Peer Revenue Now accepting positions

Papers

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

All of arXiv

Sunday 8 November

Data Structures and Algorithms

Linear Set Reconciliation with $O(d)$ Additive Redundancy

In set reconciliation, two parties holding sets $A, B \subseteq [n]$ with $|A \triangle B| \le d$ seek to recover their symmetric difference from short $\mathbb{F}_2$-linear sketches, which is equivalent to recovering a $d$-sparse vector $x = \mathbf{1}_{A \triangle B} \in \mathbb{F}_2^n$ from its syndrome $Hx \in \mathbb{F}_2^m$. While the information-theoretic counting bound requires $L(n,d) := \log \sum_{i=0}^d \binom{n}{i} = d\log(n/d) + \Theta(d)$ bits, existing polynomial-time schemes either use at least $d\lceil\log(n+1)\rceil$ bits (algebraic sketches, leaving an $\Omega(d\log d)$-bit gap) or incur a multiplicative $(1+\delta)L(n,d)$ overhead (peeling-based sketches), raising the question of whether $L(n,d) + O(d)$ bits is achievable with fast updates and decoding. We construct a public-coin randomized $\mathbb{F}_2$-linear sketch of length $L(n,d) + O(d) + \lambda\log^{O(1)} n$ that recovers each fixed input of weight at most $d$ with probability at least $1 - 2^{-\lambda}$, with $\log^{O(1)} n$ update time and $d\exp(O((\log\log n)^2)) = dn^{o(1)}$ worst-case decoding time. In particular, for $n \le 2^{\operatorname{polylog}(d)}$ and $\lambda \le d^{1-\Omega(1)}$, this achieves $L(n,d) + O(d)$ bits with $\operatorname{polylog}(d)$ update time and $d^{1+o(1)}$ decoding time. Our construction encodes $d$ permuted rows via randomly scaled Pinsketch syndromes and sketches the resulting bit array column by column. Decoding columns sequentially from prefix predictions cancels recovered rows and exposes undecided rows as decoder-side hints, reducing each column to a hinted sparse recovery problem that we solve via a recursive sketch, HintSketch.
Read the full abstract

All authors Tianren Liu, Rui Xu, Tong Yang, Kangyang Zhou

Probability

Rate of convergence for the Wiener–Itô chaos in $W_p$ for $1\leq p<\infty$

Inspired by previous work of Nualart, Ledoux, Peccati, Nourdin, and others, we establish quantitative fourth moment theorems for random variables in a fixed Wiener–Itô chaos and for sums of two Wiener–Itô chaoses in the $W_p$-Wasserstein distance for $1\leq p<\infty$.

Combinatorics

Snowflake $q$-Flipping

We prove a weighted extension of Byun's snowflake-flipping theorem for cored hexagons with unit triangular holes along three crossing lines. The ratio of the tiling generating functions of two regions related by a snowflake flip is given by an explicit product. The result recovers Byun's theorem when all lozenge weights are 1 and resolves an open problem posed by the author in 2024. Our proof uses shuffling identities for semihexagons with dents on two sides.
Read the full abstract

Mesoscale and Nanoscale Physics

Industrial RPCVD growth of high-purity 28Si epilayers with 29Si content below 30 ppm for high-fidelity electron spin qubits

Silicon quantum-dot spin qubits are among the most promising routes to large-scale quantum computation, but the residual 29Si nuclear spins in natural silicon impose a hyperfine dephasing limit that must be overcome by isotopic purification. Here, we report the epitaxial growth of high-purity 28Si on float-zone silicon using an industrial reduced-pressure chemical vapor deposition (RPCVD) system. Secondary-ion mass spectrometry (SIMS) confirms an 28Si isotopic purity of 99.997%, with the 29Si concentration below 30 ppm. Through a systematic vacuum annealing matrix (850-1150 C), combined with atomic force microscopy and SIMS, we investigate the defect-mediated decomposition of the native oxide and the injection of non-equilibrium vacancies. We identify a thermal-budget window below 950 C within which the thin-layer structure of 28Si remains intact. At 1.5 K, magnetotransport measurements reveal a peak Hall mobility of 12,858 cm^2 V^-1 s^-1 and a quantum scattering lifetime of 0.659 ps, corresponding to a single-particle level broadening of 499 micro-eV. A double quantum dot fabricated on this platform exhibits a T2* of 0.62 us, a Hahn-echo coherence time of 28.8 us, and single-qubit gate fidelities of up to 99.9% measured by randomized benchmarking. Our results establish industrial RPCVD as a promising route toward high-coherence, high-fidelity silicon spin qubits.
Read the full abstract

