Peer Revenue Play credits. Real papers.

Papers

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

Mathematics

math Watch Mathematics (sign in to use)

Saturday 11 April

Operator Algebras

Pureness of Certain Crossed Product C*-Algebras

We establish comparison and divisibility properties for crossed product C*-algebras arising from automorphisms of algebras C (X, D) which lie over minimal homeomorphisms, from actions of compact groups which have finite Rokhlin dimension with commuting towers, and from actions of compact groups which have the restricted tracial Rokhlin property with comparison. We deduce that these crossed products we consider are pure, and conclude they have stable rank one, and in certain cases have real rank zero. We give examples in which these properties do not follow from previous results, in the case of C (X, D) due to the lack of Z-stability of D, the underlying topological spaces not being finite dimensional, or both.
Read the full abstract

All authors Dawn Archey, Julian Buck, Javad Mohammadkarimi, N. Christopher Phillips, Apurva Seth

Analysis of PDEs

Sharp hypocoercive convergence estimates for underdamped Langevin dynamics via the modified $L^2$ method

In this note, we consider the underdamped Langevin dynamics with invariant measure $\mu(\mathrm{d}x\,\mathrm{d}v) \propto e^{-U(x)-|v|^2/2}\,\mathrm{d}x\,\mathrm{d}v$. Assume that the position marginal $\mu_x(\mathrm{d}x)\propto e^{-U(x)}\,\mathrm{d}x$ satisfies a Poincaré inequality with constant $m>0$, and that $\nabla^2 U\ge -K\,\mathrm{Id}$ for some $K\ge 0$. We revisit the modified $L^2$ method of Dolbeault–Mouhot–Schmeiser, employing a shifted corrector \begin{equation*} A_\alpha=(\alpha- L_{\mathrm o})^{-1}(L_a\Pi_v)^*, \qquad \alpha\ge0, \end{equation*} where ${L}_{\mathrm{o}}=\Delta_x-\nabla U\cdot\nabla_x$ is the overdamped generator, ${L}_a$ is the generator of the Hamiltonian flow, and $\Pi_v$ denotes averaging over the velocity variable. We establish an explicit hypocoercive $L^2$-convergence rate $\Lambda_{\alpha,\gamma}$ for every shift $\alpha\ge0$ and friction coefficient $\gamma>0$, and show that, for each fixed $\gamma$, the rate is maximized at $\alpha=0$. Optimizing further over $\gamma$ gives \begin{equation*} \Lambda_{0,\gamma_*} = \frac{\sqrt m} {2\left(2+\sqrt{2+\frac{2K}{m}} +\sqrt{6+\frac{2K}{m}}\right)}\,, \qquad \text{at} \quad \gamma_*=\sqrt{6m+2K}. \end{equation*} For convex $U$, this recovers the optimal $O(\sqrt m)$ rate.
Read the full abstract

Probability

Ergodicity of the voter model with dynamic anti-voter bonds

The voter model with anti-voter bonds is a variant of the classical voter model in which the edges of the underlying graph are assigned signs. At each update, a voter chooses a neighbour according to a transition kernel; interactions across a positive edge follow the usual voter dynamics, so that a site adopts the current opinion of its chosen neighbour, whereas interactions across a negative edge lead to the adoption of the opposite opinion. In this work, we introduce a new variant in which the edge signs evolve dynamically according to dynamical percolation with density parameter $p \in (0,1)$ and speed $\mathsf{v} \in (0,\infty)$, where the two states of the process represent positive and negative edges. This defines a joint spin-bond Markov process. Following Liggett's notion of ergodicity, we prove that this process is ergodic on any simple graph with countably many vertices, with an arbitrary transition kernel of adoption rates and for all choices of the parameters of the edge dynamics.
Read the full abstract

Analysis of PDEs

On the classification of solutions to a class of $N$-Liouville equations in $\mathbb{R}^N$

