Peer Revenue Past citations are no guarantee of future returns

Papers

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

Mathematics

math Watch Mathematics (sign in to use)

Monday 4 May

Dynamical Systems

Continuity of Lyapunov Exponent for Quasi-Periodic Gevrey Cocycles

It is shown that for the quasi-periodic cocycles in Gevrey space $G^{s}$ and subexponential Brjuno class frequency $\Omega(\eta)$, the Lyapunov exponent is continuous provided that $1<s+\eta<2$.

Optimization and Control · Cross-listed from cs.RO

Sampling-Based Control via Entropy-Regularized Optimal Transport

Sampling-based model predictive control methods like MPPI and CEM are essential for real-time control of nonlinear robotic systems, particularly where discontinuous dynamics preclude gradient-based optimization. However, these methods derive from information-theoretic objectives that are agnostic to the geometry of the control problem, leading to pathological behaviors such as mode-averaging when the cost landscape is complex. We present OT-MPC, a sampling-based algorithm that overcomes these limitations through an entropy-regularized optimal transport formulation. By computing an optimal coupling between candidate control sequences and low-cost proposals, OT-MPC refines candidates toward nearby promising samples while coordinating updates across the ensemble to maintain coverage of the solution space. We derive closed-form, gradient-free updates via the Sinkhorn algorithm, enabling real-time performance. Experiments on navigation, manipulation, and locomotion tasks demonstrate improved success rates over existing methods.
Read the full abstract

Classical Analysis and ODEs

Inhomogeneous Picard-Fuchs equations of Abelian integrals in piecewise smooth near-Hamiltonian systems

In this paper, we explicitly obtain inhomogeneous Picard-Fuchs equations for Abelian integrals $I_{i,j}^+(h)$, where $I_{i,j}^+(h)$ is an integral along orbital arcs defined by polynomials $\frac{1}{2}y^2 + F(x)=h$. Moreover, we discuss the method of using Picard-Fuchs equations to recursively compute the asymptotic expansions of genearating functions of Abelian integrals near a homoclinic loop. As an application, we derive the maximum number of isolated zeros of Melnikov functions near a nilpotent saddle homoclinic loop for piecewise polynomials perturbations with the inclination $\theta$ of the separation line as a free parameter.
Read the full abstract

Statistics Theory · Cross-listed from cs.LG

On the Optimal Sample Complexity of Offline Multi-Armed Bandits with KL Regularization

Kullback-Leibler (KL) regularization is widely used in offline decision-making and offers several benefits, motivating recent work on the sample complexity of offline learning with respect to KL-regularized performance metrics. Nevertheless, the exact sample complexity of KL-regularized offline learning remains largely from fully characterized. In this paper, we study this question in the setting of multi-armed bandits (MABs). We provide a sharp analysis of KL-PCB (Zhao et al., 2026), showing that it achieves a sample complexity of $\tilde{O}(\eta SAC^{\pi^*}/\epsilon)$ under large regularization $\eta = \tilde{O}(\epsilon^{-1})$, and a sample complexity of $\tilde{\Omega}(SAC^{\pi^*}/\epsilon^2)$ under small regularization $\eta = \tilde{\Omega}(\epsilon^{-1})$, where $\eta$ is the regularization parameter, $S$ is the number of contexts, $A$ is the number of arms, $C^{\pi^*}$ policy coverage coefficient at the optimal policy $\pi^*$, $\epsilon$ is the desired sub-optimality, and $\tilde{O}$ and $\tilde{\Omega}$ hide all poly-logarithmic factors. We further provide a pair of sharper sample complexity lower bounds, which matches the upper bounds over the entire range of regularization strengths. Overall, our results provide a nearly complete characterization of offline multi-armed bandits with KL regularization.
Read the full abstract

