Peer Revenue Equity in science

Papers

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

Mathematics

math Watch Mathematics (sign in to use)

Monday 23 February

Algebraic Geometry

Explicit formulas and exact values for the number of rational points on singular curves over finite fields

We provide new explicit formulas for bounding the number of rational points on singular curves over finite fields. This enables us to obtain exact values of N q (g, $\pi$) which is defined as the maximum number of rational points over F q on a curve of geometric genus g and arithmetic genus $\pi$. We also give special attention to the case g = 2 in order to extend the work of Aubry and Iezzi on N q (0, $\pi$) and N q (1, $\pi$).
Read the full abstract

Probability

Regularization of a stationary point process by a stationary increments perturbation

We present a novel procedure where a stationary point process is regularized through the convolution with a continuous random field with stationary increments, in the sense that the dependency between distant points is weakened; and the potential peaks in the spectrum (or Bragg peaks), reminiscent of a periodic behavior, are erased. We use this procedure to efficiently generate a hyperuniform point process in dimension 1 using a fractional Brownian Motion; simulating n points with complexity n log(n).
Read the full abstract

Number Theory

The stacky Batyrev-Manin conjecture and modular curves

Let $\mathcal{X}_0(N)$ be the Deligne–Rapoport modular stack of elliptic curves endowed with a cyclic rational $N$-isogeny over a number field $F$. Let $N\in\{1,2,3,4,5,6,7,8,9,10,12,13,16,18,25\},$ which are precisely the values for which the coarse moduli space of $\mathcal{X}_0(N)$ is isomorphic to $\mathbb{P}^1$. We show that the stacky Batyrev–Manin conjecture [DY24] holds for the naive height on $\mathcal{X}_0(N)$ when $F=\mathbb{Q}$. In the process, we give a concrete description of $\mathcal{X}_0(N)$ as a square root stack over a stacky curve.
Read the full abstract

Algebraic Geometry

Relevant maps and the algebraic skeleton of simplicial toric prevarieties

Morphisms between schemes arising from multigraded rings are essential for understanding geometric relationships in algebraic geometry, yet a systematic theory for such maps has been lacking. In this paper, we develop a comprehensive framework for rational maps between multigraded Proj schemes by introducing several notions of maps between their underlying multigraded rings. A key challenge is that to induce actual morphisms (rather than just rational maps), the ring homomorphism $\varphi\colon R \to S$ must hit every relevant element in $S$. To address this, we introduce the use of relevant subsets $B \subseteq S_+$ (where $S_+$ is the ideal generated by all relevant elements), $B \unlhd S$, which allow us to control this condition more flexibly. As an application, we show that multigraded noetherian polynomial rings naturally encode combinatorial data, giving rise to systems of fans and thus to toric prevarieties. By leveraging our notion of rational maps with those relevant subsets, we prove that the category of triples $(D, S, B)$ - where $D$ is a finitely generated abelian group, $S$ is a $D$-graded noetherian polynomial ring, and $B \unlhd S$ is a subset of $S_+$ - together with rational maps of conical rings, is anti-equivalent to the category of simplicial toric prevarieties.
Read the full abstract

Numerical Analysis

Extreme $L_p$ discrepancy, numerical integration and the curse of dimensionality

The classical notion of extreme $L_p$ discrepancy is a quantitative measure for the irregularity of distribution of finite point sets in the $d$-dimensinal unit cube. In this paper we find a dual integration problem whose worst-case error is exactly the extreme $L_p$ discrepancy of the underlying integration nodes. Studying this integration problem we show that the extreme $L_p$ discrepancy suffers from the curse of dimensionality for all $p \in (1,\infty)$. It is known that the problem is tractable for $p=\infty$; the case $p=1$ stays open.
Read the full abstract