Given $N\geq 2$ and $\alpha>-1$, we consider the following weighted Liouville-type equation involving the $N$-Laplacian: \begin{equation*} \left\{ \begin{aligned} -& \Delta_N u = |x|^{N\alpha} e^u \quad \text{ in } \mathbb{R}^N && , \\ & \int_{\mathbb{R}^N} |x|^{N\alpha} e^u \, dx < + \infty\,. &&\end{aligned} \right. \end{equation*} Solutions have been completely classified when $N=2$ via complex analysis, and when $\alpha=0$ using Pohozaev identities and an isoperimetric argument. In this paper, we first devise a $P$-function approach to the classification result for all $\alpha>-1$ when $N=2$. Since it is not based on complex analysis, this alternative and more PDE-oriented approach naturally extends to $N\geq 3$ by providing the classification for any $-1<\alpha\leq 0$. In particular, the explicit radial solutions are the unique ones for $-1<\alpha\leq0$ but become degenerate for special values $\alpha_k>0$, a hint that non-radial solutions might arise for $\alpha>0$ as it happens when $N=2$.
Read the full abstract

Operator Algebras

$C(SO_q(4)/SO_q(2))$ as a Groupoid $C^*$-algebra

In this paper, we prove that $C(SO_q(4)/SO_q(2))$ is isomorphic to the $C^*$-algebra of the tight groupoid $\mathcal{G}_{\mathrm{tight}}$ associated with the inverse semigroup generated by the standard generators of its classical limit $C(SO_0(4)/SO_0(2))$. We show that all four orbits of the unit space $\mathcal{G}_{\mathrm{tight}}^{(0)}$ under the natural action of $\mathcal{G}_{\mathrm{tight}}$ are locally closed, and that the associated isotropy groups are isomorphic to $\mathbb{Z}$. Consequently, every irreducible representation of $C^*(\mathcal{G}_{\mathrm{tight}})$ is induced from an irreducible representation of $C^*(\mathbb{Z})$, which are parametrized by $\mathbb{T}$. In this way, we obtain four families of irreducible representations parametrized by $\mathbb{T}$, and we explicitly construct their equivalence with the corresponding Soibelman irreducible representations of $C(SO_q(4)/SO_q(2))$.
Read the full abstract

Probability

The directed landscape in half-space

We prove that two half-space models in the KPZ universality class, exponential last-passage percolation and a family of Poisson-avoiding metrics generalizing colored TASEP, converge to a common scaling limit. This scaling limit is the directed landscape in half-space, a random directed metric in the half-plane indexed by a parameter which determines the strength of the boundary interaction. As part of our analysis, we characterize the half-space directed landscape in terms of the half-space KPZ fixed point, and prove convergence of geodesics. We also give an explicit construction of joint stationary measures (or horizons) in half-space for the log-gamma polymer, the KPZ equation, exponential and geometric last passage percolation, and the directed landscape itself.
Read the full abstract

Optimization and Control · Cross-listed from stat.ME

Toward Exact Convergence in Byzantine-Robust Decentralized Learning: A Statistical Identification Approach

To defend against Byzantine attacks in decentralized learning, most existing methods rely on robust aggregation rules to mitigate the influence of malicious machines. However, these strategies inherently introduce bias, leading to inexact convergence with non-vanishing steady-state errors. In this paper, we propose a strategic shift from passive aggregation to active identification by introducing the Decentralized Rescaled Stochastic Gradient Descent with Byzantine Machine Identification (DRSGD-ByMI) framework. The core of our approach is an identification-based “detect-then-optimize” pipeline, where a p-value-free detection procedure is developed to accurately prune malicious nodes from the network. By leveraging sample-splitting score statistics, this identification mechanism achieves false discovery rate control without requiring restrictive distributional assumptions. We theoretically demonstrate that this precise identification allows the decentralized network to recover sufficient connectivity among the normal nodes, thereby enabling DRSGD-ByMI to match, even in the presence of Byzantine machines, the same order-optimal convergence rate as standard decentralized stochastic first-order methods. Numerical experiments validate our theoretical results and demonstrate the effectiveness of DRSGD-ByMI for decentralized robust learning problems.
Read the full abstract

All authors Siyuan Zhang, Chengde Qian, Xin Liu, Changliang Zou

Optimization and Control

A simulation-optimization approach for fractional, profitability-oriented inventory control under service-level type constraints