All authors Zhenzhen Kong, Weizhu Liao, Ning Chu, Yiwen Zhang, Ying Zhang, Jixuan Li, Changyang Xu, Yongliang Hao, Yi Zhou, Yanpeng Song, Xiaomeng Liu, Hailing Wang, Xiangsheng Wang, Jinbiao Liu, Chao Zhao, Hao Meng, Gang Cao, Guo-Ping Guo, Hai-Ou Li, Guilei Wang

Number Theory

Revisiting the conjectures of Shapiro and Schanuel

We tackle Shapiro's conjecture for exponential polynomials where the coefficients are polynomials, and show that it follows from Schanuel's conjecture. More concretely, given exponential polynomials $f$ and $g$ with infinitely many common zeros, we show that Schanuel's conjecture implies that there is an exponential polynomial $h$ vanishing precisely at all the common zeros of $f$ and $g$.
Read the full abstract

General Mathematics

A note on two functional inequalities

Motivated by some recent works on the following composite functional equation: $$f(x^2+yf(z)) = xf(x)+zf(y), \quad x, y, z \in \R,$$ we solve two related functional inequalities. In each case the solutions are only the constant zero function and the identity on $\R$.

Hardware Architecture

A Per-Step Load-Balancing Engine for Expert-Parallel MoE Serving

In expert-parallel Mixture-of-Experts serving, a layer finishes only when the most loaded GPU of its group does. Token routing shifts at every decode step and concentrates on a few experts, so most of the group's reserved GPU time waits on stragglers. Recovering that time means re-deciding, at every layer of every decode step, how each replicated expert's tokens divide between its copies. The exact linear-program solvers that make this decision optimally take longer than many of the layers they serve, and their working set outgrows GPU shared memory at deployment scale. We introduce Relief, a bidirectional pairwise relaxation that makes the linear program unnecessary, coming within a fraction of a point of the per-step optimum in a few passes of integer halving over the token counts. We build and measure three deployments, CUDA kernels on the serving GPU, the same verified engine on an FPGA over PCIe, and a fixed-function block beside the accelerator. On decode routing from three open MoE families, Relief holds the balance near the per-step optimum at every group size and layer time measured. Scored as effective utilization relative to that optimum, the kernel alone matches or exceeds every deployable scheme at every group size, from the hundred-microsecond layer on. The on-die engine widens that lead at every point, recovering up to 37.3 percentage points over the deployed static placement, while adding 0.02% of a current accelerator's transistors.
Read the full abstract

Classical Analysis and ODEs

The bilinear Hilbert transforms are not of restricted weak type $(1,1,1/2)$

For every real $\alpha\neq0,1$, we show that the bilinear Hilbert transform \[ H_\alpha(f_1,f_2)(x) =\text{p.v.}\int_{\mathbb{R}} f_1(x-\alpha y)\,f_2(x-y)\,\frac{\mathrm{d} y}{y} \] fails to map $L^1(\mathbb{R})\times L^1(\mathbb{R})$ to $L^{1/2,\infty}(\mathbb{R})$ when restricted to characteristic functions of sets of positive finite measure. The counterexample is quantitative and can be realized by sets that are finite unions of intervals of a common arbitrarily small length.
Read the full abstract

Differential Geometry

Global Classification of the Flat Singular Escobar-Yamabe Equation

In every dimension $n\geq3$, we classify positive solutions of the flat Escobar–Yamabe equation on the Euclidean half-space with a possible isolated boundary singularity at the origin. Solutions with a removable singularity are boundary bubbles. Up to dilation, the nonremovable solutions are the cylindrical solution and a family of periodic Delaunay solutions uniquely parametrized by the necksize $\epsilon \in (0,1)$, the minimum of the normalized boundary trace. In each fundamental period $T_\epsilon$, the trace has exactly two critical points, a nondegenerate minimum and a nondegenerate maximum. The period and the solution centered at the trace minimum and rescaled to period one depend smoothly on $\epsilon$. As $\epsilon \to 0$, $T_\epsilon = \frac{4}{n-2}\log(1/\epsilon)+O(1)$. As $\epsilon \to 1$, the solutions converge to the cylindrical solution and their periods tend to that of its linearized oscillation. We use uniqueness in the odd and even classes for linear Robin equations to obtain reflection symmetry and continuation in the necksize.
Read the full abstract

