ports and memory when local equilibria are detected, with stage-dependent horizons treated in the analysis as a direct generalization. This design preserves a passivity-compatible structure while allowing the same trained policy to use clean, stochastic, or estimated gradient inputs. Experiments on fixed-budget nonconvex optimization tasks show that SHAPE improves best-so-far performance compared with fixed-policy optimizers. These results suggest that adaptive Hamiltonian energy shaping provides a principled mechanism for balancing descent, exploration, and budget allocation in difficult optimization landscapes.
Read the full abstract

Analysis of PDEs

Asymptotic analysis of the energy for a ferroelectric nematic

The variational model for a ferroelectric nematic bears close resemblance to the well-known energy model for micromagnetics. Despite this similarity, the two models operate in fundamentally distinct parameter regimes describing different physics. In this paper we establish that the ferroelectric nematic energy functional $\Gamma$-converges to the energy of a nematic with high elastic anisotropy.
Read the full abstract

Optimization and Control · Cross-listed from cs.LG

A Sharp Finite-Iteration Theory for Asynchronous Categorical Distributional Temporal-Difference Learning

We study finite-iteration behavior of asynchronous categorical distributional temporal-difference methods, covering scalar categorical TD (CTD) in the Cramér geometry and multivariate signed-categorical TD (MTD) in the maximum mean discrepancy (MMD) geometry. We establish high-probability guarantees under i.i.d. sampling and under a continuing Markovian trajectory. For CTD, the Cramér sample complexity to the true return law has leading term $\tilde O(\rho_{\min}^{-1}(1-\gamma)^{-2}\varepsilon^{-2})$ in the i.i.d. and $\tilde O(\mu_{\min}^{-1}(1-\gamma)^{-2}\varepsilon^{-2}+\mu_{\min}^{-1}t_{\mathrm{mix}})$ in the Markovian setting, without using variance reduction or data dropping. For MTD, the MMD sample complexity has leading terms $\tilde O(\rho_{\min}^{-1}(1-\gamma)^{-1-c}\varepsilon^{-2})$ and $\tilde O(\mu_{\min}^{-1}(1-\gamma)^{-1-c}\varepsilon^{-2}+\mu_{\min}^{-1}t_{\mathrm{mix}})$, where $c\in(0,2)$ is the homogeneity exponent of the MMD kernel, and they reduce to the CTD rates when $c=1$. These are, to our knowledge, the first finite-iteration rates for MTD. For undiscounted fixed-horizon policy evaluation, the same analysis applies to fixed-horizon versions of CTD and MTD under i.i.d. and episodic sampling, and the rates match the discounted ones with the effective horizon $(1-\gamma)^{-1}$ replaced by the horizon $H$ in the leading term. Matching minimax lower bounds show that the leading terms, and the implied $1$-Wasserstein rates, are optimal under i.i.d., Markovian, and episodic sampling. Together, these results provide a unified non-asymptotic analysis of asynchronous categorical distributional TD across scalar, multivariate, discounted, and fixed-horizon settings, with rates that are minimax optimal in their leading terms.
Read the full abstract

Numerical Analysis · Cross-listed from cs.LG

Christoffel-DPS: Optimal sensor placement in diffusion posterior sampling for arbitrary distributions

State estimation is a critical task in scientific, engineering and control applications. Since the reliability of reconstructions depends on the number and position of sensors, optimal sensor placement (OSP) is essential in scenarios where measurements are sparse and expensive. Classical OSP approaches rely on Gaussian assumptions and are consequently unable to account for the complex distributions encountered in many real-world systems. Generative-model-based reconstruction using sensor guided diffusion posterior sampling (DPS) has emerged as a promising technique for reconstructing states from highly complex distributions. However, existing sensor-selection methods either require unrealistically many sensors or emulate classical OSP, creating a mismatch between modern recovery models with classical OSP tools motivating the need for fundamentally new ideas towards OSP that match the recent advances made in powerful recovery models. We introduce a distribution-free sensor placement framework based on the Christoffel function: a mathematical formulation of optimal sampling and recovery guarantees for posterior sampling with arbitrary sensors and signal distributions, from which we derive a new OSP strategy with non-asymptotic bounds on the number of sensors needed for recovery. We develop Christoffel-DPS, with offline and online variants, instantiating Christoffel sampling for generative models. Christoffel-DPS outperforms Gaussian OSP baselines and existing generative-model placement methods, validating that distribution-free sensing is both theoretically principled and practically superior. The framework is model-agnostic; we demonstrate its application to a range of unconditional DPS and flow-matching models on structurally non-Gaussian benchmarks, showing the efficacy of Christoffel-DPS in low sensor budget regimes.
Read the full abstract