Managing stock efficiently remains a core issue in modern logistics, where companies must reconcile cost efficiency with dependable service despite unpredictable market conditions. Conventional models often overlook the direct connection between investment in inventory and overall financial performance. This study introduces a data-driven decision framework that combines stochastic simulations with a profit-oriented optimization routine to enhance decision-making under uncertainty. The simulation stage generates performance estimates across multiple operating scenarios, providing realistic data on expenditures, revenues, and service reliability. These outcomes inform a fractional optimization process that searches for policies yielding the highest financial returns while maintaining required availability levels. The algorithm iteratively refines parameter values through feedback between simulated outcomes and optimization results, ensuring adaptability to dynamic enterprise systems. Computational experiments using representative business settings confirm that this approach improves both service consistency and financial yield. Overall, the framework demonstrates a practical, data-driven path for firms seeking to align operational responsiveness with sustainable profitability.
Read the full abstract

Analysis of PDEs

Long Time Behavior of Stochastic Thin Film Equation

We consider the stochastic thin-film equation with linear deterministic and stochastic Itô perturbations. The existence of nonnegative weak martingale solutions on the semi-axis is established, and their asymptotic behavior as $t \to \infty$ is investigated. It is shown that in square mean the $L^\infty$ norm of the solution converges to the spatial mean value of the initial condition, multiplied by a random factor similar to a geometric Wiener process.
Read the full abstract

All authors Oleksiy Kapustyan, Olha Martynyuk, Oleksandr Misiats, Oleksandr Stanzhytskyi

Differential Geometry

On weak formulations of (super) Ricci flows

We present two characterizations of smooth compact Ricci flow solutions solely in terms of metrics and measures (one of them only works under positive scalar curvature along the flow); thus, provide weak formulations that are generalized to the singular setting in a straightforward manner. These formulations are achieved by weakly formulating super Ricci flows and imposing a saturation condition (solely in terms of metric and measure) to ensure the super Ricci flow inequality is an equality.
Read the full abstract

Probability · Cross-listed from cond-mat.stat-mech

Universality and ambiguity in extremes of anomalous diffusion

Many biophysical processes begin when the fastest searcher finds a target out of many random searchers, which is called an extreme or fastest first passage time (fFPT). In some models, (i) the fFPT vanishes logarithmically as the number of searchers grows, and (ii) the fFPT can be faster for subdiffusive search compared to normal diffusion. Though mathematically rigorous, the relevance of (i) and (ii) to actual physical systems is suspect since their derivations involve searchers which move with unbounded speed. Indeed, we previously proved that the fFPT for searchers with bounded speed converges exponentially to a strictly positive minimal search time as the number of searchers grows. In this paper, we study fFPTs for a broad class of anomalous and normal diffusion models with bounded or unbounded speed. These models include scaled Brownian motion, Riemann-Liouville fractional Brownian motion, and fractional Brownian motion. For all of these models, we show that the fFPT decays logarithmically in the number of searchers and that subdiffusion can be faster than normal diffusion (we further show that superdiffusion can be slower than normal diffusion). In this sense, features (i) and (ii) are rather universal. On the other hand, we show that the parameter regimes in which (i) and (ii) are valid depend on the particulars of the individual model, and thus ambiguities remain in the relevance of these features to specific physical systems.
Read the full abstract

Functional Analysis

A generalization of the inverse mapping theorem in infinite dimensions

We present a generalization of the inverse mapping theorem, where variations of a weaker non-expansiveness property (referred to as property ${\sf A}$) replace the key $\mathsf{C}^1$ condition. We also obtain inverse mapping theorems that can be applied to non-smooth maps. Also as a by-product of the generalized inverse mapping theorem, we prove generalizations of the implicit function theorem and existence and uniqueness theorem of abstract PDE systems as well.
Read the full abstract

Optimization and Control · Cross-listed from cs.RO

GPU-Accelerated Continuous-Time Successive Convexification for Contact-Implicit Legged Locomotion