Cryptography and Security

RoCoMark: Robust Code Watermarking Against LLM-Based Functionality-Preserving Rewriting Attacks

Large Language Models (LLMs) present significant challenges to source tracing of generated code, as LLM rewriting can substantially alter the syntactic and structural characteristics of watermarked code. However, existing watermarking methods often suffer severe performance degradation under LLM-based functionality-preserving rewriting attacks. We attribute this vulnerability not merely to the intensity of rewriting, but to a fundamental distribution mismatch between the perturbations encountered during watermark training and the global, non-deterministic rewriting behaviors induced by LLMs. In order to deal with this challenge, we propose RoCoMark, a Robust Code waterMarking framework that explicitly learns and adapts to LLM rewriting behaviors. RoCoMark is a multi-round minimax game between the watermark embedder and LLM rewriter. In each round, the embedder first learns threat profiling from LLM-rewritten code by optimizing the watermark encoder through contrastive learning and knowledge distillation while mitigating catastrophic representation drift. Then the embedder improves strategies by optimizing the watermark encoder and extractor simultaneously. Finally, an adaptive attack-evolution mechanism based on Reflexion and short- and long-term memory pools is applied, enabling continuous interaction between the watermark embedder and evolving rewriting strategies. This iterative attacker-defender process progressively improves the robustness of the watermark against diverse and previously unseen LLM rewriting behaviors. Extensive experiments on four typical benchmarks, under attacks generated by five LLMs, show the effectiveness of RoCoMark. RoCoMark achieves an average detection AUROC of 74.2% and a Rewrite BitACC of 75.3%, while maintaining a code functionality pass rate above 99.7%.
Read the full abstract

All authors Hongjie Zhang, Xingcheng Li, Hanzhou Wu, Zhihua Xia, Xinpeng Zhang

Machine Learning

Decomposing Earth Embeddings with Sparse Autoencoders

Earth observation foundation models produce global metre-resolution embeddings that underpin work across agriculture, ecology, hydrology, and beyond. Yet these embeddings are opaque: each dimension entangles unrelated physical concepts, which makes them difficult to interpret and their robustness hard to assess. Here, we propose sparse autoencoders (SAEs) as a principled way to decompose Earth embeddings into sparse, approximately monosemantic features over an overcomplete dictionary. Because LLM judges of interpretations do not transfer to satellite pixels, and spatial autocorrelation matters, we introduce a tailored evaluation protocol. It scores any SAE on reconstruction at matched sparsity, alignment with reference concepts, and spatial scale, each against baselines that include an SAE whose decoder is a frozen random dictionary. On this protocol, SAEs reconstruct AlphaEarth and Tessera embeddings better than every baseline at matched sparsity. Their features align with far fewer reference concepts than raw embedding dimensions (on AlphaEarth, 88% align with at most one of 20 concept groups, most with none, against a median of 12 per raw dimension). They also carry more of their variance at fine spatial scales (23 to 32% within a 640 m patch for BatchTopK and Matryoshka, against 9%). An SAE with a frozen random decoder shares both properties, which therefore follow from sparsity, whereas only a learned dictionary reconstructs the embedding accurately: on AlphaEarth, trained SAEs leave eight times less variance unexplained. We test the sparse code on two applications, discovering land-cover sub-concepts and detecting out-of-distribution pixels by area-of-applicability estimation. Both indicate how sparse features could be used, but neither is yet reliable enough for practical use. Together, these results are a step toward Earth observation models whose features can be inspected.
Read the full abstract

Rings and Algebras

Inverse problem in difference Galois theory

Let $C$ be an algebraically closed field of characteristic zero, and let $K=C(x)$ with $\sigma(x)=x+1$. We show that every closed subgroup $G\subseteq\GL_n(C)$ with cyclic component group $G/G^\circ$ can be realized as the \PV{} group of a system $\sigma Y=A(x)Y$. Here $G^\circ$ denotes the identity component of $G$. Moreover, we can choose the coefficient matrix to have the formal expansion $A(x)=h+O(x^{-2})$ at infinity, where $h\in G(C)$ is semisimple. This gives an affirmative answer to the conjecture of van der Put and Singer in their monograph Galois Theory of Difference Equations (Lecture Notes in Mathematics, vol.~1666). The original results in this paper were entirely generated by ChatGPT 6 and the MechMath Agent Team (MMAT) developed by the Key Laboratory of Mathematics Mechanization, AMSS, CAS. The author reviewed and simplified the initial proofs and verified the correctness of the results.
Read the full abstract