Combinatorics

Explicit construction of spherical $5$- and $7$-designs

This paper develops an explicit and implementable framework for constructing spherical designs by lifting point sets from tight fusion frames. By combining existing ingredients, we obtain, in every dimension, explicit spherical $5$-designs with $|X|=\mathcal{O}(d^3)$. As a core component of the method, we give an explicit construction of simplex $3$-designs realized as orbits of the symmetric group. Using these simplex designs as input, we further construct spherical $7$-designs in arbitrary even dimensions; more precisely, for every even integer $d\ge 6$ we obtain spherical $7$-designs in dimension $d$, and if $\frac{d}{2}-1$ is a prime power then the number of points is $\mathcal{O}(d^6)$.
Read the full abstract

Analysis of PDEs

A Lagrangian approach for prescribed mass solutions of cubic-quintic Schrödinger equations and $L^2$-supercritical problems

We study the existence of radially symmetric solutions of the following nonlinear scalar field equations in $\mathbb R^N$ ($N \geq 2$): $$ (*)_m \quad - \Delta u + \mu u = g(u) \quad \text{in}\ {\mathbb R}^N, \quad {1\over 2} \int_{{\mathbb R}^N} u^2\, dx = m,$$ where $g(s) \in C({\mathbb R},{\mathbb R})$, $m > 0$ and $\mu \in {\mathbb R}$ is an unknown Lagrangian multiplier. We take an approach using a Lagrangian formulation of $(*)_m$: $$J_m(\mu,u)={1\over 2}\int_{{\mathbb R}^N} |\nabla u|^2\,dx -\int_{{\mathbb R}^N} G(u)\,dx +\mu\left({1\over 2}\int_{{\mathbb R}^N} u^2\, dx-m\right) \in C^1((0,\infty)\times H_r^1({\mathbb R}^N), {\mathbb R})$$ and we give new general existence results through the function: $$ b_m:\, (0,\infty) \to {\mathbb R};\ \mu \mapsto \text{Mountain Pass minimax value for}\ (u\mapsto J_m(\mu,u)).$$ We will show the existence of solutions of $(*)_m$ related to local minima and local maxima of $b_m(\mu)$. As applications, we study cubic-quintic type equations and $L^2$-supercritical problems. In particular, when $N=2,3$, we show new existence results of normalized solutions without assuming global Ambrosetti-Rabinowitz type conditions, which partially improve the preceding results due to Jeanjean [24] and Jeanjean-Lu [26, 28].
Read the full abstract

Differential Geometry

The character of ideal circle patterns

Let $S$ be an oriented closed surface with a cellular decomposition $\mathcal{D}$ and a weight $\Phi\in(0, \pi)$. It is crucial to determine when $S$ supports an ideal $\mathcal{D}$-type circle pattern $\mathcal{P}$ with the exterior intersection angles given by $\Phi$. Rivin, Bobenko-Springborn and Ge-Hua-Zhou provided perfect solutions and gave wonderful criteria for the existence and uniqueness of ideal circle patterns. However, all criteria established by Rivin, Bobenko-Springborn and Ge-Hua-Zhou are extremely difficult to verify for the given cellular decomposition $\mathcal{D}$ and the weight $\Phi$. In this paper, we introduce the character $\mathcal{L}(\mathcal{D},\Phi)$ depends only on the data of the weighted cellular decomposition $(\mathcal D, \Phi)$ on $S$, and give some quite simple criteria for the existence of ideal circle patterns realizing $(\mathcal{D},\Phi)$. It seems that our character-type criteria are the first conditions totally different from criteria of Rivin, Bobenko-Springborn and Ge-Hua-Zhou, and provide more easily verifiable criteria. Our new character-type theorems may be of some independent interest. As an application, we give a new descriptions of the curvature image set $\mathbf{K}(\mathbb{R}^N_{>0})$. To approach our results, we shall use the combinatorial Ricci flows with ideal circle patterns introduced by Ge-Hua-Zhou as a fundamental tool. The main difficulty in the proof of our results is to establish the compactness of the solution to the flows. To circumvent the difficulty, we borrow the techniques developed by Ge and his collaborators.
Read the full abstract