Contact-implicit trajectory optimization (CITO) enables the automatic discovery of contact sequences, but most methods rely on fine time discretization to capture all contact events accurately, which increases problem size and runtime while tying solution quality to grid resolution. We extend the recently proposed sequential convex programming (SCP) approach for trajectory optimization, continuous-time successive convexification (ct-SCvx), to CITO by introducing integral cross-complementarity constraints, which eliminate the risk of missing contact events between discretization nodes while preserving the flexibility of contact mode changes. The resulting framework, contact-implicit successive convexification (ci-SCvx), models full multibody dynamics in maximal coordinates, including stick-slip friction and partially elastic impacts. To handle complementarity constraints, we embed a backtracking homotopy scheme within SCP for reliable convergence. We implement this framework in a stand-alone Python software, leveraging JAX for GPU acceleration and a custom canonical-form parser for the convex subproblems of SCP to avoid the overhead of general-purpose modeling tools such as CVXPY. We demonstrate ci-SCvx on diverse legged-locomotion tasks. In particular, we validate the approach in MuJoCo with the Gymnasium HalfCheetah model against the MuJoCo MPC baseline, showing that a tracking simulation with the optimized torque profiles from ci-SCvx produces physically consistent trajectories with lesser energy consumption. We also show that the resulting software achieves faster solve times than existing state-of-the-art SCP implementations by over an order of magnitude, thereby demonstrating a practically important contribution to scalable real-time trajectory optimization.
Read the full abstract

Information Theory

Deep Reinforcement Learning for Cognitive Time-Division Joint SAR and Secure Communications

Synthetic aperture radar (SAR) imaging can be exploited to enhance wireless communication performance through high-precision environmental awareness. However, integrating sensing and communication functionalities in such wideband systems remains challenging, motivating the development of a joint SAR and communication (JSARC) framework. We propose a dynamic time-division JSARC (TD-JSARC) framework for secure aerial communications that is relevant for critical scenarios, such as surveillance or post-disaster communication, where conventional localization of mobile adversaries often fails. In particular, we consider a secure downlink communication scenario where an aerial base station (ABS) serves a ground user (UE) in the presence of a ground-moving eavesdropper. To detect and track the eavesdropper, the ABS uses cognitive SAR along-track interferometry (ATI) to estimate its position and velocity. Based on these estimates, the ABS applies adaptive beamforming and artificial-noise jamming to enhance secrecy. To this end, we jointly optimize the time and power allocation to maximize the worst-case secrecy rate, while satisfying both SAR and communication constraints. Using the estimated eavesdropper trajectory, we formulate the problem as a Markov decision process (MDP) and solve it via deep reinforcement learning (DRL). Simulation results show that the proposed learning-based approach outperforms both learning and non-learning baseline schemes employing equal-aperture and random time allocation. The proposed method also generalizes well to previously unseen eavesdropper motion patterns.
Read the full abstract

All authors Mohamed-Amine Lahmeri, Ata Khalili, Yujiao Liu, Anke Schmeink, Robert Schober

Classical Analysis and ODEs

Volterra periodic chain

In this paper, the inverse spectral problem is applied to the integration of a periodic Volterra chain. A generalization of the Lagrange interpolation formula has been made.

Combinatorics

A Recursive Characterization of Laplacian Spectral Radii of Trees with Bounded Maximum Degree

For a positive integer $r$ and a real number $\alpha\ge 2$, let $\mathscr L_r(\alpha)$ be the set of positive real numbers containing $\alpha-1$ and closed under the following operation: if $q_1,\ldots,q_s\in\mathscr L_r(\alpha)$, where $1\leq s\leq r-1$, and $q=\alpha-1-s-\sum\limits_{i=1}^{s}q_i^{-1}>0$, then $q\in\mathscr L_r(\alpha)$. We prove that there exists a tree $T$ with $\Delta(T)\leq r$ and Laplacian spectral radius $\mu(T)=\alpha$ if and only if $(\alpha-1)^{-1}\in\mathscr L_r(\alpha)$. Consequently, the Laplacian spectral radii of nontrivial trees are precisely the real numbers $\alpha\geq2$ satisfying $(\alpha-1)^{-1}\in\mathscr L_{\lfloor\alpha\rfloor-1}(\alpha)$. As an application, we prove that, for integers $k\geq2$ and $r\geq2$, a tree $T$ with $\mu(T)=k^2$ and $\Delta(T)=r$ exists if and only if $(k-1)^2+2\leq r\leq k^2-1$.
Read the full abstract