Computation and Language

Search-by-Answer Bias in Retrieval Models

Dense retrievers are meant to select passages that match a query's semantic specification. We show they suffer from search-by-answer bias: they favor passages that mention answer objects the model knows (scores as likely), over passages that support valid alternative answers. We test this with controlled factual contrasts from LAMA/TREx. For each retriever, we measure how strongly the model scores the canonical object over an alternative object using a token-prediction head, and ask whether this score predicts passage rankings. Across eight retrievers, including DPR, Spider variants, Contriever, E5, and Qwen3-Embedding, larger token-prediction scores consistently predict higher retrieval scores for passages containing the corresponding object. This holds when passages differ only in the answer object, and even when the passage that mentions the high-scoring object expresses a wrong relation. Dense retrieval thus exhibits search-by-answer bias during passage selection, shaping which evidence is retrieved.
Read the full abstract

Quantum Physics

High-Dimensional Maximally Entangled OAM States by Joint Pump and Modal-Space Engineering

Spontaneous parametric down-conversion (SPDC) generates correlated photon pairs that inherently exhibit high-dimensional spatial entanglement in orbital angular momentum (OAM) modes. In practice, the resulting two-photon states are typically not maximally entangled due to a highly non-uniform distribution of generated OAM modes. Although pump-beam engineering has emerged as a tool to prepare high-dimensional entangled states, it has usually been restricted to low-dimensional subspaces, i.e., small sets of selected OAM modes for each photon within the full infinite-dimensional two-photon Hilbert space. A general understanding of the requirements for the generation of maximally entangled states (MESs) for any dimensionality $d$ with $d>2$ is lacking. Here, we demonstrate how pump engineering enables the generation of high-dimensional MESs for arbitrary $d$ defined in $d \times d$ subspaces. By identifying the interplay between the spatial pump structure and the choice of OAM subspace, the required pump profile for a target MES is directly determined. Our results reveal that maximal entanglement with OAM requires a careful and nontrivial design of the effective high-dimensional Hilbert subspace.
Read the full abstract

All authors Richard Bernecker, Baghdasar Baghdasaryan, Fabian Steinlechner, Stephan Fritzsche

Cryptography and Security

Self-Healing by design: Immune-system inspired autonomous cyber resilience

Digital innovations have strongly improved ship capabilities, including command, control & communication, navigation and combat management. In the wake of digitisation and geopolitical stress, resilience against cyber-attacks has become a key challenge. Autonomous Cyber Resilience (ACR) methods, inspired by features of the human immune system, provide a promising perspective. Proofs of concept of autonomous self-healing and self-protecting software have been realised in cloud environments using containerisation, monitoring and deployment technology. However, developing a suitable "cyber immune system" for more complex digital systems is still an intricate challenge. In this paper, we share insights on how ACR methods can be combined with resilience mechanisms present in critical naval and aeronautical manoeuvring systems. Using TNO's "OpPic" application, we show how TNO's Self-Healing for Cyber Security software can provide autonomous mission- and threat-adaptive resilience. On this basis, six guidelines for the design of autonomous cyber resilient systems are derived. The ultimate goal is to enable operators to oversee and control their system's "cyber immune response", to be able to continuously adapt to evolving threats and mission priorities.
Read the full abstract

Quantum Physics

Information–control limits for record-indistinguishable reset of a monitored qubit