Mathematical Physics · Cross-listed from quant-ph

Symmetry and Exact Solutions of General Spin-Boson Models

Spin-boson models are the canonical benchmark for quantum dissipation. We show the symmetry structure of general spin-boson Hamiltonians and obtain their spectra explicitly by exploiting the symmetry. As an illustration of the general case, we numerically demonstrate the exact solution for the two-mode case.
Read the full abstract

Dynamical Systems

Invariant measures for piesewise fractional linear maps

The first part deals with piecewise fractional linear maps with three branches. Given a map $T$ a map $S$ is called a related map if some branches of $T$ are replaced by a 'flipped' branch, namely a branch of $1-T$. The main question is if $T$ and $S$ have a common invariant measure. The short second part presents invariant measures of a new type.
Read the full abstract

Mathematical Physics · Cross-listed from cond-mat.stat-mech

Two-parameter families of matrix product operator integrals of motion in Heisenberg spin chains

Recently, Fendley et al. (2025) [arXiv:2511.04674] revealed a new simple way to demonstrate the integrability of XYZ Heisenberg model by constructing a one-parameter family of integrals of motion in the matrix product operator (MPO) form with bond dimension 4. In this work, I report on the discovery of two-parameter families of MPOs that commute with Heisenberg spin chain Hamiltonian in case of various anisotropies (XXX, XXZ, XX, XY and XYZ). These solutions are connected by taking appropriate limits. For all cases except XYZ, I also write down Floquet charges of two-step Floquet protocols corresponding to the Trotterization. I describe a symbolic algebra approach for finding such integrals of motion and speculate about possible generalizations and applications.
Read the full abstract

Differential Geometry

Projective Maps from the Perspective of Elliptic Differential Operators

This paper develops an analytical approach to the study of the geometry of projective maps using the theory of elliptic differential operators. We construct two elliptic operators of second and fourth order, whose kernels characterize projective diffeomorphisms between Riemannian manifolds and one-parameter groups of projective diffeomorphisms (transformations) of a Riemannian manifold onto itself, respectively. This approach establishes a natural correspondence between analytical and geometric properties, enabling the study of projective diffeomorphisms via operator-theoretic methods. The proposed framework provides a new understanding of projective structures on Riemannian manifolds and extends classical results in differential geometry.
Read the full abstract

Optimization and Control · Cross-listed from cs.LG

Understanding the Curse of Unrolling

Algorithm unrolling is ubiquitous in machine learning, particularly in hyperparameter optimization and meta-learning, where Jacobians of solution mappings are computed by differentiating through iterative algorithms. Although unrolling is known to yield asymptotically correct Jacobians under suitable conditions, recent work has shown that the derivative iterates may initially diverge from the true Jacobian, a phenomenon known as the curse of unrolling. In this work, we provide a non-asymptotic analysis that explains the origin of this behavior and identifies the algorithmic factors that govern it. We show that truncating early iterations of the derivative computation mitigates the curse while simultaneously reducing memory requirements. Finally, we demonstrate that warm-starting in bilevel optimization naturally induces an implicit form of truncation, providing a practical remedy. Our theoretical findings are supported by numerical experiments on representative examples.
Read the full abstract

Number Theory

On the analytic continuation of Dirichlet series with missing digits

We study the Dirichlet series associated with the integers whose radix-$b$ representation misses certain (fixed) digits. The existence of a meromorphic continuation to the entire complex plane, which was already well-known as a general fact valid for $b$-automatic Dirichlet series, is proven anew from a representation as an everywhere defined series with good convergence properties. A generating function related to the residues on the real axis is shown to be the multiplicative inverse of the moment generating function for the associated Cantor set in the unit interval. This makes the (normalized) residues some sort of generalized Bernoulli numbers.
Read the full abstract