All authors Kaixuan Ji, Qiwei Di, Heyang Zhao, Qingyue Zhao, Quanquan Gu

Combinatorics · Cross-listed from cs.DM

Improving SAT Solvers on Orthogonal Latin Square Problems

Latin squares are $n\times n$ matrices containing $n$ symbols, where each symbol appears exactly once in each row and column. They were studied by Euler, later popularized through Sudoku, and remain a rich source of difficult combinatorial search problems. Two Latin squares are orthogonal mates if, when overlaid, no ordered pair of symbols repeats. Pairs of orthogonal Latin squares exist for every order except 2 and 6, but finding orthogonal Latin squares computationally can be challenging. Satisfiability (SAT) solvers are strong at combinatorial search and have been used to resolve a number of various kinds of orthogonal Latin square problems. On the other hand, SAT solvers lack domain knowledge about Latin squares, such as the Euler-Parker algorithm for orthogonal mate construction. In this paper, we propose a hybrid method combining a SAT solver with the Euler-Parker algorithm (implemented using a Diophantine system solver) and show that the resulting solver is effective at finding certain kinds of orthogonal Latin squares. For example, certain pairs of $10\times10$ orthogonal Latin squares whose existence was unknown for over 25 years were recently found by Bright, Keita, and Stevens using a SAT solver. The hardest cases could not be solved by the SAT solver CaDiCaL within seven days, but CaDiCaL augmented with an external Euler-Parker algorithm solves these cases in a median of around 5,100 seconds.
Read the full abstract

Algebraic Geometry

From Diaz's Enriques Product to an $n$-Fold Cup-Product Bockstein Family of Integral Hodge Counterexamples

We reinterpret Diaz's construction of Chow-trivial smooth projective varieties violating the integral Hodge conjecture as the level-two case of an \(n\)-fold cup-product Bockstein mechanism. Diaz's dimension-four example is \(V=S_1\times S_2\), where \(S_1,S_2\) are Enriques surfaces, and its obstruction is the Bockstein of $\pi_1^*\alpha_1\cup\pi_2^*\beta_2\in H^3(V,\mathbb Z/2(2))$. Here \(\alpha_1\) is the K3 double-cover class and \(\beta_2\) is an Enriques Brauer-detecting class. We extend the finite-coefficient source construction to \(X_n=S_1\times\cdots\times S_n\) by forming $\Theta_n=\pi_1^*\alpha_1\cup\pi_2^*\beta_2\cup\cdots\cup\pi_n^*\beta_n \in H^{2n-1}(X_n,\mathbb Z/2(n))$,with Bockstein $\Delta_n=\delta(\Theta_n)\in H^{2n}(X_n,\mathbb Z(n))$. Using external products of perverse sheaves, categorical Bockstein compatibility, and a Leibniz rule for the MacPherson–Vilonen boundary, we prove unconditionally that \(\Delta_n\) has nonzero image in a distinguished Enriques–Brauer component of the MV obstruction channel. Under the Brauer-separation hypothesis, which asserts that algebraic codimension-\(n\) cycle classes have zero image in this same component, the class \(\Delta_n\) is a non-algebraic \(2\)-torsion integral Hodge class. We verify this separation for decomposable algebraic cycles and reduce the remaining non-decomposable case, via integral even Chow–Künneth projectors on the Enriques factors, to a single coefficient-level algebraic-control problem involving the \(H^1(S_1,\mathbb Z/2(1))\) Enriques double-cover direction. We also record a motivic finite-coefficient lift of the tower via the finite-coefficient cone \(\mathbf 1_X(n)/2:=\operatorname{Cone}(\mathbf 1_X(n)\xrightarrow{\times 2} \mathbf 1_X(n))\), and explain which formal part of the MacPherson–Vilonen zig-zag construction lifts motivically under Betti realization.
Read the full abstract

Optimization and Control

A Parameter-Free First-Order Algorithm for Non-Convex Optimization with $\tilde{\mkern1mu O}(\epsilon^{-5/3})$ Global Rate

We introduce PF-AGD, the first parameter-free, deterministic, accelerated first-order method to achieve $O(\epsilon^{-5/3}\log(1/\epsilon))$ oracle complexity bound when minimizing sufficiently smooth, non-convex functions; this is the best-known bound for first-order methods on smooth non-convex objectives. Unlike existing methods possessing this rate that require a priori knowledge of smoothness constants, we use an adaptive backtracking scheme and a gradient-based restart mechanism to estimate local curvature. This yields a practical algorithm that matches best-known theoretical rates. Empirically, PF-AGD outperforms the practical variant of AGD-Until-Guilty (Carmon et al., 2017), as well as other parameter-free variants, and is a viable alternative to nonlinear conjugate gradient methods.
Read the full abstract