A qubit that is continuously monitored from the moment it is prepared in its excited state writes evidence of that preparation into its measurement record. We ask how much control is needed to reset such a qubit to its ground state while keeping its complete record statistically close to the record of an undriven ground-state qubit. The qubit is measured in $\sigma_z$ at rate $\Gamma_m$ with efficiency $\eta$ and has no relaxation. For feedback that acts causally on the record, distinguishability is the Kullback–Leibler divergence $D$ of the controlled record law from the reference law, and the control cost is the quadratic effort $J$, the integrated squared Rabi frequency. We prove that every open-loop drive leaves $D$ infinite and that every admissible control obeys $DJ\ge\Gamma_mU(\eta,D/\kappa_0)$, where $U$ is given by a one-parameter family of integrals and $\kappa_0=1-\ln2$. Our main result is the exact high-effort constant $\lim_{\epsilon\downarrow0}\inf\{DJ/\Gamma_m: D\le\epsilon\}=K(\eta)$, where $K(\eta)=[\int_0^\pi\sqrt{2\eta(1+\cos\theta)^2+(1-\eta)\sin^2\theta}\,d\theta]^2$; an explicit sequence of two-phase feedback policies attains it. The first term of the integrand is the population evidence that the record receives while the qubit is rotated toward the ground state. The second is the price of the impurity created when the unrecorded part of the measurement dephases the coherence that the rotation needs; the record receives this evidence later, when the measurement repurifies the qubit. The global constant $c(\eta)=\inf DJ/\Gamma_m$ satisfies $K/2\le F_2\le c\le K$, where $F_2$ is again given by an integral; $c$ is known exactly at $\eta=1$ and in the limit $\eta\to0$. With any nonzero relaxation rate the unconstrained infimum of $DJ$ vanishes, so $c(\eta)$ is specific to the relaxation-free model.
Read the full abstract

Systems and Control

A Distributed Observer with Random Agent Activations over Random Communication Networks

This paper develops a distributed observer for the forward-Euler discretization of a continuous-time linear time-invariant system under random local operations and random communications. Under uniform stochastic strong connectivity, we establish almost sure exponential convergence of all local state estimates for sufficiently small discretization step sizes. We further examine how the underlying parameters and components affect the admissible range of the discretization step size and the guaranteed convergence rate bound. Unlike related approaches, the proposed method does not require each local unobservable subspace to admit an $A$-invariant complement.
Read the full abstract

Computer Vision

Let the Gallery Learn Itself: Training-Free Iterative Side-View Face Recognition

What if a face recognition system could learn from every encounter without a single gradient update? We present Gallery Self-Evolution (GSE), a training-free framework where the gallery itself adapts at test time. The core mechanism—dedup-FIFO curation—archives novel appearance samples after successful recognition while permanently locking the enrollment template, bounded at 30 embeddings per identity. On CFP-FP and IJB-C benchmarks, GSE closes the side-view gap by +23 points at 45{\deg} yaw and boosts TAR@FAR=0.001 by +9.4 points, all at CPU-level cost. No retraining. No GPU. Just a gallery that grows smarter every time it meets you.
Read the full abstract

Analysis of PDEs

Quantitative Homogenization and Global Lipschitz Estimates for Elasticity with Periodic Rigid Inclusions

We consider the pure traction problem for linear elasticity in a bounded domain containing an $\varepsilon$-periodic array of perfectly rigid inclusions. For compatible $L^2$ boundary tractions on a Lipschitz domain, we prove an $O(\varepsilon^{1/2})$ error estimate in $H^1$ for a smoothed first-order two-scale expansion. If the exterior and inclusion boundaries are $C^{2,\beta}$, we also obtain a global Lipschitz estimate, uniform in $\varepsilon$, for compatible $C^{0,\sigma}$ tractions, where $0<\sigma<\beta$. The displacement is normalized modulo rigid motions, and the estimate controls the full gradient in the matrix and the inclusions. The boundary argument uses a local Neumann comparison estimate whose constant is independent of the choice of inclusions retained in cells cut by an artificial localization boundary. We prove this estimate by combining cellwise rigidification, divergence-free stress extensions, and duality, and then apply a boundary Campanato iteration and local estimates at the microscopic scale.
Read the full abstract

Number Theory

Smooth approximations to character sums and zeta sums

We study smooth approximations to Dirichlet character sums and zeta sums. The uniform upper-bound consequence of the Granville–Soundararajan conjecture, when imposed on all nonprincipal characters including imprimitive ones, implies the Riemann hypothesis for every primitive nonprincipal Dirichlet \(L\)-function. The proof uses characters induced to moduli divisible by all small primes. In the other direction, assuming the Riemann hypothesis for \(L(s,\chi)\), we establish the conjectured approximation uniformly for \(q^a\le x\le q\), where \(0<a\le1\) is fixed, with \(y=(\log q+(\log x)^2)(\log\log q)^C\) for every fixed \(C>6\). In this range both the character sum and its smooth restriction, and hence their difference, are exponentially small relative to the number of \(y\)-smooth integers coprime to \(q\). We establish the corresponding approximation for \(\sum_{n\le x}n^{-it}\) under the Riemann hypothesis for the zeta function, uniformly for \(T^a\le x\le T\) and \(2T\le t\le5T\). We also show that the full zeta-sum smooth-approximation conjecture implies the Lindelöf hypothesis.
Read the full abstract