Number Theory

Arithmetic Aspects of Number Fields Generated by Polynomial Families

Let $f(x)=(x^{k}+c)^{m}-ax^{n}\in\mathbb{Z}[x]$ be an irreducible polynomial over $\mathbb{Q}$, where $k,m,n\in\mathbb{N}$ with $km>n$, and let $K=\mathbb{Q}(\theta)$, where $\theta$ is a root of $f(x)$. We investigate the arithmetic properties of the number fields that arise from this family. We first obtain an explicit formula for the discriminant of $f(x)$. Using this formula, we establish necessary and sufficient conditions for the monogeneity of $f(x)$, expressed in terms of the prime divisors of $a$ and $c$ and the parameters $k,m,n$. This yields infinite families of monogenic polynomials of arbitrary degree, including families with a non-square-free discriminant. Building on these results, we extend our algebraic characterization to composite polynomials, establishing some explicit conditions for the monogeneity of the composition of $f(x)$ with an arbitrary polynomial $g(x)$. From an analytic point of view, we derive asymptotic estimates for the number of monogenic polynomials in these families under natural assumptions. We further study non-monogeneity via the field index $i(K)$ and, for each prime $p$, provide sufficient conditions ensuring $\nu_p(i(K))=1$, yielding partial progress toward a problem of Narkiewicz. We also highlight a connection with a class of differential equations naturally associated with $f(x)$. As an application, we determine the conditions under which the splitting field of $f(x)$ has a full symmetric Galois group. Several explicit examples illustrate our results.
Read the full abstract

All authors Rupam Barman, Anuj Jakhar, Ravi Kalwaniya, Prabhakar Yadav

Complex Variables

Generalized stationary discs attached to degenerate submanifolds in $\mathbb{C}^N$

We study the family of generalized stationary discs attached to a Levi degenerate submanifold M of codimension d in $\mathbb{C}^{n+d}$. We show, under suitable geometric assumptions on M, that this family forms a finite dimensional real submanifold of the Banach space of analytic discs.
Read the full abstract

All authors Mohammad Tarek Al Masri, Florian Bertrand, Francine Meylan, Lea Oueidat, Hadi Zoghaib

2602.19717 · 23 Feb · This work was done in the framework of the Summer Research Camp in Mathematics designed by the Department of Mathematics at the American University of Beirut (AUB) and that benefited from a generous support from the Center for Advanced Mathematical Sciences and the Faculty of Arts and Sciences at AUB

Statistics Theory

On Expectation Propagation and the Probabilistic Editor in some simple mixture problems

As for other latent-variable problems, exact Bayesian analysis is typically not practicable for mixture problems and approximate methods have been developed. Variational Bayes tends to produce approximate posterior distributions for parameters that are too tightly concentrated in having variances that are too small. The paper identifies a few mixture problems in which Expectation Propagation and variations thereof lead to approximate posterior distributions that asymptotically exhibit `correct' variances and therefore stand to provide reliable interval estimates for the unknown parameter or parameters.
Read the full abstract

Optimization and Control · Cross-listed from cs.RO

CACTO-BIC: Scalable Actor-Critic Learning via Biased Sampling and GPU-Accelerated Trajectory Optimization

Trajectory Optimization (TO) and Reinforcement Learning (RL) offer complementary strengths for solving optimal control problems. TO efficiently computes locally optimal solutions but can struggle with non-convexity, while RL is more robust to non-convexity at the cost of significantly higher computational demands. CACTO (Continuous Actor-Critic with Trajectory Optimization) was introduced to combine these advantages by learning a warm-start policy that guides the TO solver towards low-cost trajectories. However, scalability remains a key limitation, as increasing system complexity significantly raises the computational cost of TO. This work introduces CACTO-BIC to address these challenges. CACTO-BIC improves data efficiency by biasing initial-state sampling leveraging a property of the value function associated with locally optimal policies; moreover, it reduces computation time by exploiting GPU acceleration. Empirical evaluations show improved sample efficiency and faster computation compared to CACTO. Comparisons with PPO demonstrate that our approach can achieve similar solutions in less time. Finally, experiments on the AlienGO quadruped robot demonstrate that CACTO-BIC can scale to high-dimensional systems and is suitable for real-time applications.
Read the full abstract