All authors James Rowbottom, Nick Huang, Carola-Bibiane Schönlieb, Ben Adcock

Optimization and Control

Simplicial Regularizability of the Pseudo-Moment Cone and Carathéodory-Type Atomic Decomposition of Moment Matrices

We study the facial geometry of the homogeneous pseudo-moment cone \(\Sigma_{n,2d}^*\) and its implications for atomic decomposition of moment matrices. For fixed \(d \ge 2\), we show that if a moment matrix is formed by \(O(n^d)\) generically chosen weighted atoms, then its minimal face in the matrix realization of the pseudo-moment cone is simplicial and generated by the planted rank-one atoms. Based on this geometric result, we develop a Carathéodory-type extreme-ray decomposition algorithm for spectrahedral cones and show that, when specialized to the pseudo-moment cone, it yields an efficient atomic decomposition method for generically generated moment matrices in the same regime. A stabilized numerical implementation demonstrates strong recovery performance and suggests that, outside the guaranteed regime, the algorithm may serve as a practical sampler of high-rank extreme rays.
Read the full abstract

Geometric Topology

Intrinsic Linking of 2-complexes in $\mathbb{R}^4$

We produce an infinite family of $2$-complexes that are intrinsically linked when embedded into four dimensions. In particular, we show that any embedding into $\mathbb{R}^4$ of the suspension of a graph containing $K_6$ as a minor contains a non-trivially linked 1 and 2-cycle.
Read the full abstract

All authors Nathan Huber, Ishaan Raghavendra Rao, Hannah Schwartz Joseph, Tanishga Thankaraj Vijay

Statistics Theory

Convergence Rates for Latent Mixing Measures in Infinite Homoscedastic Location-Scale Mixture Models

We study posterior contraction rates for mixing measures in homoscedastic location-scale mixture models with infinitely many components. While posterior convergence at the level of densities is well understood, ensuring convergence of the latent mixing measure is more challenging and has remained an open problem in settings where both location and scale parameters are unknown. We address this by deriving novel lower-bounds that connect the $L^1$ distance between mixture densities to discrepancies, based on the Wasserstein distances and the operator norm, between the underlying mixing measures and scale matrices. Our approach combines the dual formulation of the $W_1$ distance with functional-analytic approximation techniques. This leads to general inequalities, whose strength is determined (i) by the smoothness of the mixture kernel via the rate of decay of its characteristic function, and (ii) by a key lower-bound on the $L^1$ metric involving the operator norm discrepancy between scale parameters. Moreover, a novel PDE inversion condition yields a sharper inequality for important ordinary-smooth cases. We specialize these bounds to popular mixtures based on multivariate Gaussian, Cauchy, and Laplace kernels. As a consequence, we obtain first-of-their-kind contraction rates in the context of Dirichlet process mixtures with an unknown scale parameter shared across components. As a byproduct of our inequalities, we can distinguish the convergence behavior of the location mixing measure from that of the scale parameter across a range of kernel choices, leading to nuanced insights into their respective rates.
Read the full abstract

All authors Nicola Bariletto, Dung Le, Alessandro Rinaldo, Nhat Ho

Mathematical Physics

A Note on the Construction of Trial States for the Dilute Bose Gas

We review how the local particle number cutoff introduced in [11] is used to build trial states for the dilute Bose gas that capture the substantial correlation structure of the ground state in the thermodynamic limit. In particular, we provide a simplified derivation of the Lee-Huang-Yang correction as an upper bound for the ground state energy.
Read the full abstract

Combinatorics

The strong and doubly metric dimensions of Johnson and Kneser graphs

In this paper, the strong and doubly metric dimensions of Johnson and Kneser graphs are considered. The exact value of the strong metric dimension of Johnson graph $J_{n,k}$ is obtained using the well-known results from the literature. The strong metric dimension of Kneser graph $K_{n,k}$ has been obtained for $n \ge 3k-1$. Finally, it has been shown that the doubly metric dimensions of Johnson graph $J_{n,2}$ and Kneser graph $K_{n,2}$ are both $\lceil \frac{2n}{3} \rceil$.
Read the full abstract

