Complex Variables
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Sunday 13 September
Let $X$ be a compact connected Riemann surface equipped with two nonconstant meromorphic projections $q,z:X\to\mathbb{P}^1$. Their differentials determine a canonical meromorphic relative differential morphism $Z:T_q=q^*T\mathbb{P}^1\to T_z=z^*T\mathbb{P}^1$, whose local coefficient in coordinate frames is $Z=z_q\,\partial_z\otimes dq$. Its divisor is $\operatorname{Div}(Z)=R_z-R_q$, and its local logarithmic derivative defines a rank-one logarithmic connection on $L=\operatorname{Hom}(T_q,T_z)$. The general theory of logarithmic connections is classical. The main point of the present paper is that, in the dual conformal-net setting, the Levi–Civita curvature data of the two nets realize this logarithmic connection by the symmetric formula $H^{-1}\widetilde{\omega}\,dz-H\omega\,dq =2i\,d\log z_q =2i\,Z^{-1}dZ$, where the last expression is understood through local trivializations of $L$. Thus the two dual conformal connection differentials occur with opposite signs, and their difference is the logarithmic Maurer–Cartan form of the relative differential morphism. Its residues recover the relative ramification multiplicities, $\operatorname{Res}_p\Theta=2i(r_z(p)-r_q(p))$, and globally $\sum_{p\in X}\operatorname{Res}_p\Theta =4i(\operatorname{deg}z-\operatorname{deg}q)$. Thus local conformal connection data determine both relative ramification and the difference of the degrees of the two projections. We also discuss the dual morphism, the square-root issue, and a hyperelliptic example.
Complex Variables
Generators of stability-preserving semigroups, spectral gaps, and classical ground-state computation
We classify the generators of stability-preserving semigroups on polynomial spaces with bounded coordinate degrees. The generators have differential order at most two, with principal coefficients characterized by low-degree nonnegativity conditions. In disk coordinates, positive degree damping gives strict zero-freeness and a sharp spectral gap, extending the Hermitian gap of Bravyi, Gosset, Liu, and Wong to complex generators. The resulting analytic domain yields deterministic classical algorithms for ground energies and stable product-state queries for bounded-degree Suzuki-Fisher Hamiltonians with bounded local strength and fixed positive fields. Field perturbation gives zero-field energy-value approximation schemes for bounded-degree unweighted EPR and bipartite Quantum MaxCut. We also prove that, up to scalars, the Hermitian qubit Hamiltonians whose full Gibbs tensors are Lee-Yang at every temperature in a fixed basis are precisely the edgewise phase-rotated Suzuki-Fisher family. With strictly positive longitudinal fields, these tensors are zero-free on a polydisk of radius greater than one at each positive inverse temperature. These two statements answer questions of Wong, Bravyi, Gosset, and Liu. Degree-preserving, coefficient-positive stability semigroups have concave sector growth rates, yielding token-graph concavity. We also give a counterexample to a proposed explicit ground-state radius.
Read the full abstract
Instrumentation and Detectors
Cross-Detector Transfer Learning with Parnassus: From ALEPH to SLD
Parnassus is a fast detector-simulation and reconstruction framework that maps truth-level particles directly to reconstructed particles. In this work, we investigate cross-detector transfer learning by adapting a Parnassus model for the ALEPH detector to the SLD detector. ALEPH and SLD have similar detector responses, as they were both targeting hadronic $Z$-pole $e^+e^-$ events, while differing substantially in the underlying detector technology, reconstruction, and archived data representation. We initialize the SLD particle model with weights learned on ALEPH, fine-tune it on SLD, and compare it with the same architecture trained directly on SLD. The ALEPH-initialized model gives substantially improved particle- and jet-level agreement with the SLD reference, including a factor of 5.2 improvement in the charged-particle angular response and an improvement in the jet angular resolution from $1.73$ to $1.15$ times the SLD reference. The results show that detector-response information learned on ALEPH can be reused when modeling SLD and motivate reusable pretrained Parnassus models for legacy $e^+e^-$ detectors, which is especially critical for experiments without access to the original software pipeline. With this paper, we also release an AI-ready version of simulated SLD $e^+e^-$ events.