All authors Elisa Alboni, Pietro Noah Crestaz, Elias Fontanari, Andrea Del Prete

Dynamical Systems

Existence of periodic measure-valued solutions to the nonlocal continuity equation via optimal transport

We investigate the existence of periodic solutions for a class of nonlocal continuity equations, which include mean-field equations derived from systems of coupled oscillators. While periodic solutions at the particle level have been studied through the construction of a Poincaré map on a section of an invariant set, extending this analysis to the level of continuity equations presents nontrivial challenges. In particular, setting an appropriate topology for the infinite-dimensional space to show invariance and apply the fixed point argument is not easy. To overcome this difficulty, we use fixed point theorem for geodesically convex spaces constructed by optimal transportation. Specifically, from the disintegration with respect to stationary variable, we define a metric using the Wasserstein-$2$ distance over one-dimensional space, which yields a $CAT(0)$ space. In this topology, we construct an invariance set of probability measures and prove the existence of the periodic measure-valued solution from Schauder's fixed point theorem on geodesically convex spaces. As a corollary, our method directly gives an existence of periodic graph measure solution.
Read the full abstract

All authors Seung-Yeal Ha, Gyuyoung Hwang, Philippe Thieullen, Jaeyoung Yoon

Numerical Analysis · Cross-listed from stat.ML

Smoothness Adaptivity in Constant-Depth Neural Networks: Optimal Rates via Smooth Activations

Smooth activation functions are ubiquitous in modern deep learning, yet their theoretical advantages over non-smooth counterparts remain poorly understood. In this work, we study both approximation and statistical properties of neural networks with smooth activations for learning functions in the Sobolev space $W^{s,\infty}([0,1]^d)$ with $s>0$. We prove that constant-depth networks equipped with smooth activations achieve smoothness adaptivity: increasing width alone suffices to attain the minimax-optimal approximation and estimation error rates (up to logarithmic factors). In contrast, for non-smooth activations such as ReLU, smoothness adaptivity is fundamentally limited by depth: the attainable approximation order is bounded by depth, and higher-order smoothness requires proportional depth growth. These results identify activation smoothness as a fundamental mechanism, complementary to depth, for achieving optimal rates over Sobolev function classes. Technically, our analysis is based on a multi-scale approximation framework that yields explicit neural network approximators with controlled parameter norms and model size. This complexity control ensures statistical learnability under empirical risk minimization (ERM) and avoids the impractical $\ell^0$-sparsity constraints commonly required in prior analyses.
Read the full abstract

All authors Yuhao Liu, Zilin Wang, Lei Wu, Shaobo Zhang

Differential Geometry

Classification of quadratically pinched self-shrinkers in higher codimension

We classify properly immersed self-shrinkers of the mean curvature flow in arbitrary codimension under a quadratic pinching condition of Andrews-Baker type on the second fundamental form that is preserved along the flow. Under this assumption, such self-shrinkers reduce effectively to codimension one and are therefore generalized self-shrinking cylinders. In contrast to previous works, our approach is purely elliptic: it relies on parabolicity in a weighted setting and is tailored specifically to self-shrinkers, rather than to general ancient solutions of the flow. This allows us to avoid assuming any uniform pinching condition, to treat in any dimension the sharp Andrews-Baker pinching constant $\frac{4}{3n}$ and hence to sharpen, in the self-shrinker setting, the pinching constants appearing in recent classification results for ancient solutions.
Read the full abstract

Number Theory

A cohomological invariant for algebras of degree 8 and exponent 2 in characteristic 2

Our aim in this paper is to extend a work of Sivatski to characteristic 2. More precisely, for $F$ a field of characteristic $2$ and a central simple algebra $A$ of exponent 2 that splits over a triquadratic extension of $F$ of separability degree at least 4, we attach a cohomological invariant $\inv(A) \in H_2^3(F) / G$, where $H_2^3(F)$ is the third Kato-Milno cohomology group and $G$ is a subgroup of $H_2^3(F)$ divisible by the Brauer class of $A$. As an application, we will relate the decomposability of the algebra in degree 8 to the vanishing of $\inv(A)$. Moreover, we will use this invariant to prove some descent results for central simple algebras and quadratic forms over biquadratic extensions.
Read the full abstract