Information Theory · Cross-listed from cs.LG

A Unified Measure-Theoretic View of Diffusion, Score-Based, and Flow Matching Generative Models

We survey continuous-time generative modeling methods based on transporting a simple reference distribution to a data distribution via stochastic or deterministic dynamics. We present a unified framework in which diffusion models, score-based generative models, and flow matching are instances of learning a time-dependent vector field that induces a family of marginals $(\rho_t)_{t \in [0,1]}$ governed by continuity and Fokker-Planck equations. Such a unified theory is timely because these methods are converging methodologically, yet fragmented notation and competing derivations continue to obscure their shared structure and the practical tradeoffs governing sampling, stability, and computation. Within this framework, we (i) derive reverse-time sampling for diffusion and score-based models as controlled stochastic dynamics, (ii) show that the probability flow ODE yields identical marginals and connects diffusion to likelihood-based normalizing flows, and (iii) interpret flow matching as direct regression of the velocity field under a chosen interpolation, clarifying when it coincides with or differs from score-based training. We compare objectives, sampling schemes, and discretization errors under unified notation, discuss connections to Schrodinger bridges and entropic optimal transport, and summarize theoretical guarantees and open problems on approximation, stability, and scalability.
Read the full abstract

Information Theory · Cross-listed from stat.ML

How Does Attention Help? Insights from Random Matrices on Signal Recovery from Sequence Models

We study the spectral properties of sample covariance matrices constructed from pooled sequence representations, where token embeddings are drawn from a fixed two-class Gaussian mixture table and pooled via (fixed) attention weights. Working in the high-dimensional regime $d,V,N\to\infty$ with $d/V\to\delta$ and $d/N\to\gamma$, we derive exact characterizations of the limiting eigenvalue distribution, outlier eigenvalues, and eigenvector alignment with the hidden signal. The bulk spectrum follows a non-Marchenko–Pastur law given by the free multiplicative convolution $\kappa(MP_\delta\boxtimes MP_\gamma)$, reflecting the finite vocabulary structure. Signal recovery undergoes two successive BBP-type phase transitions characterized by the scalars: $\delta,\gamma,\alpha=w^{\top} R w$ and $\kappa=\|w\|^2$, where $w$ denotes the attention pooling weights and $R$ the positional correlation matrix. An aftermath of our analysis demonstrates that the optimal attention weights maximizing the signal-to-noise ratio $\alpha/\kappa$ are given by the (normalized) top eigenvector of $R$, and we show (as a particular case of our analysis) that parameter-free causal self-attention with $\tau/d$ score scaling yields deterministic harmonic weights that improve signal recovery over mean pooling whenever early tokens carry more signal. Extensive simulations confirm sharp agreement between theory and finite-dimensional experiments.
Read the full abstract

Geometric Topology

Plumbed 3-Manifolds and Neumann Moves

We give a constructive proof that every weakly negative definite plumbing tree can be transformed into a negative definite one by a finite sequence of Neumann moves. The argument combines Neumann's plumbing calculus with the diagonalization algorithm of Duchon, Eisenbud, and Neumann, which extracts the eigenvalues of the framing matrix directly from the combinatorics of the tree. We show that any positive eigenvalues are supported on linear branches and can be eliminated systematically via controlled applications of Neumann moves. This provides an explicit algorithm reducing weakly negative definite plumbing trees to negative definite ones.
Read the full abstract

Optimization and Control · Cross-listed from cs.LG

A Rod Flow Model for Adam at the Edge of Stability

Cohen et al. (arXiv:2207.14484) observed that adaptive gradient methods such as Adam operate at the edge of stability. While there has been significant work on continuous-time modeling of gradient descent at the edge of stability, extending these models to momentum methods remains underdeveloped. In the gradient descent setting, Regis et al. (arXiv:2602.01480) introduced rod flow, which models consecutive iterates as an extended one-dimensional object – a "rod." Here we extend rod flow to Adam by working in the joint phase space of parameters and first moment $(w, m)$ and treating the second moment $\nu$ as a smooth auxiliary variable. We also develop rod flows for heavy ball momentum, Nesterov momentum, and scalar and per-component versions of RMSProp, Adam, and NAdam. For all eight optimizers, we empirically evaluate rod flow on representative machine learning architectures, where it tracks the discrete iterates through the edge-of-stability regime significantly more accurately than the corresponding stable flow.
Read the full abstract

