Information Theory · Cross-listed from quant-ph
Quantum Prediction of Transport Dynamics in Discretized State Spaces
Felix Govaers
We propose a gate-based
quantum algorithm for the prediction step of
Bayesian state estimation based on the Fokker-Planck equation on a discretized position-velocity state space. The probability density is encoded in the amplitudes of a
quantum state, enabling a compact representation of high-dimensional distributions. Exploiting the circulant structure of finite-difference operators, the
evolution is realized in the spectral domain using
quantum Fourier transforms and phase rotations. A key result is that the drift component can be implemented exactly in amplitude space, leading to an accurate reproduction of the classical transport dynamics. In contrast, the
diffusion term does not admit a linear representation in amplitude space due to the nonlinear relation between probability density and wave function. To enable a
quantum implementation, we introduce a
unitary surrogate based on a Wick rotation, transforming
diffusion into a dispersive phase
evolution. This yields a fully
unitary propagation that can be implemented efficiently on a gate-based
quantum computer. The proposed method is evaluated numerically for different scenarios and shows strong agreement with the exact solution of the Fokker-Planck equation. The approach demonstrates the potential of
quantum computing for
Bayesian state estimation, as the representable state space grows exponentially with the number of
qubits. This allows the efficient representation and propagation of probability densities that would otherwise require complex tensor decompositions on classical hardware, making the method a promising candidate for high-dimensional filtering problems.
Read the full abstract ▾
Dynamical Systems
Semi-discrete approximation of Mañé, Aubry and Mather sets
Fabio Camilli and Cristian Mendico
We study a variational discretization in time of Aubry–Mather theory for Tonelli Lagrangians on the flat
torus. We define discrete Mañé, Aubry and Mather sets, establish their compactness and inclusion properties, and investigate their convergence as the time step tends to zero. We prove upper-limit results for the Mañé sets and, when the continuous Aubry set has finitely many static classes, for the Aubry sets. We obtain full Kuratowski and Hausdorff convergence of the Aubry sets when this set is a single
equilibrium or periodic orbit, and, under additional assumptions, a finite union of
equilibria. Every weak limit of discrete Mather measures is a continuous Mather measure. For
equilibrium Aubry sets, we give sufficient conditions for exact preservation of Mather sets and measure families at small time steps. A mechanical example illustrates these results and exhibits exact preservation of the Aubry set as well.
Read the full abstract ▾
General Topology
A Descriptive Perspective on Devaney's Chaos and Some Results on Topologically Conjugate Systems
Fatih Ucan and Tane Vergili
In this study, Devaney's chaos conditions are revisited within the framework of descriptive proximity. The concepts of descriptive transitivity, the density of descriptive periodic objects, and descriptive sensitivity are defined. The most notable finding of the study is that Banks
Theorem, which establishes the hierarchy among these conditions in classical
topology, does not generally hold in the descriptive perspective, and some of the concepts above remain invariant under topological conjugacy certain conditions.
Read the full abstract ▾
Combinatorics
The minimum number of detours in a connected graph of minimum degree three
Xining Liu, Pu Qiao and Xingzhi Zhan
A longest path in a graph is called a detour. Denote by $a(k,n)$ the minimum number of detours in a connected graph with minimum degree $k$ and order $n,$ and denote by $b(k,n)$ the minimum odd number of detours in such a graph. X. Zhan has posed the problem of determining $a(k,n)$ and $b(k,n).$ It is known that $a(2,n)=4$ for $n\ge 4$ and $b(2,n)=9$ for $n\ge 9.$ In this paper we prove that $a(3,n)=36$ for $n\ge 18,$ $a(k,n)\le (k!)^2$ for $n\ge k^2+2k+3$ and $b(3,n)\le 225$ for $n\ge 11.$ We also pose several related unsolved problems.
Read the full abstract ▾
Optimization and Control
The Benjamin-Ono equation with 2D control input: approximate controllability and its application
Jia-Cheng Zhao
We establish the approximate controllability in $L^2$ for the nonlinear Benjamin-Ono equation on
torus via two-dimensional control input. Our proof is based on adaptations of
geometric control approach introduced by Agrachev and Sarychev. As an application of this control result, we study long-time dynamics of a randomly forced equation. It is proved that the trajectories are unbounded in Sobolev norms almost surely, when the random force is nondegenerate and statistically periodic in time.