Probability · Cross-listed from cs.LG

Soft-to-Hard Routing in Sparse Mixture-of-Experts Models

Softmax routing approaches hard top-1 routing as the temperature tends to zero, but the limiting passage is singular at router ties. This paper develops a boundary-layer calculus for this soft-to-hard limit in population squared-loss mixture-of-experts regression. For a router with logits $a_k(x;\phi)$, the relevant local quantity is the top-two margin $\Delta(x;\phi)$, and the relevant global quantity is the boundary mass $\mathbb{P}(\Delta(X;\phi)\le w)$. Under smoothness and transversality assumptions, coarea and tubular-neighborhood estimates show how this mass scales with the slab width; in the binary case the leading coefficient is an explicit surface integral over the routing interface. These geometric estimates give quantitative bounds between the soft objective $L_\tau$ and the hard objective $L_0$, including an $O(\tau^\alpha)$ uniform comparison under a margin-tail condition, and yield $\Gamma$-convergence of the soft objectives on compact parameter spaces. The main conclusion is that the zero-temperature approximation is controlled by the probability carried by an $O(\tau)$ neighborhood of the routing interfaces, not by temperature alone. After isolating this boundary-layer part of the problem, we record a conditional landscape-transfer theorem from hard to small-temperature soft routing and a reduced two-expert Gaussian calculation illustrating local symmetry breaking. Synthetic diagnostics are included only as controlled checks of the boundary-layer predictions.
Read the full abstract

Functional Analysis

A lifting theorem for operators between Lipschitz spaces

We prove that every bounded linear operator between Lipschitz spaces admits a lifting along the de Leeuw embedding. More precisely, given pointed metric spaces $M$ and $N$ and $\epsilon>0$, every bounded linear operator $S:\mathrm{Lip}_0(M)\to \mathrm{Lip}_0(N)$ admits a lifting $\mathfrak{S}:C(\beta \widetilde{M})\to C(\beta \widetilde{N})$ such that $\|\mathfrak{S}\|\leq \|S\|+\epsilon$ and $\mathfrak{S}(\varPhi_M(f))=\varPhi_N(S(f))$ for every $f\in \mathrm{Lip}_0(M)$. Moreover, compact operators admit compact liftings.
Read the full abstract

History and Overview of Mathematics

Robert V. Kohn (1953-2026)

The article is dedicated to the memory and enduring legacy of Professor Robert V. Kohn, Courant Institute, NYU. In this memorial article, we record thoughts and reminiscences of his exemplary life.

Category Theory

Geometric Categories and Sheaves on Topoi

We introduce the notion of a geometric $(\infty,1)$-category, the protopyical example of which is an $(\infty,1)$-topos. We study (hyper)sheaves on geometric $(\infty,1)$-categories, proving that these are characterized by a form of Čech (hyper)descent. As an application we study (hyper)sheaves on $(n,1)$-topoi for all $n\in \mathbf{Z}_{\geq 1}\cup \{\infty\}$, and prove that the effective epimorphism topology on an $(n,1)$-topos $\mathcal{X}$ may be identified as the canonical topology on $\mathcal{X}$. Moreover, we show that for finite $n\in \mathbf{Z}_{\geq 1}$ the study of sheaves on an $(n,1)$-topos $\mathcal{X}$ is equivalent to the study of $(n-1)$-truncated sheaves on certain $(\infty,1)$-topoi. We then globalize our study to consider sheaves on $\infty\mathcal{T} op$. In the appendix, we study the behavior of modules under a reflective monoidal $(\infty,1)$-functor $L^\otimes:\mathcal{C}^{\otimes}\rightarrow \mathcal{D}^{\otimes}$, and study (hyper)sheafification under a change of universe.
Read the full abstract