Mathematical Physics · Cross-listed from hep-th

The Conformal Grassmannian: A Symplectic Bi-Grassmannian for $CFT_ 4$ Correlators

We introduce a formalism for conformal field theory in four dimensions: a symplectic bi-Grassmannian representation of CFT$_4$ Wightman correlators. Working in Klein space with off-shell spinor-helicity variables, we show that correlators of $\Delta = 2$ scalars and symmetric-traceless conserved currents are encoded by integrals over a pair of $n$-planes in a $2n$-dimensional symplectic vector space. These planes are constrained to be mutually symplectically orthogonal and aligned with the external kinematics. Conformal invariance, momentum conservation, and little-group covariance all follow geometrically from this structure. We derive all two- and three-point functions involving scalars, fermions, conserved currents, and stress tensors. As a non-trivial test, we show that the construction reproduces the full set of independent conformally invariant structures of $\langle JJJ\rangle$ and $\langle TTT\rangle$ in CFT$_4$. The resulting expressions are considerably more compact than their momentum-space counterparts. They also make manifest the double copy between Yang–Mills $\langle JJJ \rangle$ and Einstein-gravity $\langle TTT \rangle$. We further present a helicity-basis reformulation that makes the GL(1,R) and SL(2,R) weights of individual helicity components explicit. This basis also provides a natural starting point for a twistor-space formulation of the correlators.
Read the full abstract

Information Theory

A Framework of Variable-Length Source Encryption using Mutual Information Security Criterion: Universal Coding, Strong Converse Theorem

In this paper we consider the variable-length lossless source coding for discrete memoryless sources. We proposes a new encryption framework for securely transmitting codewords over a noiseless channel. The proposed source encryption framework is based on the secure communication framework of the Shannon cipher system. In the proposed framework, we use the mutual information as a measure of information leakage to an adversary. We establish the necessary and sufficient condition for secure communication under the condition that the information leakage is upper bounded by a constant $\delta\in (0,\infty)$, thereby providing a complete solution to the problem. We also show that the obtained necessary and sufficient condition does not depend on the constant $\delta \in (0,\infty)$, demonstrating that we have the strong converse coding theorem for the proposed framework of source encryption. We further prove the existence of encryption/decryption schemes, which are universal in the sense that they work effectively for any distributions of the plain text and those of the key used for the encryption.
Read the full abstract

2605.06802 · 7 May · 7 pages, 5 figures. This is an extended version of the paper submitted for presentation in ISITA 2026, which will be held at Okinawa, Japan form November 1 to 4 in 2026. The submitted paper has several errors. We correct those errors in this arXiv paper. arXiv admin note: substantial text overlap with arXiv:2605.04720

Information Theory

Cryptographic and Information-theoretic Security Capacities for General Arbitrarily Varying Wiretap Channels

We compare the strong secrecy capacities of Arbitrarily Varying Wiretap Channels (AVWCs) and General Arbitrarily Varying Wiretap Channels (GAVWCs) with their capacities under semantic secrecy constraint and other equivalent cryptographic secrecy constraints. It turns out that the average error and strong secrecy capacity of an AVWC is always equal to its maximal error and semantic secrecy capacity. However, this equivalence does not hold for all general communication systems, and we prove this by a counterexample. We also show that, for the GAVWC, semantic security and the other cryptographic security measures considered achieve the same capacity values. Finally, we bound the gap between the strong secrecy capacity and the semantic secrecy capacity for the GAVWC. The gap vanishes if the choice of the jammer is sub-double-exponential with respect to the block length n, which gives a sufficient condition for the strong and semantic secrecy capacities to be equal for GAVWCs.
Read the full abstract