Read the full abstract ▾
Differential Geometry
Finite-Gap Solutions of the Pohlmeyer–Lund–Regge Equation and the Associated Lund-Regge Surfaces
Yuhei Kogo
We develop a finite-gap construction for the Pohlmeyer-Lund-Regge (PLR) equation and the associated Lund-Regge surfaces. From the hyperelliptic spectral data we build a Baker-Akhiezer function and an SU(2)-frame, yielding an explicit theta-quotient formula for the PLR solution. We then derive criteria for s-periodicity and t-periodicity: under the stated theta-period assumptions, s-periodicity and t-periodicity are each equivalent to a critical-point condition for the corresponding normalized differential together with a phase condition at the reconstruction point. This provides a PLR analogue of the periodicity mechanism of Calini-Ivey.
Read the full abstract ▾
Algebraic Geometry
Dimension of the motivic Galois group of a 1-motive
Cristiana Bertolin
We compute the dimension of the motivic Galois group of a 1-motive M defined over the field of complex numbers, expressing it explicitly in terms of the rank of the multiplicative group generated by the points defining M. As an application, we obtain a new formulation of the Grothendieck–André periods
Conjecture in the setting of 1-motives.
Read the full abstract ▾
Mathematical Physics · Cross-listed from hep-th
Super Landau Model and Howe Duality: From Supermonopole Harmonics to Quantum Matrix Geometry
Kazuki Hasebe
Landau models serve as
quantum mechanical systems for generating
quantum matrix geometries. In this paper, we demonstrate that Howe duality provides the underlying structure of the super Landau model, reflecting a general feature of coset-type Landau models. The (super) Howe duality relates different Landau levels and accounts for the emergence of a dual fuzzy geometry. By employing super-spinor derivative operators, the supermonopole harmonics in both integer and half-integer Landau levels are explicitly constructed and the algebraic structure of the super-
Hilbert space is revealed. We propose a consistent probabilistic interpretation for these wavefunctions defined on a supermanifold. Through a level projection method, we derive the
matrix coordinates of fuzzy supersphere geometries for arbitrary Landau levels, along with a precise determination of the non-commutative scale factor. It is shown that the theta correspondence of Howe duality induces a
geometric transformation between fuzzy objects. Finally, we point out that Howe duality realizes an internal-external space duality and underlies
quantum matrix geometries, suggesting that it may play a fundamental role in understanding
Matrix model geometries.
Read the full abstract ▾
Optimization and Control
Optimal control for a fourth-order nonisothermal tumor growth model of Caginalp type
Giulia Cavalleri, Pierluigi Colli and Elisabetta Rocca
We study a distributed optimal control problem for a nonisothermal Caginalp-type phase-field model that describes tumour growth under thermal therapy. The PDE system couples a possibly viscous Cahn-Hilliard equation, governing the
evolution of the healthy and tumor phases, with an equation for the heat balance, and a reaction-
diffusion equation for the nutrient concentration. Chemotaxis and active transport effects are taken into account, and hyperthermia appears as a control variable. We introduce a suitable tracking-type cost functional and show the existence of optimal controls. Then, we analyse the differentiability of the control-to-state operator and establish necessary first-order conditions expressed through a variational inequality involving the adjoint state variables.
Read the full abstract ▾
Algebraic Geometry
Birational boundedness of stable families
Paolo Cascini, Jihao Liu, Calum Spicer et al. (+1 more)
We prove that normal projective stable families of maximal variation, of fixed dimension, and with bounded adjoint volume are birationally bounded. This is a consequence of a substantially stronger statement, formulated a priori independently of stable families: algebraically integrable foliations of fixed dimension and bounded adjoint volume are log birationally bounded. In this way, the birational geometry of foliations provides a systematic framework for approaching classical boundedness problems for fibrations. A key input is our proof of M\textsuperscript{c}Kernan's ACC
conjecture for interpolated log canonical thresholds of algebraically integrable foliations. This may be viewed as the foliated analogue of Shokurov's ACC
conjecture for log canonical thresholds, proved in the classical setting by Hacon–M\textsuperscript{c}Kernan–Xu. As applications, we establish two boundedness criteria for Fano algebraically integrable adjoint foliated structures: Birkar's criterion for exceptional Fanos, and Jiang's criterion for Fanos for which both Tian's $\alpha$-invariant and the anti-canonical volume are bounded away from zero. We also obtain several results on the birational geometry of algebraically integrable adjoint foliated structures, including lower bounds for adjoint volumes, boundedness of
automorphism groups, and ACC
theorems for pseudo-effective thresholds, $\mathbb{R}$-complementary thresholds, and the Fano spectrum.