Geometric Topology

Kauffman bracket skein module of the connected sum of two solid tori

We determine the structure of the Kauffman bracket skein module of the connected sum of two genus one handlebodies over the ring of Laurent polynomials $\mathbb Z[q^{\pm 1}]$, thereby proving a conjecture posed by the first and third authors. Our results lay the groundwork for computing the Kauffman bracket skein module of arbitrary connected sums over the ring $\mathbb Z[q^{\pm 1}]$.
Read the full abstract

Optimization and Control · Cross-listed from cs.LG

LoDAdaC: a unified local training-based decentralized framework with adaptive gradients and compressed communication

In the decentralized distributed learning, achieving fast convergence and low communication cost is essential for scalability and high efficiency. Adaptive gradient methods, such as Adam, have demonstrated strong practical performance in deep learning and centralized distributed settings. However, their convergence properties remain largely unexplored in decentralized settings involving multiple local training steps, such as federated learning. To address this limitation, we propose LoDAdaC, a unified multiple Local Training (MLT) Decentralized framework with Adam-type updates and Compressed communication (CC). LoDAdaC accommodates a broad class of optimizers for its local adaptive updates, including AMSGrad, Adam, and AdaGrad; it is compatible with standard (possibly biased) compressors such as low-bit quantization and sparsification. MLT and CC enable LoDAdaC to achieve multiplied reduction of communication cost, while the technique of adaptive updates enables fast convergence. We rigorously prove the combined advantage through complexity analysis. In addition, experiments on image classification and GPT-style language model training validate our theoretical findings and show that LoDAdaC significantly outperforms existing decentralized algorithms in terms of convergence speed and communication efficiency.
Read the full abstract

All authors Wei Liu, Anweshit Panda, Ujwal Pandey, Haven Cook, George M. Slota, Naigang Wang, Jie Chen, Yangyang Xu

Number Theory

Low moments of random multiplicative functions twisted by Fourier coefficients of modular forms

Let $\lambda(n)$ denote the Fourier coefficients of a fixed modular form and $h(n)$ a Steinhaus or Rademacher random multiplicative function. In this paper, we determine the order of magnitude of \[ \E|\sum_{n \leq x} h(n)\lambda(n)|^{2q} \] for real $x$, $q$ with $0 \leq q \leq 1$.
Read the full abstract

Logic

On Cohesive Products of Fields

We develop the foundations of effective ultraproducts of fields and their Galois groups using the methods of computability theory. These computability-theoretic analogs of ultraproducts are called cohesive products, since the role of an ultrafilter is played by a cohesive set. A set of natural numbers is cohesive if it is infinite and cannot be partitioned into two infinite subsets by any computably enumerable set. In particular, we investigate the way cohesive products interact with field extensions with emphasis on both finite and infinite Galois extensions, and the associated Galois groups. We study the first-order theories and definability of cohesive powers of number fields, and characterize the infinite Galois groups of cohesive powers for a large class of infinite Galois extensions. Finally, we introduce hyper-automorphisms, which are automorphisms of a cohesive power that respect non-standard field operations, and give a complete description of the hyper-automorphism groups of cohesive powers of a large class of computable Galois extensions, and use them to describe the classical infinite Galois groups of such fields.
Read the full abstract

All authors Rumen Dimitrov, Valentina Harizanov, Henry J. Klatt, Keshav Srinivasan

Algebraic Geometry

Descendant and Fourier-Mukai equivalences for simple flops

For a simple flop $X\dashrightarrow X'$, we construct a correspondence between genus $0$ descendant Gromov-Witten theories of $X$ and $X'$. We show that the Fourier-Mukai equivalence induced by $X\dashrightarrow X'$ is compatible, in a precise sense, with the descendant correspondence.
Read the full abstract

Friday 10 April

General Topology

A Complete, Bounded and Almost Convex Metric Space which is Not Totally Convex

In the 2006 paper “Exponentiation in V-categories” by M. M. Clementino and D. Hofmann, exponentiable objects in the category of metric spaces with non-expansive maps are characterised as those metric spaces which are bounded and almost convex. A corollary to this characterisation is made claiming that every complete, exponentiable metric space is totally convex. A counterexample to this claim is provided in the present paper, and a correct version of the claim is suggested and proved.
Read the full abstract