All authors Holger Boche, Ning Cai, Yiqi Chen, Marc Geitz

Spectral Theory

On the Spectral Region of 4-Cycle Stochastic Matrices

We study the spectrum of 4-cycle row-stochastic matrices. For real eigenvalues the spectral region is [-1,1]. For nonreal eigenvalues a+ib we derive necessary conditions in terms of the real and imaginary parts, including the inequality a+|b| <= 1 and the condition (b^2+a^2+a)^2+2a^2-b^2 >= 0. We also prove conversely that every point in the corresponding interior region occurs as an eigenvalue of a 4-cycle matrix. The proof is organized through a reformulation of the characteristic equation, an argument parametrization, a convex-analytic criterion, and explicit boundary constructions. Hence, the spectral region for the 4-cycle row-stochastic matrices is exactly and explicitly determined.
Read the full abstract

All authors Brando Vagenende, Brecht F. Verbeken, Andres Algaba, Marie-Anne Guerry

Group Theory

Forest Diagrams and Lengths for the Generalised Thompson's Group $F(n)$

We extend the concept of two-way forest diagrams, introduced by Belk and Brown in 2003, to represent elements of $F(n)$ as a pair of infinite, bounded $n$-ary forests together with an order-preserving bijection of the leaves. This representation allows us to develop an alternative way to compute the length of an element of $F(n)$, distinct from the formula established by Fordham and Cleary in 2009. As an application of our length formula, we re-prove the existence of dead end elements in $F(n)$ and show that their depth is always two, first proved by Wladis in 2009.
Read the full abstract

Numerical Analysis

Error estimation for numerical approximations of ODEs via composition techniques. Part II: BDF methods

Integration of Ordinary Differential Equations (ODEs) using Backward Difference formula (BDF) methods with p backward steps achieves order p accuracy if specific conditions are met. This work extends the composition technique with complex coefficients to the implicit BDF schemes, increasing the approximation order by one without additional backward points. The imaginary part of the composed flow provides an error estimate of order p + 1. Linear stability analysis reveals that the composed schemes break the Dahlquist barrier, achieving stability up to order eight. The computational performance of the composed flow outperforms BDF schemes when using the same number of backward points, allowing for higher accuracy with lower CPU time. For non-uniform meshes, the ratio of consecutive time steps, which influences stability, appears as a parameter in the roots of algebraic equations relative to the composed flow. Having a complex root with a real positive part implies a lower bound to this ratio depending on the order. For example, the bound is 0.4506 for order three and 0.6806 for order four. Numerical tests demonstrate the effectiveness of this technique in improving the accuracy and stability compared to BDF methods.
Read the full abstract

Combinatorics

Blow-up trick in Combinatorics

Blow-up in graph theory is a procedure in which each vertex is replaced by copies of itself, and two copies are adjacent if and only if the original vertices are adjacent. In this paper, we extend the concept of graph blow-up to a more general combinatorial context and discuss its potential applications.
Read the full abstract

Spectral Theory

Necessary Conditions for Single-Critical-Point Higher-Order Szegő Sum Rules in OPUC

We prove the necessity part of the higher-order Szegő theorem on the unit circle for the single-critical-point weights $H_m(e^{i\theta})=(1-\cos\theta)^m$, $m\ge1$. If $\{\alpha_n\}_{n\ge0}$ are the Verblunsky coefficients of a nontrivial probability measure $d\mu=w(\theta)d\theta/(2\pi)+d\mu_{\mathrm s}$, then the weighted Szegő condition $\int_0^{2\pi} (1-\cos\theta)^m\log w(\theta)\frac{d\theta}{2\pi}>-\infty$ implies $\Delta^m\alpha\in\ell^2, \,\, \alpha\in\ell^{2m+2}.$ The proof uses a finite-volume version of Yan's higher-order sum rule. The quadratic part yields the $m$-th difference energy, and the logarithmic tail yields the $\ell^{2m+2}$-control. The non-sign-definite critical terms are treated in two steps. First, the quartic principal critical block is isolated using the Yan quotient-algebra normal representative and shown to have a positive semidefinite Gram representation. Second, the remaining non-principal critical terms are controlled by the diagonal-vanishing property $\mathcal Y_{k,\mathrm{crit}}^{(m)} \in \mathfrak I_k^{\,m+1-k}, \,\, 2\le k\le m,$ together with the Breuer–Simon–Zeitouni normal form, discrete interpolation, and Young's inequality. These estimates yield a uniform finite-volume coercive bound, from which the necessity theorem follows for all $m\ge1$.
Read the full abstract