Read the full abstract ▾
All authors Paolo Cascini, Jihao Liu, Calum Spicer, Roberto Svaldi
Numerical Analysis
Quasi-Monte Carlo with a Hankel random digital net
Takashi Goda, Yang Liu and Raúl Tempone
This paper proposes a new randomized design of digital nets in which the generating
matrices are chosen to be random Hankel
matrices. Compared with previous randomized designs of digital nets, this approach simplifies the construction process and reduces the number of random variables required, while still achieving desirable convergence rates when combined with appropriate
estimators. We analyze the properties of the proposed design, derive bounds for Walsh coefficients, and provide error analysis for both the median-of-means
estimator and a newly proposed greedy selection
estimator, i.e. the selection of the best design from a batch in terms of a worst-case error bound. Numerical experiments validate our theoretical findings and demonstrate the practical performance of the proposed methods.
Read the full abstract ▾
Group Theory
The Number of Isomorphism Classes of Beauville Surfaces with Beauville $p$-Group
Şükran Gül
A Beauville surface is a rigid complex surface of general type, isogenous to a higher product by the free action of a finite group $G$, called a Beauville group. In \cite{GT}, González-Diez and Torres-Teigell find the number of isomorphism classes of Beauville surfaces for which the group $G$ is $\PSL(2,p)$ with particular types of `Beauville structures'. On the other hand, in \cite{GJT}, González-Diez, Jones and Torres-Teigell give an explicit formula for this number when the group $G$ is abelian. To the best of the author's knowledge, in the literature, the exact number of isomorphism classes of Beauville surfaces is given only for $\PSL(2,p)$ and for abelian groups. In this paper, we extend the result for Beauville surfaces with abelian $p$-group to Beauville surfaces for which the Beauville group is either a non-abelian metacyclic $p$-group or a $p$-group of nilpotency class $2$.
Read the full abstract ▾
Numerical Analysis
Submatrices with the best-bounded inverses: an asymptotically tight upper bound for $\mathbb{C}^{n \times 2}$
Yuri Nesterenko
The long-standing hypothesis formulated by Goreinov, Tyrtyshnikov and Zamarashkin \cite{GTZ1997} has recently been solved affirmatively in the case of real two-column
matrices by Sengupta and Pautov \cite{SP2026}. In this paper, we consider the complex variant of this problem and prove the asymptotically tight upper bound for spectral norms of the best-bounded inverse $2 \times 2$ submatrices of an arbitrary complex $n \times 2$
matrix with orthonormal columns.
Read the full abstract ▾
Combinatorics
An exponentially small gap of the Perron vector on independent sets
Hongzhang Chen, Jianxi Li, Yongtao Li et al. (+2 more)
A classical result of Cioabă states that if $G$ is a connected graph with the unit Perron vector $\mathbf{x}$, then any independent set $S$ of $G$ satisfies $\sum_{v\in S} x_v^2 \le \frac{1}{2}$, with equality if and only if $G$ is a bipartite graph and $S$ is one of the partite sets. Let $\chi(G)= k $ be the chromatic number of $G$. A well-known
conjecture of Gregory asserts that any independent set $S$ of $G$ satisfies $\frac{1}{2} - \sum_{v\in S}x_v^2 = \Omega ((k/n)^{1/2})$. Recently, Liu and Ning [J. Combin. Theory Ser. B 176 (2026)] disproved Gregory's
conjecture by constructing a graph $G$ and an independent set $S$ such that $\frac{1}{2}- \sum_{v\in S}x_v^2 = O(k^5/n^3)$. Furthermore, they conjectured that this bound is tight up to a constant factor. In this paper, we first show that any cycle $C_n$ with odd integer $n\ge 7$ provides a simple counterexample to Gregory's
conjecture. Second, we establish that for any independent set $S$, we have $\frac{1}{2} - \sum_{v\in S}x_v^2 = \frac{q}{4\lambda -2q}$, where $\lambda$ is the spectral radius of $G$, and $q$ is the Rayleigh quotient of $\mathbf{x}$ restricted to $\bar{S} :=V(G)\setminus S$. Third, we construct a graph with arbitrarily large chromatic number and find an independent set $S$ such that $\sum_{v\in S}x_v^2$ can be arbitrarily close to $\frac{1}{2}$, with an exponentially small gap. Our construction shows that there is no universal lower bound of the form $\Omega (k^{\alpha}/n^{\beta})$ for any $\alpha, \beta >0$. This settles both Gregory's original
conjecture and the modified
conjecture of Liu and Ning in the negative. Finally, we show the tightness of our construction and provide some local weighted lower bounds.