Differential Geometry · Cross-listed from cs.LG

Geometric and Spectral Alignment for Deep Neural Network II

This paper develops the angular and static-channel component of Geometric and Spectral Alignment for residual Jacobian chains. Starting from Cartan-coordinate rigidity and fitted effective-rank windows, we study how dominant singular subspaces are transported across adjacent layers and how the resulting finite matrices can be displayed in physical channel coordinates. The main results are deterministic, margin-verified results. We bound the error between full interface transport and its dominant-window truncation, add fitted-tail errors so that empirical spectra can be certified against the Gibbs–Cartan tail model, and distinguish source-mode incidence from fully physical input-output channel incidence. Given row groups and active supports, the Physical Alignment Matrix decomposes orthogonally as core plus overlap plus noise. Active-column gaps, pairwise overlap margins, and noise bounds combine into a static certificate radius under which the full transport and the truncated transport induce the same active supports, pairwise incidence graph, SRS sets, hub columns, and core/overlap/noise masks. The finer SC/SA/ST labels of the Invariant Channel Mapping require additional row-energy and profile-correlation margins, stated as explicit perturbation tests. The empirical section reports the matrices and block-energy heatmaps that measure these certificate quantities across CNNs, language models, and vision/diffusion backbones. The figures are interpreted as finite-dimensional measurements; complete membership in the Physical GSA certificate domain requires checking the numerical margin protocol stated in Section 10.
Read the full abstract

All authors Ziran Liu, Wei Wang, Jinhao Wang, Pengcheng Wang, Xinyi Sui, Cihan Ruan, Nam Ling, Wei Jiang

Differential Geometry · Cross-listed from cs.LG

Geometric and Spectral Alignment for Deep Neural Network I

Deep residual architectures are modeled as products of near-identity Jacobians. This paper proves deterministic quotient-geometric estimates for singular spectra of Frobenius-normalized layer factors, emphasizing a normalized top-radial Cartan coordinate and fitted power-law chart. Full-rank factors are mapped from $\mathrm{GL}(d)$ to the positive cone by $A\mapsto A^\top A$, then to ordered eigenvalue data. Under Frobenius normalization, exact power-law spectra form a trace-normalized Cartan orbit. This orbit is a Gibbs family on ranks, a Fisher information line, and a Bures–Wasserstein curve with line element $d/4$ times Fisher information. The main rigidity theorem is a slack-aware margin inequality: interface radial amplitude, non-backtracking slack, and signed residual variation control displacement of the fitted Cartan coordinate. In the exact-chart zero-slack case, a depth-$L$ budget gives exponent drift of order $(\log M)/L$; generally, slack and residual increments augment the bound. We separate scalar top-radial from full-Cartan spectral control, which also needs Bures/Hellinger residual variation. We prove approximate-power-law and metric-chart versions, converse lower bounds, Fisher–KL/Bures action estimates, and near-identity expansions for normalized residual chains. Near-identity results verify transport budgets; chart quality remains measurable. Effective rank is a spectral-energy quantile, giving finite-width power-law tail bounds and robust rank-window transition estimates. Empirical static-weight exponent profiles serve as diagnostics; full verification also requires interface budgets, slacks, and residuals for the same operator chain.
Read the full abstract

All authors Ziran Liu, Wei Wang, Jinhao Wang, Pengcheng Wang, Xinyi Sui, Cihan Ruan, Nam Ling, Wei Jiang

Probability

Chart-Generated Probability Geometry and Kolmogorov Expectations

This paper develops a geometric reinterpretation of probability in which a cumulative distribution function is not used merely as a passive coordinate label, but as a generator of metric geometry. An admissible probability chart $G:I\to(0,1)$ pulls the Euclidean distance of the probability interval back to value space. This creates a decisive distinction from ordinary coordinate invariance: if $G$ is replaced by another chart $H$ while the observations and their law are kept fixed, the induced metric-measure structure changes. Probability charts preserve ordinal structure while altering metric notions such as distance, dispersion, boundary proximity, and barycentric centrality. Averaging linearly in the chart coordinate and pulling the result back defines the Fréchet barycenter of $X$ under $d_G$. These functionals coincide with Kolmogorov-Nagumo means, but the chart-generated viewpoint explains their chart dependence geometrically: changing the chart changes the loss being minimized, rather than merely rewriting one fixed centre in new coordinates. A rigidity result shows that, within normalized probability charts, preserving the barycenter for every law forces the chart itself to remain unchanged. Under the intrinsic chart the coordinates are uniform and the barycenter is the median; under an external benchmark chart, tail mismatch is represented by excess boundary occupation. Laws of large numbers, central limit theorems, and a non-asymptotic concentration inequality are established for the transformed functionals. Probability-coordinate moments exist for every law supported on the chart domain and determine the law uniquely, in contrast with classical moment non-existence and indeterminacy on unbounded value spaces.
Read the full abstract