Information Theory

Near-field Channel Estimation for XL-RIS-aided mmWave MIMO Systems

Extremely large-scale reconfigurable intelligent surfaces (XL-RISs) have emerged as a promising technology for millimeter-wave (mmWave) communications. However, the exceedingly large aperture of XL-RISs renders the RIS-user links likely to operate in the near-field region, where the conventional planar-wave assumption and angular-domain sparse representation become invalid, thus making channel estimation significantly more challenging. In this paper, we investigate cascaded channel estimation for an XL-RIS-aided multi-user multiple-input multiple-output (MU-MIMO) system, in which the BS-RIS channel is modeled in the far field, while the RIS-user channels exhibit near-field spherical-wave characteristics. To tackle the resulting hybrid-field estimation problem, we propose a low-overhead two-stage channel estimation scheme by jointly exploiting the common BS-RIS link shared by all users and the polar-domain sparsity of the RIS-user channels. Specifically, the multi-antenna users are firstly decomposed into multiple virtual single-antenna users, based on which the common BS-RIS parameters are extracted from a typical virtual user and the RIS-user channels are initialized via compensated polar-domain sparse recovery. Then, an alternating least-squares refinement procedure is developed to jointly improve the common BS-RIS operator and the user-specific RIS-side channels. Simulation results show that the proposed scheme achieves competitive channel estimation performance with substantially reduced pilot overhead compared with the existing near-field benchmarks.
Read the full abstract

All authors Erkang Dong, Taihao Zhang, Cunhua Pan, Hong Ren, Jiangzhou Wang

Rings and Algebras

Mean weak length

We introduce a weak version of the classical length function, termed the weak length function, defined on subsets of $R$-modules relative to the ambient $R$-modules over a unital ring $R$. We further consider the concept of mean weak length for each $R{\Gamma}$-module relative to a bigger $R{\Gamma}$-module associated with an amenable group ${\Gamma}$. Under an appropriate upgrading condition together with certain mild assumptions, we establish that the mean weak length satisfies an addition formula with respect to short exact sequences. This result has three applications. First, we provide a purely algebraic proof of the additivity of algebraic entropy, which is a property originally established via topological entropy methods. Second, within our unified framework, we give an alternative and conceptual proof of the additivity of mean length, previously obtained by Li-Liang and Virili using different approaches. Third, we recover a particular case of amenable extension for Sylvester rank functions in the spirit of Jiang-Li.
Read the full abstract

Algebraic Geometry

Rational homology disk degenerations of elliptic surfaces

In this paper, a $\mathbb{Q}$HD singularity is a weighted homogeneous normal surface singularity admitting a rational homology disk ($\mathbb{Q}$HD) smoothing. These singularities are rational but often not log canonical. We classify all $\mathbb{Q}$HD degenerations of nonsingular projective elliptic surfaces, extending Kawamata's classification of the case with only Wahl singularities (i.e., log terminal $\mathbb{Q}$HD singularities). We also realize all $\mathbb{Q}$HD degenerations of Dolgachev surfaces $D_{a,b}$ with one $\mathbb{Q}$HD singularity, for every pair of integers $a,b$. For each such degeneration, we construct a minimal semi log canonical (slc) birational model via a Seifert partial resolution in the sense of Wahl followed by semistable flips. Finally, we prove that these minimal slc models are unobstructed and deform to the recent degenerations of Dolgachev surfaces constructed by D. Lee and Y. Lee.
Read the full abstract

Mathematical Physics · Cross-listed from cond-mat.str-el

Pro-Tensor Network

We introduce the pro-tensor network, a categorification of the tensor network, as a fully rigorous yet graphically transparent framework for studying the collection of many many-body theories, which we dub many-many-body theory. We provide a comprehensive toolbox for the graphical calculations using pro-tensor networks. As applications, we recover the Levin-Wen model as a "uniform" pro-tensor network and generalize a result of Kitaev and Kong by characterizing particles as modules over promonads. One can also interpret the string-net pro-tensor network as the space of symmetric tensor networks, thus our framework also applies to the study of generalized symmetry and topological holography. Notably, our generalization dispenses with the assumptions of semisimplicity, finiteness, and rigidity, potentially facilitating the exploration of many-body physics beyond these constraints.
Read the full abstract