Read the full abstract ▾
All authors Hongzhang Chen, Jianxi Li, Yongtao Li, Lele Liu, Bo Ning
Optimization and Control
An online adaptive finite-element method for nonsmooth PDE-constrained optimization
Harbir Antil, Robert J. Baraldi, Rohit Khandelwal et al. (+1 more)
We present a trust-region-based adaptive finite-element
algorithm for numerically solving a class of nonsmooth PDE-constrained optimization problems that includes problems with sparsifying regularizers and convex constraints. In particular, we consider the class of problems whose objective function is the sum of a smooth, possibly nonconvex, function and a nonsmooth extended real-valued convex function. Our method combines the robustness of inexact trust-region
algorithms for nonsmooth problems with the efficiency of adaptive finite-element discretizations. Starting from a coarse mesh, the
algorithm automatically refines the discretization based on reliable a posteriori error
estimators for both the state and adjoint equations, systematically controlling the accuracy of the computed smooth objective function value and gradient. This adaptivity mechanism balances computational cost and solution accuracy, enabling high resolution of localized phenomena and sparsity structures in the state and control variables. We demonstrate the performance of our
algorithm through numerical experiments on representative control and
topology optimization examples.
Read the full abstract ▾
All authors Harbir Antil, Robert J. Baraldi, Rohit Khandelwal, Drew P. Kouri
Dynamical Systems
A robust obstruction to full strong pluripotency for wild blender-horseshoes
Shin Kiriki, Xiaolong Li, Yushi Nakano et al. (+1 more)
Suppose that $M$ is a closed
manifold of dimension greater than two and $r \ge 2$. We show that there exists a \(C^r\)-diffeomorphism \(f:M\to M\) with a wild affine blender-horseshoe \(\Lambda_f\) such that \(f\) is \(C^r\)-robustly strongly pluripotent, in the sense that after arbitrarily small perturbations a positive-Lebesgue-measure family of orbits can shadow a prescribed orbit in time average, for the dense subset \(\Lambda_f^{(\mathrm{mj})}\) consisting of itineraries satisfying the majority condition, namely, those for which the symbol \(0\) occupies at least half of certain growing terminal blocks, but not for the whole horseshoe \(\Lambda_f\).
Read the full abstract ▾
All authors Shin Kiriki, Xiaolong Li, Yushi Nakano, Teruhiko Soma
Probability
A characterization of ruin-inducing probability measures in a renewal risk model
Spyridon M. Tzaninis and Apostolos Bozikas
In this work, we derive a complete characterization of all ruin-inducing probability measures that preserve the structure of a given compound renewal process in terms of suitable pairs of functions $(\gamma,\delta)$. This result allows us to obtain an explicit representation of the infinite-time ruin probability as an expectation under any ruin-inducing probability measure. A key feature of our approach is that the construction of these measures does not rely on the existence of moment generating functions, and is therefore applicable to heavy-tailed claim size distributions. The proposed framework includes the classical Esscher transform as a special case.
Read the full abstract ▾
Dynamical Systems
Quantitative Hölder Regularity, Concentration, and Spectral Applications for Lyapunov Exponents of Random $\operatorname{GL}(2,\mathbb{R})$ Cocycles, with Extensions to $\operatorname{GL}(d,\mathbb{R})$
Abdoulaye Thiam
This paper develops a quantitative regularity theory for the Lyapunov exponents of random products of
matrices in $\operatorname{GL}(2,\mathbb{R})$, with extensions to $\operatorname{GL}(d,\mathbb{R})$ for all $d \geq 2$. At every compactly supported measure $\nu$ with simple Lyapunov spectrum, we give an explicit closed-form Hölder exponent $\beta_*(\nu, \theta)$ and constant in the modulus of continuity of $\lambda_\pm$ in the Wasserstein-plus-Hausdorff metric, depending only on the eccentricity of $\mathrm{supp}\,\nu$, the Lyapunov gap, and the Hölder index $\theta$. At every $\nu \in \textit{M}_c(\operatorname{GL}(2,\mathbb{R}))$ we identify the log-Hölder exponent of Tall and Viana as $\kappa_*(\nu, \theta) = \theta/(2+\theta)$ under a natural mixing hypothesis, and $\theta/(8(1+\theta))$ in the perpetuity regime. The same spectral-gap method yields a large deviation principle with explicit rate function, Hoeffding-Azuma concentration inequalities, an extension to Markov-chain driven cocycles with closed-form exponent, and a quantitative log-Hölder modulus of continuity for the integrated density of states of one-dimensional random Schrödinger operators with absolutely continuous disorder. The Hölder theory extends to $\operatorname{GL}(d,\mathbb{R})$ for the top exponent under spectral simplicity, and to the partial sums $\Lambda_k = \lambda_1 + \cdots + \lambda_k$ under strong $k$-irreducibility, yielding Hölder continuity of each individual sub-top exponent. A method-optimality proposition shows that $\beta_*$ is the best exponent obtainable from the linear balance of axioms (A1)-(A3) of the spectral-gap method; strict improvement requires either modifying these axioms or adopting a different proof strategy. A lower-bound proposition adapted from Duarte, Klein, and Santos rules out uniform Hölder continuity across $\textit{M}_c(\operatorname{GL}(2,\mathbb{R}))$.