Optimization and Control

Discrete Coplanarity Rigidity and Local Mahler Volume Growth Near the Cube

We study the finite-dimensional isovertex radial stratum of the cube, P(c)=conv{(1+c_q)q : q in Q_n={+/-1}^n}, with c(-q)=c(q), and establish quantitative rigidity and volume-product estimates for n>=3. The first main result identifies the kernel of the linearized facet-coplanarity operator R_n with the homothety direction span{1} and proves that the coplanarity defect is locally equivalent to both the transverse amplitude dist_infty(c,span{1}) and the distance d_GL(P(c),C_n) to the GL(n)-orbit of the cube. After normalizing out the homothety mode onto 1^perp, a key step is a second-order comparison between the linearized vertex-displacement seminorm and its optimization over small linear transformations. The second main result converts transverse radial perturbations c in 1^perp into a linear volume increment. A facetwise concave-envelope construction and an associated pocket functional yield the explicit primal lower bound vol(P(c))-2^n >= 8/(3(n-1))||c||_infty - O(||c||_infty^2) on 1^perp. Combining this with a chamber-independent first-order expansion of the polar volume (whose linear part vanishes on 1^perp with a uniform quadratic remainder) gives a linear lower bound for the Mahler product V_M(P(c))-V_M(C_n) in terms of d_GL(P(c),C_n). In dimension three, an exact chamber analysis on 1^perp yields vol(P(c))=8+(8/3)||c||_infty+O(||c||_infty^2), and the Mahler product has one-sided directional derivative (32/9)||u||_infty along every nonzero ray in 1^perp. In dimension four, a symmetry-orbit classification of the seventy balanced Boolean directions yields six combinatorial types and explicit first-order pocket bounds. We also distinguish the frozen-combinatorics KKT balance from geometric stationarity.
Read the full abstract

2606.14709 · 10 Apr · Substantial revision and deliberate narrowing of scope relative to arXiv:2606.14709v3. Broader discrete KKT and stability claims are withdrawn. This version focuses on the isovertex radial stratum of the cube, with quantitative coplanarity rigidity, linear Mahler lower bounds on 1^perp, explicit first-order coefficients for n>=3, and exact chamber and parity-orbit analysis in dimensions 3 and 4

Probability · Cross-listed from q-fin.ST

Systemic Risk and Default Cascades in Global Equity Markets: A Network and Tail-Risk Approach Based on the Gai Kapadia Framework

This study extends the Gai-Kapadia framework, originally developed for interbank contagion, to assess systemic risk and default cascades in global equity markets. We analyze a 30 asset network comprising Brazilian and developed market equities over the period 2015-2026, constructing exposure based financial networks from price co-movements. Threshold filtering (theta = 0.3 and theta = 0.5) is applied to isolate significant interconnections. Cascade dynamics are analyzed through a combination of deterministic propagation and stochastic Monte Carlo simulations (n = 1000) under varying shock intensities. The results show that the system exhibits strong global resilience, with a negligible probability of large scale failure, while maintaining localized vulnerability within highly clustered subnetworks. In particular, shocks lead to an average of 1.0 failed asset for single shocks and 2.0 for simultaneous shocks, indicating limited propagation below a critical threshold. Network analysis reveals a clear structural asymmetry: Brazilian assets display high clustering (Ci approx 0.8-1.0) and dense connectivity, which amplifies local shock propagation, whereas developed market assets exhibit lower connectivity (Ci approx 0.2-0.5), limiting systemic spread. Tail risk analysis, based on empirical CCDF and Hill estimators, confirms the presence of heavy tailed loss distributions, particularly in emerging markets, reinforcing their exposure to extreme events. These findings demonstrate that systemic risk arises from the interaction between network topology and tail behavior, rather than from isolated asset characteristics. The proposed framework provides a scalable and empirically grounded approach for stress testing and systemic risk assessment, offering relevant insights for regulators and portfolio managers in increasingly interconnected financial markets.
Read the full abstract