Sunday 3 May

General Mathematics

On the Constructive Dimension Spectrum of Polynomials

Recently, Stull [18], [17] resolved a long-standing open problem posed by Lutz, on whether the set of effective Hausdorff dimensions of points on a straight line in $\mathbb{R}^2$ – the effective dimension spectrum of the line – contains a unit interval. This question is related to problems in classical fractal geometry like the Kakeya conjecture and Furstenberg sets. Stull posed an open question on the dimension spectra of polynomial curves. For the first result, with new techniques which adapt the theory of classical real root-finding of polynomials to the current setting, we show that the dimension spectra of every polynomial curve contains at least two points. This answers an open question posed by Stull [18], [17]. We use the main result to construct a class of polynomials which have width strictly greater than 1, answering a second problem stated in [18],[17]. Stull [18] resolved the dimension spectrum conjecture for planar lines, showing that it contains a unit interval. For the second result, we resolve the conjecture for a subfamily of polynomials whose coefficients form a "low" dimension point in $\mathbb{R}^{d+1}$.
Read the full abstract

Information Theory · Cross-listed from q-bio.NC

Information as Maximum-Caliber Deviation: A bridge between Integrated Information Theory and the Free Energy Principle

The Free Energy Principle (FEP) is a leading framework for mathematically modeling self-organization and learning, while Integrated Information Theory (IIT) is a computational ontology of consciousness oriented around irreducible cause and effect. While conceptual unifications have been proposed and appear to be supported by empirical findings, the absence of a rigorous mathematical mapping places upper bounds on their precision and testability. This work proposes that information can be defined as the deviation $\psi$ of realized dynamics from a constrained maximum-caliber (MaxCal) path ensemble over a finite time horizon. Under this definition, each of the cause/effect repertoires central to IIT 3.0 emerge directly from MaxCal variational principles, allowing IIT's phenomenological calculus to be re-derived from constrained entropy-maximization (CMEP). This framework supplies a theoretical bridge to active inference, which is mathematically dual to CMEP under Langevin dynamics, and offers a principled route for extending IIT to new dynamical regimes. When the approach is applied under the Central Limit Theorem (CLT) for Markov chains and via large deviations theory (LDT) to Ising models, information $\psi$ is shown to be equivalent to prediction error under accompanying predictive coding models. This may hold relevance to the “hill-shaped trajectory” of $\Phi$ observed in neuronal cultures adapting to sensory inputs. Together, these results provide a physically and mathematically grounded rationale for studying the convergence of FEP, IIT, and thermodynamic frameworks of cognition such as recent work grounding consciousness in violations of the Fluctuation-Dissipation Theorem (FDT).
Read the full abstract

General Mathematics

Discontinuity at the fixed point in suprametric spaces

The aim of this paper is to generalize some fixed point theorems in the class of convex contraction of order $m$ on a complete suprametric space. Then, we will prove that the class of convex contraction of order m is strong enough to generate a fixed point on a complete suprametric spaces but do not force the mapping to be continuous at the fixed point, and it can be replaced by relatively weaker conditions of $k$-continuity or $T$-orbitally lower semi-continuous. On this way a new and distinct solution to the open problem of Rhoades (Contemp Math 72:233-245,1988) is found. In sequel, we will prove some fixed point results in the setting suprametric spaces which are generalizations of the results regarding Sehgal, Ćirić and Fisher's quasi-contraction. Some examples and application will be approved our results.
Read the full abstract

Dynamical Systems

Non-Monotone Response Modules and Cascades from the EML Operator for Reduced Models of Biological Dynamics