Read the full abstract ▾
Optimization and Control
Analysis of Stability and Performance of Economic Model Predictive Control with State-Independent Costs
Alireza Arastou, Ye Wang and Erik Weyer
This paper studies economic model predictive Control (EMPC) schemes, where the stage cost depends only on control inputs. Such problems arise in applications like water distribution networks and differ from standard EMPC since multiple steady states can correspond to the unique optimal steady input. We show that, under a strict dissipativity assumption related to the set of optimal steady states, the closed-loop trajectories converge asymptotically to this set, ensuring convergence of the economic cost to the optimal steady state cost. To enhance Lyapunov stability, we propose a modified stage cost that preserves the optimal input while guaranteeing asymptotic stability of a specific
equilibrium with a slight performance loss. The approach is further extended to EMPC of a class of linear systems with periodic costs and disturbances by lifting it to a multi-step EMPC problem for periodic operations. A case study with a water distribution network demonstrates the effectiveness of the proposed methods in achieving both asymptotic convergence and stability.
Read the full abstract ▾
Information Theory · Cross-listed from cs.LG
Generalising maximum mean discrepancy: kernelised functional Bregman divergences
Russell Tsuchida and Frank Nielsen
Bregman divergences play a pivotal role in statistics, machine learning and computational information geometry. Particularly in the context of machine learning, they are central to clustering, exponential families, parameter estimation and optimisation, among other things. Despite this, the full toolkit of
Hilbert spaces and in particular reproducing kernel
Hilbert spaces have not been systematically developed and applied to functional Bregman divergences, where points are functions rather than finite-dimensional parameter vectors. While other types of functional Bregman divergences have been studied, these are typically in a
Banach space rather than more directly aligned with kernel methods and
Hilbert-space geometry commonly used in machine learning. We consider functional Bregman divergences on a
Hilbert space, where the self-dual pairing and Riesz representer afford us particularly convenient calculus. Further specialising Bregman generators as a composition involving a kernel mean
embedding makes such divergences easy to estimate. We discuss applications in clustering, universal estimation, robust estimation and generative modelling, and contrast our approach with other types of Bregman divergences.
Read the full abstract ▾
Dynamical Systems · Cross-listed from quant-ph
Lobe Dynamics, Phase-Space Transport, and Non-Adiabatic Leakage Thresholds in the Nonautonomous Kerr-Cat Qubit
Stephen Wiggins
The Kerr-nonlinear parametric oscillator (KPO) provides a foundational semiclassical model for cat-state
quantum hardware. Standard analyses of the KPO typically rely on autonomous, frozen-time approximations to describe the stabilization of macroscopic coherent states. However, state preparation and gate manipulation are driven by explicitly time-dependent microwave pulses, so the operational dynamics are inherently nonautonomous. In this paper, we show that static algebraic
equilibrium pictures are incomplete for describing both state formation and gate-induced transport in the Kerr-cat
qubit. For nonautonomous state preparation, we analyze the ramped resonant model by combining a linear nonautonomous stability analysis with a local invariant-graph reduction near the
vacuum trajectory. This yields a quintic reduced normal form in the critical direction and identifies two symmetric post-threshold moving branches that organize the local state-formation dynamics. The associated diagnostics separate the reduced branch dynamics from the full two-dimensional phase-twist relaxation observed in the hardware coordinates. For gate execution, we model a fast pulse as a weak aperiodic perturbation of the conservative resonant figure-eight separatrix and apply Melnikov's method to derive a leading-order transport criterion. In this framework, transient lobe dynamics emerge as a semiclassical mechanism for non-adiabatic leakage, and the resulting amplitude-width threshold curve provides a leading-order
geometric indicator for the onset of gate-pulse-induced transport.