Geometric Topology

Isospectrality and isometry groups for infinite-type hyperbolic surfaces with discrete length spectrum

We study infinite-type hyperbolic surfaces with discrete length spectrum. In this setup, we show that Sunada's method can only produce finite isospectral families, but that there is no bound on the cardinality of an isospectral family (for any infinite-type surface without planar ends). We then prove that, given any infinite-genus surface satisfying an additional topological assumption, any finite group can be realized as isometry group of a hyperbolic structure with discrete length spectrum.
Read the full abstract

Combinatorics

A Note on Grünbaum's Conjecture about Longest Cycles and Paths

Let $c(G)$ denote the circumference of a graph $G$, i.e., the number of vertices in its longest cycle. For positive integers $n$ and $k$ with $n>k$, let $\varGamma(n;k)$ be the class of graphs of order $n$ with $c(G) = n-k$ such that every induced subgraph of order $n-k$ is Hamiltonian. When $k=$, the class $\varGamma(n; 1)$ coincides with the family of hypohamiltonian graphs-non-Hamiltonian graphs in which the deletion of any single vertex yields a Hamiltonian graph.Replacing Hamiltonian with traceable and $c(G)$ with $p(G)$, the order of a longest path, defines the analogous class $\varPi(n;k)$.Grünbaum (1974) conjectured that both $\varGamma(n; k)$ and $\varPi(n; k)$ are empty for all $n>k \ge 2$. In this note, we first establish upper bounds on the maximum degree of graphs in the classes $\varGamma(n; k)$ and $\varPi(n; k)$. Using these bounds, we show that $\varGamma(n; k)$ is empty when $n<k^2+2k+3$, and that $\varPi(n; k)$ is empty when $n<k^2+2k+2$. These results provide further evidence supporting Grünbaum's conjecture.
Read the full abstract

Analysis of PDEs

Gradual smoothing: strong hypercontractivity and logarithmic Sobolev inequalities

We study the possibility of a gradual improvement as time progresses of the regularity of solutions to evolution problems of parabolic type driven by Lévy-type operators, not necessarily translation invariant. In the course of our analysis we study the equivalence between general smoothing effects and a family of logarithmic Sobolev inequalities. This equivalence allows us to identify a new type of regularization, strong hypercontractivity, characterized by the existence of a time at which solutions belong to every $L^p$ space with $p$ finite. It can also be used to prove logarithmic Sobolev inequalities in a context not previously seen in the literature. We then show that any purely nonlocal Lévy-type operator whose kernel is comparable to that of $\log(I-\Delta)$ is strongly hypercontractive, but fails to be supercontractive and, consequently, also fails to be ultracontractive. Furthermore, in the translation-invariant case, we also prove that solutions get bounded eventually and start improving in differentiability right after doing so. Finally, we show that this behaviour only appears if the kernel defining the operator behaves as $|x-y|^{-N}$ for small interactions ($0^+$-order operators): more singular kernels yield instantaneous smoothing, while less singular ones do not produce any regularization.
Read the full abstract

All authors Arturo de Pablo, David Lee, Fernando Quirós, Jorge Ruiz-Cases

Number Theory

$L^4$-norms of automorphic forms in the depth aspect

Let $p$ be an odd prime, and suppose $f$ is an $L^2$-normalised newform for $\Gamma_0(p^n)$ with bounded spectral parameters and trivial central character. We prove the optimal $L^4$-norm bound $\lVert f \rVert_4 \ll_{p,\varepsilon}(p^{n})^\varepsilon $ for all $\varepsilon >0$ as $n \rightarrow \infty$.
Read the full abstract