Standard saturating response functions, such as the Hill function, are monotone and therefore cannot represent recruitment-induced overshoot or adaptive transients with a single block. Reproducing such non-monotone responses from saturating primitives requires at least a difference of two blocks with opposing amplitudes, doubling the static-block parameter count. Here, building on a recent mathematical result that a single binary operator, EML, generates all standard elementary functions, we use EML as a structured grammar for reduced nonlinear ODEs. This yields an activation-suppression module that captures overshoot directly. We validate the framework in three settings. First, on PKA-R relocalization data, the EML grammar discovers a reduced surrogate consistent with established mechanistic biology. Second, on Rho-GTPase recruitment data, an exhaustive search over EML expression trees selects the same compositional form across all four perturbation-response traces. Third, a 50-state simulated network is compressed by an EML cascade acting as a fixed temporal basis. Thus we demonstrate the power and potential of EML for reduced models of biological dynamics.
Read the full abstract

Optimization and Control · Cross-listed from cs.LG

Analytic Bridge Diffusions for Controlled Path Generation

Most modern bridge-diffusion methods achieve finite-time transport by specifying an interpolation, Schrodinger-bridge, or stochastic-control objective and then learning the associated score or drift field with a neural network. In contrast, we identify a restricted but sufficiently broad analytically solvable class in which, for a deterministic source and a Gaussian-mixture target, the score and all intermediate marginals are explicit and protocol objectives of the type used in this paper can be differentiated without inner stochastic simulation loops. We recast the classical linear–quadratic–Gaussian stochastic-control structure as a transport problem of the Path Integral Diffusion type. Linear dynamics, Gaussian noise, and quadratic running costs reduce the bridge calculation to a matrix Riccati cascade, while the terminal state cost is replaced by a prescribed Gaussian-Mixture terminal probability density. Linear Quadratic – Gaussian Mixture – Path Integral Diffusion (LQ-GM-PID) thereby turns bridge diffusion from terminal target matching alone into an analytically controlled laboratory for path shaping. We demonstrate this on a 2D corridor task, a 2D multi-entrance task, and a high-dimensional study reaching d=32 and M=16 terminal modes in separate scaling sweeps. We position LQ-GM-PID as an analytically solvable reference model in which score approximations, path-shaping objectives, and protocol-learning procedures can be tested against explicit quantities.
Read the full abstract

Statistics Theory

Entropic Strict Minimum Message Length and Its Connections to PAC-Bayes and NML

We introduce entropic strict minimum message length (SMML), a risk-sensitive generalization of strict minimum message length coding. The proposed criterion replaces expected two-part codelength under the prior predictive distribution with an exponential certainty equivalent, thereby defining a one-parameter family of coding rules that interpolates between Bayesian average-case coding and worst-case minimax coding. We show that ordinary SMML is recovered in the risk-neutral limit, while the extreme risk-sensitive limit yields a minimax codelength criterion. Applying the same entropic soft maximum to regret relative to the oracle maximum likelihood codelength recovers the normalized maximum likelihood (NML) minimax-regret principle. We further prove that entropic SMML admits a variational characterization as a Kullback–Leibler-regularized worst-case expected codelength, giving it a PAC–Bayes-type interpretation. We establish joint \(n\)–\(\tau\) asymptotics that identify how the risk parameter must scale with sample size in order to recover Bayesian average-case, intermediate robust, and worst-case minimax coding behavior. For regular exponential families, the fixed-codebook partition remains affine in sufficient-statistic space, while the codepoints satisfy a tilted moment-matching condition and admit an interpretation as tilted Bregman centroids. These results position entropic SMML as an information theoretic bridge between MML, PAC–Bayes, and MDL.
Read the full abstract

Information Theory · Cross-listed from quant-ph

Separability from Multipartite Measures

We show that multi-invariants provide necessary and sufficient criteria for full separability in finite-dimensional multipartite systems. For tripartite three-qubit states, the third-order negativity gives an exact criterion for full separability for both pure and mixed states. We then extend the framework to four-qubit pure states, whose full characterization requires eighteen independent real parameters. We identify eighteen independent entanglement measures, consisting of six bipartite, eight tripartite, and four quadripartite measures, with the third-order negativity serving as a key separability criterion. Finally, we generalize the construction to multipartite qudit systems and discuss an application to conformal field theory.
Read the full abstract

Probability

An analytic approach to the finite R-transform

We revisit Marcus' finite free analogue of Voiculescu $R$-transform from an analytic viewpoint. By relating the finite free Fourier transform to the Laplace transform, we study the finite $R$-transform through logarithmic potentials and Legendre transforms. Under suitable assumptions, we prove that the finite $R$-transform of a polynomial differs from the Voiculescu $R$-transform of its empirical root distribution by $O(N^{-1})$. As an application, we obtain an analytic proof of the convergence of finite free additive convolution to free additive convolution.
Read the full abstract