Read the full abstract ▾
Information Theory · Cross-listed from cs.CL
Improving Robustness of Tabular Retrieval via Representational Stability
Kushal Raj Bhandari, Adarsh Singh, Jianxi Gao et al. (+2 more)
Transformer-based table
retrieval systems flatten structured tables into
token sequences, making
retrieval sensitive to the choice of serialization even when table semantics remain unchanged. We show that semantically equivalent serializations, such as $\texttt{csv}$, $\texttt{tsv}$, $\texttt{html}$, $\texttt{markdown}$, and $\texttt{ddl}$, can produce substantially different
embeddings and
retrieval results across multiple
benchmarks and retriever families. To address this instability, we treat serialization
embedding as noisy views of a shared semantic signal and use its centroid as a canonical target representation. We show that centroid averaging suppresses format-specific variation and can recover the semantic content common to different serializations when format-induced shifts differ across tables. Empirically, centroid representations outrank individual formats in aggregate pairwise comparisons across $\texttt{MPNet}$, $\texttt{BGE-M3}$, $\texttt{ReasonIR}$, and $\texttt{SPLADE}$. We further introduce a lightweight residual bottleneck adapter on top of a frozen encoder that maps single-serialization
embeddings towards centroid targets while preserving variance and enforcing covariance regularization. The adapter improves robustness for several dense retrievers, though gains are model-dependent and weaker for sparse lexical
retrieval. These results identify serialization sensitivity as a major source of
retrieval variance and show the promise of post hoc
geometric correction for serialization-invariant table
retrieval.
Read the full abstract ▾
All authors Kushal Raj Bhandari, Adarsh Singh, Jianxi Gao, Soham Dan, Vivek Gupta
Statistics Theory · Cross-listed from cs.LG
A Limit Theory of Foundation Models: A Mathematical Approach to Understanding Emergent Intelligence and Scaling Laws
Jun Shu, Junxiong Jia, Deyu Meng et al. (+1 more)
Emergent intelligence have played a major role in the modern AI development. While existing studies primarily rely on empirical observations to characterize this phenomenon, a rigorous theoretical framework remains underexplored. This study attempts to develop a mathematical approach to formalize emergent intelligence from the perspective of limit theory. Specifically, we introduce a performance function E(N, P, K), dependent on data size N, model size P and training steps K, to quantify intelligence behavior. We posit that intelligence emerges as a transition from finite to effectively infinite knowledge, and thus recast emergent intelligence as existence of the limit $\lim_{N,P,K \to \infty} \mathcal{E}(N,P,K)$, with emergent abilities corresponding to the limiting behavior. This limit theory helps reveal that emergent intelligence originates from the existence of a parameter-limit architecture (referred to as the limit architecture), and that emergent intelligence rationally corresponds to the learning behavior of this limit system. By introducing tools from nonlinear Lipschitz operator theory, we prove that the necessary and sufficient conditions for existence of the limit architecture. Furthermore, we derive the scaling law of foundation models by leveraging tools of Lipschitz operator and covering number. Theoretical results show that: 1) emergent intelligence is governed by three key factors-training steps, data size and the model architecture, where the properties of basic blocks play a crucial role in constructing foundation models; 2) the critical condition Lip(T)=1 for emergent intelligence provides theoretical support for existing findings. 3) emergent intelligence is determined by an infinite-dimensional system, yet can be effectively realized in practice through a finite-dimensional architecture. Our empirical results corroborate these theoretical findings.
Read the full abstract ▾
All authors Jun Shu, Junxiong Jia, Deyu Meng, Zongben Xu
Numerical Analysis
Numerical Analysis of a Variable-Order Time-Fractional Incompressible Magnetohydrodynamics System
Abdumauvlen Berdyshev, Dossan Baigereyev, Aibek Bakishev et al. (+2 more)
We consider an incompressible magnetohydrodynamics (MHD) model in which the classical first-order time derivatives in the momentum and
magnetic induction equations are replaced by variable-order Caputo time-fractional derivatives. This formulation allows the
memory effect to vary during the
evolution and represents a time-fractional generalization of the incompressible MHD system with nonstationary
memory. To approximate the problem, we use a fully discrete scheme combining a finite element discretization in space with an L1-type approximation of the variable-order Caputo operators in time. For this discretization, we establish a discrete stability estimate and also derive an auxiliary corrected discrete energy estimate for the fully discrete solution. Convergence is proved by showing that the kernels generated by the variable-order L1 approximation satisfy the assumptions of an abstract discrete fractional Grönwall
theorem, which is then applied to the coupled MHD system. The numerical study consists of four parts. First, representative order profiles are used to examine temporal convergence. Second, consistency with the classical incompressible MHD equations is studied as the fractional orders approach one, using norms of solution differences and deviations in kinetic and
magnetic energies. Third, the influence of the variable-order fractional terms on nonlinear
evolution is investigated through the periodic divergence-free
vortex benchmark, with comparisons based on energy and enstrophy histories, divergence errors, Reynolds-number dependence, and time-integrated diagnostics. Fourth, parameter-space maps show how the parameters defining the variable orders affect global indicators. The results show that the variable orders can noticeably affect the
evolution of the energy, enstrophy, and current enstrophy even when the Reynolds number is fixed.