Read the full abstract
Classical Analysis and ODEs
A reciprocal-Gamma Mehler–Heine limit for Krawtchouk–Sobolev type orthogonal polynomials with an exterior mass point
We study the monic polynomials $\Sob{n}$ orthogonal with respect to a Sobolev-type modification of the discrete Krawtchouk inner product in which a single point mass at an exterior point $\alpha<0$, outside the support $\{0,1,\dots,N\}$ of the binomial measure, acts through the $j$-th forward difference $\dif^{j}$ (here $\Nzero=\{0,1,2,\dots\}$, $\Z_{+}=\{1,2,\dots\}$, $\R_{+}=(0,\infty)$). In the joint scaling $n,N\to\infty$, $n/N\to r$ with $0<p<r<1$ (the reciprocal-Gamma regime of Dominici's classification), we prove the Mehler–Heine formula. The exterior mass survives the large-degree limit exactly through the linear factor $z-\alpha$, with the explicit universal constant $C_{\ast}$, and the limit is independent of both $\lambda>0$ and $j\ge1$; the same conclusion holds in the Uvarov case $j=0$ (Corollary~\ref{cor:uvarov}). The derivation is pointwise and self-contained: it uses only a rank-one connection formula (Theorem~\ref{thm:connection}), the reproducing-kernel recursion, Dominici's pointwise limit and the classical three-term recurrence; the normalised Sobolev correction obeys an exact affine recursion at fixed $N$, controlled by a triangular contraction argument over two backward windows with an explicit boundary estimate, and the contraction ratio $\varrho<1$ is exactly the condition $r>p$. We also discuss the scale of the theorem: in this normalisation the classical term is asymptotically negligible, and the independence of $\lambda$ concerns each fixed $\lambda>0$, not the continuity at $\lambda=0$. Multiprecision computations illustrate the convergence and the exact value $C_{\ast}=2.8$ for $p=0.3$, $r=3/5$, $\alpha=-2.5$, $j=2$, $\lambda=5$.
Read the full abstract
General Mathematics
1729 in the Mirror: Reversal and Affine Multiplication
The decimal reversals 1729 and 9271 have multiplicative orders of $2$ equal to 36 and 63. Motivated by this symmetry, we classify the affine commutation equation $R_B(hx+1)=hR_B(x)+1$, where $R_B$ reverses base-$B$ digits, for every $B\ge3$, $2\le h<B$ and positive integer $x$ such that neither $x$ nor $hx+1$ ends in zero. Comparing the two carry sequences separates the solutions by their digit lengths. Equal lengths force an odd number of digits alternating between two explicit intervals. When the output gains a digit, coprimality of $h$ and $B+1$ forces a repdigit input with palindromic output. In the noncoprime case, all inputs instead have even length and form a finite family. A slack-variable bijection gives the sharp maximum length, binomial counts at each length, and totals expressed through Fibonacci numbers or powers of two. For reversal pairs of primes already known to share an order index $2\le h<B$, the classification gives an exact digit criterion for reversal of their orders. Finally, an exhaustive computation with independently checkable primality certificates proves that $\{1729,9271\}$ is the unique pair of distinct odd decimal reversals whose orders of $2$ are distinct two-digit reversals of each other.
Read the full abstract
Robotics
OJOx: Specification-Conditioned Demonstrations for Embodied AI in Construction
Large-scale egocentric and whole-body human demonstrations are becoming a primary source of data for embodied intelligence. They record what people perceive and do, but rarely the external specification that gave an action its purpose. In construction that omission is consequential: skilled work is directed at project-specific configurations defined in a design model - configurations not yet present in the environment being observed. A mason's transferable competence is not the geometry of one wall but the ability to realise a new geometry from a specification. We introduce the specification-conditioned demonstration: a synchronised record of the physical state a demonstrator perceives, the intended state supplied to them by an external design, and the behaviour connecting the two. We present OJOx, a capture interface that realises this for construction - delivering design geometry to a headset, anchoring it in the physical workspace, rendering it into a demonstrator's stereo passthrough view, and recording that view synchronously with whole-body and hand motion. We report one fully instrumented session - a 33-component wall laid against a specification that changes while the work proceeds - and check the record against the physical scene through an external camera registered independently of the capture. Recorded sessions remain compatible with existing humanoid retargeting infrastructure and replay onto a Unitree G1 in simulation. The result is a data interface for testing whether embodied policies can learn not merely to imitate demonstrated actions, but to act toward specifications absent from their training experience.