Representation Theory

On Lusztig's canonical bases of simple Lie algebras

Let $\mathfrak{g}$ be a simple Lie algebra over~$\mathbb{C}$ with root system~$\Phi$. In the simply laced case, Frenkel and Kac found a particularly simple construction of~$\mathfrak{g}$, together with a Chevalley basis and explicitly given structure constants, in terms of a certain multiplicative $2$-cocycle $\varepsilon\colon \mathbb{Z} \Phi\times \mathbb{Z}\Phi \rightarrow\{\pm 1\}$. We show that Lusztig's canonical basis of~$\mathfrak{g}$ can also be obtained in this way, for a suitable choice of~$\varepsilon$. We also address the problem of explicitly describing the structure constants when $\Phi$ is not simply laced.
Read the full abstract

Information Theory

Nacrith: Neural Lossless Compression via Ensemble Context Modeling and High-Precision CDF Coding

We present Nacrith, a lossless compression system that combines a 135M-parameter transformer language model (SmolLM2-135M) with an ensemble of lightweight online predictors and a 32-bit arithmetic coder, achieving the best compression results among the systems evaluated in this study on natural language text. Beyond the base LLM-plus-arithmetic-coding paradigm, Nacrith introduces several contributions: (1) a CDF precision upgrade from 2^16 to 2^24 that eliminates ~75% of quantization overhead caused by minimum-probability floors in large vocabularies; (2) a token-level N-gram model for fast local predictions; (3) an adaptive log-space bias head correcting per-document LLM errors via online gradient descent; (4) confidence-based LLM skip for accelerating highly predictable tokens; (5) a hybrid binary format (NC06) extending neural compression to arbitrary binary files–to our knowledge a first among LLM-based compressors; (6) a llama cpp inference backend achieving ~7x faster single-token decode than PyTorch; (7) parallel multi-GPU compression across up to 8 workers; and (8) native KV cache sliding window reducing per-slide cost by ~37x. The system requires only ~500 MB of GGUF weights and ~1.2 GB VRAM per worker, running on consumer GPUs. On alice29 (Canterbury Corpus, 152 KB), Nacrith achieves 0.918 bits per byte (bpb)–outperforming gzip by 3.1x, bzip2 by 2.5x, CMIX v21 by 44%, and ts_zip by 20%, while compressing below the 0th-, 1st-, and 2nd-order byte-level Shannon entropy bounds. On enwik8 (100 MB), Nacrith achieves 0.9389 bpb (11.74%), surpassing ts_zip (~1.11 bpb) by 15% and FineZip (1.024 bpb) by 8% despite using a 60x smaller model with no fine-tuning. An out-of-distribution (OOD) evaluation on a document published after the model's training cutoff confirms these gains are not memorization artifacts, achieving 0.723 bpb on unseen text.
Read the full abstract

Probability

Laplace Transforms of Stopping Times for Subordinator with Applications to Inventory Control

Intermittent demand fluctuations pose significant challenges in disaster logistics and medical supply systems. In this study, we formulate cumulative demand as a generalized Lévy process composed of a drift term, Poisson jumps, and compound Poisson jumps, and analyze a continuous-time inventory model. The proposed framework provides a unified formulation that encompasses both drifted Poisson processes and drifted compound Poisson processes. From a mathematical perspective, we treat the reorder time as a first-passage problem of a subordinator and derive its Laplace transform via the Laplace exponent. In particular, for the drifted Poisson case, we obtain an explicit representation of the inverse Laplace exponent using the Lambert W function, which yields an analytic expression for the Laplace transform of the first-passage time. Furthermore, when the jump sizes follow exponential and Gamma distributions, we derive explicit formulas for the mean and variance of the reorder times, thereby clarifying the moment structure of first-passage times for generalized Lévy demand processes. From an operations research perspective, we explicitly characterize the expected total cost over a finite time horizon based on the distribution of cumulative demand. This study presents an analytical framework that integrates first-passage theory of Lévy processes with continuous-time inventory control.
Read the full abstract