Read the full abstract ▾
All authors Abdumauvlen Berdyshev, Dossan Baigereyev, Aibek Bakishev, Nurlana Alimbekova, Talgat Farkhadov
Analysis of PDEs · Cross-listed from cs.AI
QED: An Open-Source Multi-Agent System for Generating Mathematical Proofs on Open Problems
Chenyang An, Qihao Ye, Minghao Pan et al. (+1 more)
We present QED, an open-source multi-
agent system that turns human-provided research questions into complete mathematical proofs without further human guidance. Its pipeline is designed to overcome common failures of single-query proof generation by separating planning, proving, and verification: a decomposition
agent structures the proof search, prover
agents generate candidate arguments, and verifier
agents check correctness. In collaboration with domain experts, we evaluated QED on 18 research-level projects of varying difficulty. QED produced five original works across algebraic geometry, fluid PDEs, probability, and inverse problems. Expert assessments regard these works as solid specialized research contributions, with three comparable in difficulty and scope to work commonly published in established specialist mathematics venues. QED is released at https://github.com/proofQED/QED.
Read the full abstract ▾
All authors Chenyang An, Qihao Ye, Minghao Pan, Jiayaun Zhang
Statistics Theory · Cross-listed from stat.ME
Neyman Jackknife: Design-Based Variance Estimation for Causal Inference under Interference
Bryan Park and Stefan Wager
We propose a framework, the Neyman Jackknife, for conservative variance estimation in finite-population causal inference under interference. Our approach provides a general, flexible blueprint that enables conservative variance estimation whenever we are able to recompute our target
estimator with some treatment assignments omitted. In classical settings, our approach recovers
estimators closely related to the Neyman
estimator under SUTVA and the Newey-West HAC variance
estimator for time series. Numerical experiments suggest that our general-purpose framework yields variance
estimators that can match or even surpass the performance of baselines that were purpose-built for specific applications.
Read the full abstract ▾
Optimization and Control · Cross-listed from cs.LG
FedSLoP: Memory-Efficient Federated Learning with Low-Rank Gradient Projection
Yutong He, Zhengyang Huang, Jiahe Geng et al. (+1 more)
Federated learning enables a population of clients to collaboratively train machine learning models without exchanging their raw data, but standard
algorithms such as FedAvg suffer from slow convergence and high communication and
memory costs in heterogeneous, resource-constrained environments. We introduce FedSLoP, a
federated optimization
algorithm that combines
stochastic low-rank subspace projections of gradients, thereby reducing the dimension of communicated and stored updates while preserving optimization progress. On the theoretical side, we develop a detailed nonconvex convergence analysis under standard smoothness and bounded-variance assumptions, showing that FedSLoP is guaranteed to converge to a first-order stationary point at a rate of $O(1/\sqrt{NT})$. On the empirical side, we conduct extensive experiments on
federated MNIST classification with heterogeneous data partitions, showing that FedSLoP substantially reduces communication volume and client-side
memory while achieving competitive or better accuracy compared with FedAvg and representative sparse or low-rank baselines. Together, our results demonstrate that random subspace momentum methods such as FedSLoP provide a principled and effective approach to communication- and
memory-efficient
federated learning. Codes are available at: https://github.com/pkumelon/FedSLoP.git.