Group Theory

Some families of locally graded groups with finitely many orbits under automorphisms

In this work, we study three families of locally graded groups with finitely many orbits under automorphisms. We prove that: (i) a residually finite group with finitely many orbits under automorphisms is locally finite and has finite exponent; (ii) a finitely generated locally graded group with finitely many orbits under automorphisms is finite; and (iii) the Mal'cev $\mathbb{Q}$-completion of an $r$-generated free nilpotent group of class $c$ has finitely many orbits under automorphisms if and only if either $r = 2$ and $c = 3$, or $c \leq 2$
Read the full abstract

Statistics Theory · Cross-listed from econ.EM

Fast Monte-Carlo

This paper proposes an eigenvalue-based small-sample approximation of the celebrated Markov Chain Monte Carlo that delivers an invariant steady-state distribution that is consistent with traditional Monte Carlo methods. The proposed eigenvalue-based methodology reduces the number of paths required for Monte Carlo from as many as 1,000,000 to as few as 10 (depending on the simulation time horizon $T$), and delivers comparable, distributionally robust results, as measured by the Wasserstein distance. The proposed methodology also produces a significant variance reduction in the steady-state distribution.
Read the full abstract

Numerical Analysis

On the Practical Impact of Local Linear Instabilities in Entropy-Stable Schemes

Local linear instability refers to the linearized discrete operator exhibiting perturbation growth exceeding that of the corresponding continuous linearized problem. In the context of nonlinear entropy-stable discretizations, we argue that local linear instabilities should be interpreted as a source of numerical error whose practical impact is often negligible compared with other discretization errors. For split-form discretizations of the variable-coefficient linear advection equation, such as those resulting from linearizations of entropy-stable discretizations of the Burgers equation, perturbations can indeed exhibit unphysical modal growth. However, we demonstrate that this growth satisfies physically interpretable bounds and is typically small. Furthermore, through modified-equation analysis and numerical experiments, we show that the growth is dominated by highly oscillatory and boundary-localized unphysical modes, and can therefore be readily controlled by small amounts of numerical dissipation. More generally, this modal perturbation growth does not extend directly to nonlinear two-point-flux discretizations of the type used in entropy-stable discretizations of the Euler equations. Floquet analysis demonstrates that unstable spectra of frozen-baseflow Jacobians need not lead to unstable perturbation growth. Using the geometric flux for the variable-coefficient linear advection equation, we derive a sharp perturbation growth bound predicting negligible growth, then show analogous behaviour for the logarithmic flux numerically. Finally, we argue that robustness issues observed for entropy-stable schemes in density-wave problems are better attributed to poor near-vacuum behaviour of the logarithmic mean than to local linear instabilities. Overall, our results suggest that local linear instabilities do not pose a practical obstacle to the use of high-order entropy-stable schemes.
Read the full abstract

Differential Geometry

Geometric Reductions of the $G_2$-Hilbert Functional via Circle Actions

In this paper, we study critical points and gradient flows of the $G_2$–Hilbert functional on a manifolds with free $\mathbb S^1$–actions. We analyze $\mathbb S^1$–invariant $G_2$–structures under the constant fiber-length non-Kähler transverse ansatz, reducing the variational problem to the $6$–dimensional quotient and we also consider a Gibbons–Hawking-type ansatz with varying fiber length and derive the formal negative $L^2$–gradient flow. We conclude that the unnormalized flow admits only trivial stationary configurations: flat connection, scalar-flat base metric, and constant fiber length.
Read the full abstract