All authors Gen Yue, Ansi Bai, Linqian Wu, Tian Lan

Optimization and Control · Cross-listed from cs.LG

Why Global LLM Leaderboards Are Misleading: Small Portfolios for Heterogeneous Supervised ML

Ranking LLMs via pairwise human feedback underpins current leaderboards for open-ended tasks, such as creative writing and problem-solving. We analyze ~89K comparisons in 116 languages from 52 LLMs from Arena, and show that the best-fit global Bradley-Terry (BT) ranking is misleading. Nearly 2/3 of the decisive votes cancel out, and even the top 50 models according to the global BT ranking are statistically indistinguishable (pairwise win probabilities are at most 0.53 within the top 50 models). We trace this failure to strong, structured heterogeneity of opinions across language, task, and time. Moreover, we find an important characteristic - *language* plays a key role. Grouping by language (and families) increases the agreement of votes massively, resulting in two orders of magnitude higher spread in the ELO scores (i.e., very consistent rankings). What appears as global noise is in fact a mixture of coherent but conflicting subpopulations. To address such heterogeneity in supervised machine learning, we introduce the framework of $(\lambda, \nu)$-portfolios, which are small sets of models that achieve a prediction error at most $\lambda$, "covering" at least a $\nu$ fraction of users. We formulate this as a variant of the set cover problem and provide guarantees using the VC dimension of the underlying set system. On the Arena data, our algorithms recover just 5 distinct BT rankings that cover over 96% of votes at a modest $\lambda$, compared to the 21% coverage by the global ranking. We also provide a portfolio of 6 LLMs that cover twice as many votes as the top-6 LLMs from a global ranking. We further construct portfolios for a classification problem on the COMPAS dataset using an ensemble of fairness-regularized classification models and show that these portfolios can be used to detect blind spots in the data, which might be of independent interest to policymakers.
Read the full abstract

All authors Jai Moondra, Ayela Chughtai, Bhargavi Lanka, Swati Gupta

Optimization and Control · Cross-listed from cs.LG

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less

Optimizers play an important role in both pretraining and finetuning stages when training large language models (LLMs). In this paper, we present an observation that full finetuning with the same optimizer as in pretraining achieves a better learning-forgetting tradeoff, i.e., forgetting less while achieving the same or better performance on the new task, than other optimizers and, possibly surprisingly, LoRA, during the supervised finetuning (SFT) stage. We term this phenomenon optimizer-model consistency. To better understand it, through controlled experiments and theoretical analysis, we show that: 1) optimizers can shape the models by having regularization effects on the activations, leading to different landscapes around the pretrained checkpoints; 2) in response to this regularization effect, the weight update in SFT should follow some specific structures to lower forgetting of the knowledge learned in pretraining, which can be obtained by using the same optimizer. Moreover, we specifically compare Muon and AdamW when they are employed throughout the pretraining and SFT stages and find that Muon performs worse when finetuned for reasoning tasks. With a synthetic language modeling experiment, we demonstrate that this can come from Muon's strong tendency towards rote memorization, which may hurt pattern acquisition with a small amount of data, as for SFT.
Read the full abstract

Dynamical Systems

Integrable perturbations of polynomial Hamiltonian systems

We consider a Hamiltonian system on the symplectic space $({\mathbb{R}}^{2n}, dy\wedge dx)$ with a real-analytic Hamiltonian $H : {\mathbb{R}}^{2n}\to {\mathbb{R}}$. We assume that the system has a non-degenerate equilibrium position at the origin. Under some nonresonance assumptions we prove the following. For any positive integer $M$ there exists a real-analytic function $F:{\mathbb{R}}^{2n}\to{\mathbb{R}}$ such that (1) $F = O\big( (|x|+|y|)^{M+1} \big)$ at the origin, (2) the system with Hamiltonian $H+F$ is completely integrable in ${\mathbb{R}}^{2n}$.
Read the full abstract

Functional Analysis

On the plasticity of the unit spheres of $\ell_1$, $\ell_{\infty}$, $c$, and Hilbert spaces

This paper demonstrates the expand-contract plasticity of the unit spheres of $\ell_1$, $\ell_{\infty}$, and $c$. Furthermore, it establishes the strong plasticity of the unit spheres of Hilbert spaces.