Read the full abstract ▾
All authors Yutong He, Zhengyang Huang, Jiahe Geng, Kun Yuan
Numerical Analysis
Adaptive-Distribution Randomized Neural Networks for PDEs: A Low-Dimensional Distribution-Learning Framework
You Yang and Fei Wang
Randomized neural networks (RaNNs) are attractive for partial differential equations (PDEs) because they replace expensive end-to-end training with a linear least-squares solve over randomized hidden features. Their practical performance, however, depends strongly on the sampling distribution of the hidden-layer parameters, which is usually chosen heuristically and problem by problem. This distribution sensitivity is a central bottleneck in randomized neural PDE solvers. In this work, we propose Adaptive-Distribution Randomized Neural Networks (AD-RaNN), a framework that promotes randomized feature generation from a fixed heuristic choice to a low-dimensional adaptive optimization problem. Instead of training all hidden weights and biases, AD-RaNN parameterizes the hidden-feature sampling distribution by a low-dimensional vector p and optimizes only p, thereby preserving the least-squares structure of RaNNs while reducing manual distribution tuning. The method uses a two-stage strategy: ridge-regularized reduced training for stable distribution-parameter optimization, followed by an unregularized least-squares refit for final solution recovery. We develop two adaptive mechanisms, PDE-Driven Adaptive Distribution (PDAD) and Data-Driven Adaptive Distribution (DDAD), and deploy them in space-time solvers, discrete-time solvers, and operator-learning models. We also incorporate an adaptive layer-growth enhancement for localized structures. For the reduced optimization problem, we establish well-posedness of the reduced objectives, consistency of ridge-regularized minimizers, an efficient gradient formula, and a practical lower-bound estimate for the ridge parameter. Numerical experiments on
benchmark problems show that AD-RaNN provides an effective distribution-level adaptation mechanism, reduces reliance on hand-crafted hidden-feature distributions, and achieves strong empirical accuracy.
Read the full abstract ▾
Algebraic Geometry
The Birational Invariance Of Fundamental Group Schemes
Lingguang Li and Hao Wang
Let $k$ be a field, $f \colon X \to Y$ a birational morphism of integral connected schemes proper over $k$ with $Y$ normal, $x \in X(k)$ lying over $y \in Y(k)$. For Tannakian categories $\mathcal{C}_X \subset \mathfrak{Vect}(X)$ and $\mathcal{C}_Y \subset \mathfrak{Vect}(Y)$, denote by $\pi(\mathcal{C}_X,x)$ and $\pi(\mathcal{C}_Y,y)$ the corresponding Tannaka group schemes. We establish a unified Tannakian criteria for the natural homomorphism $\pi(\mathcal{C}_X,x)\to \pi(\mathcal{C}_Y,y)$ to be an isomorphism. As applications, for a birational map $X \dashrightarrow Y$ between smooth projective varieties over a perfect field $k$, we prove that there exists a natural isomorphism $\pi^{*}(X,x)\cong \pi^{*}(Y,y)$ for any $* \in \{S,N,EN,F,EF,Loc,ELoc,\acute{e}t, E\acute{e}t,uni\}$. In particular, we prove that the induced homomorphism $\pi^{str}(X,x)\to \pi^{str}(Y,y)$ is an isomorphism for any birational morphism $ X \rightarrow Y$.
Read the full abstract ▾
Dynamical Systems · Cross-listed from eess.SY
Extracting Exact Lie Derivatives Without Backpropagation: A Dual Compiler for Neural Control Barrier Functions
Mohammadreza Kamaldar
A safety filter based on a neural control barrier function (CBF) deployed in an embedded control loop evaluates, at each control cycle, the trained network and its Lie derivatives along the system vector fields, under the
memory and worst-case execution time (WCET) constraints that safety-oriented coding standards impose. Reverse-mode automatic differentiation, by which training frameworks obtain these derivatives, retains an activation cache whose size grows with the sum of the layer widths, and general-purpose differentiation runtimes allocate the computational graph from the heap at each call. This paper presents a compiler that evaluates a neural CBF and its exact Lie derivatives by forward-mode dual-number arithmetic. The compiler emits self-contained C++ code in which a single forward pass, without backpropagation, returns the barrier value and its exact Lie derivative along a given vector field; the drift and input Lie derivatives of the safety constraint are obtained from one such pass per vector field, and a second-order extension based on hyper-dual numbers returns the exact second-order Lie derivatives required by CBFs of relative degree two. The dual forward pass requires a workspace bounded by four times the widest layer, independent of network depth, and the emitted code contains no allocation call sites, so the absence of dynamic allocation is verifiable by inspection of the code. On an ESP32-S3 microcontroller, the compiled filter assembles the complete safety constraint in under one millisecond from statically allocated buffers of at most 768 bytes, and the maximum execution time over 1000 evaluations lies within 5% of the median in all three examples, whereas a heap-allocating reverse-mode baseline shows maxima 33% and 70% above its median in the two first-order examples. The compiler and the embedded experiments are released as open-source software.
Read the full abstract ▾