Probability · Cross-listed from q-fin.GN

Exploring Drivers of Extreme Housing Prices in Australia

In recent years Australia has observed a growing, unexplained resilience of increasing house price trends. Here, we seek to understand what is driving Australia's indestructible asset using insights from market experts. We construct a differential equation model of house price to develop intuition for its historical behaviour and responsiveness to changes in mortgage rates. Using this model, we identify a point of 'decoupling' between house price and mortgage rate in the system with supply limitations found to be the main driver for this change. From there, modern extreme value techniques are implemented on real-world data to investigate how the effectiveness of mortgage rate in moderating extreme house price has changed before and after this historical decoupling. We find that without an increase in the housing supply chain, through either deregulation or reduced competition with government building, an 11% increase in mortgage rate will be needed to slow extreme housing costs.
Read the full abstract

Information Theory · Cross-listed from cs.LO

Semantic Channel Theory: Deductive Compression and Structural Fidelity for Multi-Agent Communication

Shannon's information theory deliberately excludes message semantics. This paper develops a rigorous framework for semantic communication that integrates formal proof systems with Shannon-theoretic tools. We introduce an axiomatic information model comprising Lsem-definable state sets linked by computable enabling maps, and define the semantic channel as a composition of Markov kernels whose supports respect the enabling structure. A fixed proof system induces an irredundant semantic core and a derivation-depth stratification, enabling four distortion measures of increasing semantic depth: Hamming, closure, depth, and a parameterized composite. Six families of computable semantic channel invariants are defined and their inter-relationships established, including a data processing bound, a semantic Fano bound, and an ideal-channel collapse theorem. The central quantitative result is a deductive compression gain: under closure-based fidelity, the minimum block length is determined by the irredundant core size rather than the full knowledge-base size. We instantiate the framework for heterogeneous multi-agent communication, introducing an overlap decomposition that yields necessary and sufficient conditions for closure-reliable communication. A semantic bottleneck phenomenon is identified in broadcast settings: vocabulary mismatch imposes irreducible fidelity limitations even over noiseless carriers. All results are verified on an explicit Datalog instance.
Read the full abstract

Information Theory · Cross-listed from cs.LG

Sequential KV Cache Compression via Probabilistic Language Tries: Beyond the Per-Vector Shannon Limit

Recent work on KV cache quantization, culminating in TurboQuant, has approached the Shannon entropy limit for per-vector compression of transformer key-value caches. We observe that this limit applies to a strictly weaker problem than the one that actually matters: compressing the KV cache as a sequence. The tokens stored in a KV cache are not arbitrary floating-point data – they are samples from the exact formal language the model was trained on, and the model is by construction a near-optimal predictor of that language. We introduce sequential KV compression, a two-layer architecture that exploits this structure. The first layer, probabilistic prefix deduplication, identifies semantically equivalent shared prefixes across sessions using the trie metric d_T(s, s') = -log_2 P_M(s ^ s') from Probabilistic Language Tries (PLTs). The second layer, predictive delta coding, stores only the residual of each new KV vector from the model's own prediction of it, achieving a per-token entropy bound of H(KV_{i+1} | KV_{<=i}) <= H(token_{i+1} | token_{<=i}). We prove that at typical language model perplexity – approximately 10-20 for fluent English text – this bound is 3.3-4.3 bits on average per token position, compared to TurboQuant's 3 bits per vector component (with typical attention heads having 64-128 components). The theoretical compression ratio over TurboQuant is approximately 914,000x at the Shannon limit. Even at 1000x above the entropy floor – a deliberately pessimistic worst-case overhead, two orders of magnitude above the 2-5x typical of practical source coders – the ratio remains approximately 914x over TurboQuant, with compression improving rather than degrading as context length grows. The two layers are orthogonal and compose with existing per-vector quantization methods including TurboQuant.
Read the full abstract

Mathematical Physics

Characteristic polynomials of non-Hermitian random band matrices near the threshold

The paper arXiv:2510.04255 shows that the asymptotic behavior of the second correlation function of characteristic polynomials of the $N\times N$ non-Hermitian random band matrices with a bandwidth $W$ exhibits the transition at $W\sim \sqrt{N}$, as $W,N\to \infty$: it coincides with those for Ginibre ensemble for $W\gg \sqrt{N}$, and factorized as $1\ll W\ll \sqrt{N}$. In this work we extend the techniques of arXiv:2510.04255 to study the critical regime when the bandwidth $W$ is proportional to the threshold $\sqrt{N}$.
Read the full abstract