Statistics Theory

Uniform inference for $U$-statistics of exchangeable arrays under degeneracy

Many $U$-statistics of row-column exchangeable arrays become degenerate at natural null hypotheses: their convergence rates change and their limiting laws may or may not remain Gaussian. We establish conditions under which inference based on the classical unbiased variance estimator is uniformly valid across these changes, without knowledge of the degeneracy order. It is known that $U$-statistics can be decomposed into orthogonal projections indexed by bipartite graphs. Under uniform fourth-moment assumptions, we establish that the standardized statistic is asymptotically Gaussian and the classical unbiased variance estimator is ratio-consistent, as long as a certain measure of concentration for projections indexed by disconnected graphs tends to zero. Under compactness of the normalized projections, both properties hold if and only if the proportion of variance contributed by disconnected graphs tends to zero. The necessity of this condition follows from a general limit theorem for these $U$-statistics, in which each graph contributes to a Wiener chaos of order equal to its number of connected components. We note that the row-column jackknife variance estimator, although nonnegative, does not generally give a studentized statistic with a standard normal limit in degenerate regimes. We illustrate our theory with an application to estimation and testing based on a distance from product form, exhibiting Gaussian and non-Gaussian limits at various rates depending on the model. We also propose a conservative inference procedure that does not require a Gaussian limit, valid under additional moment and projection conditions.
Read the full abstract

Machine Learning

Orthogonal Updates for the Win: Towards Accelerated Adaptive Minimax Optimization

Single-loop minimax methods are appealing for modern machine learning, but standard stochastic gradient updates can be unstable and oscillatory under tightly coupled primal-dual dynamics, making performance notoriously sensitive to stepsizes. To fundamentally improve stability, we develop an orthogonal update framework tailored to minimax optimization, which applies updates along orthogonalized directions to reduce update anisotropy. However, orthogonalization reshapes the optimization geometry and makes coordinating primal-dual stepsizes even more delicate. To overcome this challenge, we develop AdaSGDA, which equips orthogonal updates with an adaptive stepsize rule driven by accumulated gradient norms. This mechanism automatically balances coupled primal-dual progress under orthogonalization, yielding an effectively parameter-free method for nonconvex-strongly-concave minimax optimization. We prove that AdaSGDA avoids problem-dependent tuning while attaining the state-of-the-art $\mathcal{O}\left(T^{-1/4}\right)$ convergence rate. We further propose AdaSGDA-VR, which incorporates a carefully designed variance reduction scheme and achieves the faster $\mathcal{O}\left(T^{-1/3}\right)$ rate. Extensive experiments demonstrate that our methods deliver competitive performance compared to existing minimax optimization methods.
Read the full abstract

All authors Zhiwei Zhai, Xinyu Wang, Wenjing Yan, Lei Ding, Ying-Jun Angela Zhang

Computation and Language

What Travels Across Languages? Reading, Transfer and Erasure of Cross-Lingual Concept Subspaces in Bangla Idioms

Cross-lingual concept analysis requires distinguishing the information that a subspace retains from the classifiers and erasers that transfer through it. We study these operations in six frozen encoders using 10,359 Bangla (Bengali) idiom senses, eight input conditions, and figurative-class and sentiment labels. Class-mean directions provide the most consistent evidence of shared coarse meaning: in post hoc analyses, they outperform random-subspace, permuted-label and covariance-matched nulls in all 12 model-label tests under balanced accuracy. A prospective internal re-split supports idiom-alone sharing in 10 of 12 tests, spanning five encoders. The original confirmatory logistic-regression family rejects both planned nulls in 4 of 12 tests, where English directions gain only a few accuracy points above the majority-class floor (3.0-7.3). Classifier transfer depends on the input map and regularization: English-trained class-mean readouts beat both zero-shot nulls in 8 of 12 discovery tests, and stronger source-selected penalties improve transfer through logistic and LEACE directions. These readouts use Bangla-validation-selected layers and unlabeled target preprocessing. Input controls show greater coarse-class sharing for explicit glosses than for idiom counterparts. Reading, however, does not imply effective erasure: English-fitted LEACE removes less than one percentage point of Bangla figurative-class accuracy, compared with 4-11 points for Bangla-fitted LEACE; INLP shows a similar asymmetry. Pattern-filter analysis, covariance-weighted input maps and a binary Gaussian overlap bound explain how these operations can diverge. The analogous multiclass ordering is an empirical diagnostic, observed in 11 of 12 cells. The results support an estimator-sensitive account of cross-lingual concept sharing and motivate evaluating reading, classifier transfer, erasure and overlap separately.
Read the full abstract