Group Theory

Higher Commutativity in Finite Groups: Exact Asymptotics and Finite Spectrum

For a finite group G, we study the higher commuting probabilities, namely the probabilities that r randomly chosen elements of G commute pairwise, together with the corresponding numbers of simultaneous conjugacy classes of commuting r-tuples. We prove an exact dominant asymptotic for the number of homomorphisms from the free abelian group of rank r to G. The exponential base is the maximum order of an abelian subgroup of G, and the leading coefficient is the number of abelian subgroups of that order. As a consequence, the r-th root of the higher commuting probability tends to this maximum abelian-subgroup order divided by the order of G, while the r-th root of the orbit count tends to the maximum abelian-subgroup order itself. We also prove that the associated rank-generating series is rational and has a finite Dirichlet-spectrum expansion supported on abelian subgroup indices. This spectrum yields a finite linear recurrence, a finite-rank Hankel matrix, and an inverse finite-spectrum theorem: the tail of the hierarchy determines the full abelian-index spectrum. For split abelian extensions, we express the dominant base through fixed-subgroup geometry, and for abelian acting quotients, we obtain an exact subgroup-lattice formula. In the cyclic and coprime cases, this gives closed formulas for all spectral coefficients.
Read the full abstract

Statistics Theory

Sharp regret-Hellinger bounds for Gaussian empirical Bayes via polynomial approximation

A central problem in the theory of empirical Bayes is to control the regret (excess risk) of a learned Bayes rule by the Hellinger distance between the estimated and true marginal densities. In the normal means model, the classical result of Jiang and Zhang (2009) achieves this only after regularizing the Bayes rule and incurs an extraneous cubic logarithmic factor through a delicate recursive argument. This paper introduces a new technique, based on polynomial approximation and Bernstein-type inequalities for weighted $L_2$ norms, that bounds the unregularized regret directly. The method is conceptually simpler and yields sharper, sometimes optimal, regret bounds. For compactly supported priors, we prove the sharp bound that the regret is $O(\epsilon^2 \frac{\log(1/\epsilon)}{\log\log(1/\epsilon)})$, where $\epsilon$ is the Hellinger distance between the marginal densities. The same method also extends to priors with exponential tails. Conversely, we show that regularization is genuinely necessary for heavy-tailed priors under only bounded moment assumptions. As statistical consequences, we obtain improved regret bounds for the nonparametric maximum likelihood estimator (NPMLE). Notably, for compactly supported priors, by determining the optimal Hellinger rate of mixture density estimation, we show that the optimal regret for sample size $n$ scales as $\Theta(\frac{1}{n}(\frac{\log n}{\log\log n})^2)$, attained by the NPMLE within a $\log\log n$ factor.
Read the full abstract

Dynamical Systems

Realizing Saddle-Node Bifurcations from Finite Data

Given a finite set of data generated by an unknown ordinary differential equation it is impossible to exactly determine the associated vector field, and hence, bifurcation theory tells us that it is impossible, in general, to correctly characterize the underlying dynamics. In this paper, we bypass the effort of obtaining an analytic approximation of the vector field, and we adopt an approach based on Occam's razor: identify the simplest robust characterization of the dynamics that is compatible with the given data. Our fundamental assumption is that the data allows for the construction of an isolating block over a parameter space whose homological Conley index is consistent with a saddle-node bifurcation. Our main result establishes that, for phase spaces of dimension greater than or equal to 6, the original vector field can be smoothly deformed into a canonical model exhibiting exactly one structurally stable saddle-node bifurcation. Crucially, this deformation leaves the vector field unaltered outside the isolating block, ensuring strict compatibility with the observed data.
Read the full abstract

Combinatorics

Faster Mixing for Triangulations via Transport Flows

We prove an $\widetilde O(n^2)$ bound for the relaxation time and the log-Sobolev time (inverse log-Sobolev constant) of the classical triangulation flip chain on a convex $(n+2)$-gon, implying a mixing time of $\widetilde O(n^2)$. The previous state of the art for the mixing time of this chain, due to Eppstein and Frishberg, was $\widetilde O(n^3)$, while the best known lower bound on the mixing time, due to Molloy, Reed, and Steiger, is $\Omega(n^{3/2})$. Our relaxation time bound makes significant progress towards Aldous' conjectured bound of $\Theta(n^{3/2})$ for the relaxation time. We improve upon the analysis of Eppstein and Frishberg by further developing the framework of transport flows introduced in the work of Chen et al. In this light, our results can be seen as a more efficient way of using combinatorial decompositions to obtain functional inequalities for Markov chains. We hope our ideas will find other applications in the future.
Read the full abstract

All authors Vedat Levi Alev, Daniel Frishberg, Michail Sarantis, Prasad Tetali