Complex Variables

Hausdorff Dimension of Escaping Sets of Transcendental Entire Functions in the Speiser Class

For every $d\in[1,2]$, we construct a transcendental entire function $f$ whose finite singular value is contained in $\{-1,0,1\}$ and whose escaping set has Hausdorff dimension $d$. For $d<2$, the construction combines spiral corridors with quasiconformal folding. A narrowing Archimedean spiral gives dimension one, while a logarithmic spiral with adjustable width gives intermediate dimensions. The endpoint $d=2$ is realized by the exponential map. This gives an affirmative answer to a question of Bishop posed in 2018.
Read the full abstract

Optics

Collective Chiral Response of Achiral Plasmonic Arrays Induced by Truncated Bloch-Floquet Effect

Chirality in plasmonic systems is commonly associated with structures differentiated by their handedness. Here, we demonstrate the emergence of collective chirality in structurally achiral periodic hole arrays induced by non-local surface-plasmon interactions. We attribute this effect to the transfer of the optical angular momentum in the system. While an individual sub-wavelength circular aperture is not expected to alter the total angular momentum, the periodic structure results in a coherent coupling of plasmons wavefronts, giving rise to a collective Bloch-Floquet-like response. The edge of the structure accumulates the helical phase that eventually results in a visible spin-dependent shifting of the plasmonic beam. We systematically study the effects of hole-array size, periodicity and the geometry. The observed phenomenon shows an unprecedented robustness and consistency. These results establish periodic plasmonic arrays as a platform for converting localized helicity-dependent responses into collective chirality, providing a route toward controlling chiral plasmonic wavefronts in structurally achiral systems.
Read the full abstract

All authors Sahil Sahoo, Ahmed Lafeef Ettapuram Naduvilepurayil, Akash Das, Yuri Gorodetski

Analysis of PDEs

Nonlinear Stability of Planar Shock Waves for the Navier-Stokes-Fourier Equations in R^3

In this paper, we investigate the global nonlinear stability and large-time decay rates of small-amplitude planar viscous shock waves for the three-dimensional (3-D) full compressible Navier–Stokes–Fourier equations in the whole space $\R^3$. {We prove that if the initial perturbation is sufficiently small in $H^3(\R^3)\cap\Hm(\R^3)$, then the Cauchy problem admits a unique global strong solution that converges to the planar shock profile without any phase shift. In the moving shock frame, both the conservative and primitive perturbations decay at the rate $(1+t)^{-1/4}$ in $H^3(\R^3)$. Their time and transverse spatial derivatives decay at the rate $(1+t)^{-1/2}$, measured in $\Hm(\R^3)$ for the conservative perturbation and in $H^1(\R^3)$ for the primitive perturbation.} No zero-mass condition or spectral stability assumption is imposed. On the whole space $\R^3$, the main analytical difficulties stem from the continuous transverse low frequencies, the opposite signs of shock compression in $L^2(\R^3)$ and unweighted $\Hm(\R^3)$, and the absence of density dissipation in the continuity equation coupled with temperature variations. To overcome these difficulties, we introduce a density-weighted $\Hm(\R^3)$ multiplier built from the inverse entropy Hessian and a modified curl operator. Through entropy symmetrization and the curl projection, all second-order acoustic derivatives of the potential cancel out and the transverse acoustic modes are controlled, so that the normal Euler drift is governed by the compressive derivative of the characteristic speed. Combined with an effective momentum variable and a momentum–density cross estimate, this negative-order energy absorbs the unfavorable compression term in the $L^2$ energy and closes the $H^3(\R^3)$ a priori estimates without loss of derivatives.
Read the full abstract

Machine Learning

The comparison of density estimation for transfer learning in Banach, Hilbert, Euclidean spaces

Domain adaptation or DA methods are indispensable when target labels or computational resources are scarce, yet the domain kernel density and subspace alignment approaches completely assume a Hilbertian geometry. They measure distributional discrepancy with symmetric L2 divergences over Euclidean feature spaces or reproducing kernel Hilbert spaces (RKHS). This preprint summarizes the manuscript Non-Hilbertian Domain Adaptation. Total Variation Alignment via L1 Density Regularization on Feature Manifolds, which proposes Banach Manifold Transfer Learning or BMTL. BMTL makes two coupled design decisions. The first one is that features are embedded on a geodesic manifold rather than treated as points of a flat Euclidean space, and the second one is that the density matching regularizer is formulated in an L1 Banach norm, whose minimization is equivalent, by Scheffe's theorem, to contracting the Total Variation or TV distance between domain densities, a bounded metric that weights all regions of the sample space uniformly. The L2 formulation, by contrast, squares deviations and therefore amplifies errors in the tails, over emphasizing density outliers that do not reflect meaningful distributional shift. Experiments on MNIST-USPS, COIL20, and Office+Caltech10 show consistent gains; the L1 variant beats an otherwise identical L2 ablation on every task, for example 92.44% vs. 89.80% on USPS->MNIST and surpasses the strongest baseline DMDA by +3.68% on average over twelve Office+Caltech transfers. This document situates the contribution with an explicit comparison of Euclidean, Hilbert, and Banach geometries.
Read the full abstract

Statistics Applications

Does Predictive Performance Translate to Causal Performance? Evaluating tabular foundation models as nuisance learners for causal effect estimation

Tabular foundation models based on prior-data fitted networks (PFNs) achieve strong predictive performance, but their value for causal inference is unclear. We evaluated whether PFNs, which predict through in-context learning, could match or outperform conventional nuisance learners in supporting accurate causal estimation and reliable uncertainty quantification (inference). Learners modeled propensity scores and outcome regressions within doubly robust estimators of average causal effects, with and without cross-fitting, across synthetic and semi-synthetic settings with varying confounding complexity and effect heterogeneity. Under simple parametric data-generating processes, the PFNs were competitive with correctly specified parametric regressions and ensembles containing them. In more complex settings, they outperformed all conventional learners evaluated, including richly specified Super Learner ensembles, for both accuracy and uncertainty quantification. As sample size increased, only the two PFN models (TabPFN and Mitra) consistently approached nominal confidence interval coverage, whereas under-coverage persisted for conventional learners. In some settings, PFNs achieved near-nominal coverage and well-calibrated standard errors even without cross-fitting. At the smallest sample size (n=200) and in the higher-dimensional semi-synthetic settings, PFNs and other flexible learners exhibited poor coverage without cross-fitting. For PFNs, this appeared to be driven by propensity score overfitting consistent with in-sample memorization of treatment labels. These findings highlight the promise of prior-data fitted networks as in-context nuisance learners for causal estimation, while underscoring the need for further evaluation in challenging settings.
Read the full abstract

All authors Andrew R Weckstein, Issa J Dahabreh, Marinka Zitnik, Sonia Hernández-Díaz, Sebastian Schneeweiss

General Relativity and Quantum Cosmology

Negative Dynamical Friction in Dark Matter Spikes around Rotating Black Holes

Dark matter may affect extreme-mass-ratio inspirals (EMRIs) through orbital-energy exchange and a conservative metric response. Here we show that dark matter spikes around rotating massive black holes can drive negative dynamical friction, supplying orbital energy to secondary black holes on stable prograde circular orbits. The energy-gain region extends beyond the innermost stable circular orbit once the black-hole spin exceeds a model-dependent threshold. For an illustrative four-year inspiral with Milky Way–like halo parameters and mass ratio $q=10^{-5}$, the total dynamical-friction contribution produces a phase difference of order $10^2$ radians relative to vacuum. For the same system, the fixed-frequency environmental redshift correction is comparable in magnitude and opposite in sign to the first-order conservative self-force correction. Both environmental energy exchange and the conservative metric response are therefore essential to precision EMRI modeling.
Read the full abstract