Read the full abstract
Signal Processing
HAPS-Enabled 6G TN-NTNs: Achieving Superior QoS-Energy Tradeoffs Compared with LEO
Non-terrestrial networks (NTNs) can reduce terrestrial energy consumption by absorbing traffic from lightly loaded base stations (BSs), allowing them to be switched off. We study how the NTN platform shapes this quality of service (QoS) energy trade-off, comparing one low Earth orbit (LEO) satellite at 600 km with 1, 4, and 9 high-altitude platform stations (HAPS) at 20 km. Using a traffic-aware framework that jointly optimizes user association, NTN bandwidth, and terrestrial BS activation over a 24-hour traffic profile, we show that 9 HAPS reduce terrestrial energy consumption by 34.1%, versus 26.0% for LEO. At the same QoS, 9 HAPS save 20% more terrestrial energy than LEO while improving cell-edge throughput.
Read the full abstract
Materials Science
Predicting energy and structural response to force correction in molecular dynamics
We predict how force correction changes energy exchange and structural statistics by measuring leading response coefficients on shared reference trajectories. Residual power and the displacement virial distinguish the transfer of energy from the change in restoring forces, including intermittent reference updates. Independent simulations then test the predicted kinetic and configurational shifts. A matched-timetable experiment shows that reference timing affects heating through its coupling to the evolving state. In an anharmonic chain, the displacement virial predicts a structural shift missed by a power-only description. Local force constants in silicon predict a complementary directional tradeoff: scalar calibration repairs optical motion while degrading an already accurate low-frequency direction. Full nonlinear trajectories confirm this tradeoff and distinguish the benefits of static curvature correction and repeated reference impulses. Independent finite-temperature integrals in orthorhombic tin selenide (SnSe) support the configurational-response direction predicted from separate reference calculations. These results provide a physical basis for choosing how reference information enters molecular dynamics. The framework assesses force correction through its effect on atomic motion and statistical observables, beyond the accuracy of individual force evaluations.
Read the full abstract
Number Theory
Superpolynomially Large Support in Every Irreducible Factor of Lacunary Polynomials
We show that there exist infinitely many polynomials $F\in\mathbb{Q}[x]$ with exactly $m$ nonzero coefficients such that every irreducible factor of \(F\) over \(\mathbb{Q}\) has \(\exp\) \(\!\bigl(\Omega(\sqrt{m/\log m})\bigr)\) nonzero coefficients. This places multiplication in a sharply different category from powers, composition, and other algebraic operations for which reverse-sparsity principles are known. It also gives an unconditional, degree-free obstruction to sparse factorization, complementing positive factor-sparsity results that impose hypotheses on coefficient height, exponent positions, reciprocal structure, or degree bounds.
Read the full abstract
Robotics
ORDER: A Fictitious-World Benchmark for Domain-Adaptive Embodied AI
Adapting language models to new domains via continual pre-training raises a basic evaluation problem: if the training corpus overlaps with what the model already knows, performance gains cannot be cleanly attributed to new learning rather than pre-existing knowledge. This matters most for knowledge-intensive, task-light (KHTL) robot deployments - pharmaceutical dispensing, hazardous-material handling, facility-specific protocols, where the physical task is simple but the governing rules are proprietary and safety-critical, and where extensive live testing is costly or unsafe. We introduce ORDER (Ontology-driven Decision-making for Embodied Reasoning), a benchmark built on a fictitious world: a 342,069-token synthetic corpus defining a self-consistent physics that cannot appear in any model's pre-training data. ORDER pairs a 500-question knowledge test (ORDER-BENCH) with a harder compositional task, ORDER-SPATIAL: ordering objects for safe manipulation across both familiar and entirely novel scenes. GPT-4.1 without adaptation scores below chance on ORDER-SPATIAL (Kendall's tau = 0.441), showing its priors actively conflict with the invented physics. After continual pre-training, small models improve substantially on both familiar and novel scenes alike evidence of genuine world-model induction rather than memorization. We then carry this through to a robot pipeline: models that answer the knowledge test well often cannot produce valid, executable plans without a further skill-adaptation stage, after which small, fully offline models outperform GPT-4.1 even when GPT-4.1 is given retrieval access to the same rules (Kendall's tau = 0.848 vs. 0.606), on a full perception-to-execution loop demonstrated on a simulated iiwa7 arm with human-in-the-loop correction. Throughout, ORDER-SPATIAL performance, not knowledge-test accuracy is what predicts real plan quality.
Read the full abstract
Computer Vision
Complementary rPPG-Derived and Lip-Region Frequency Cues for Talking-Face Deepfake Detection
Talking-face (TF) deepfakes are detected unevenly by rPPG-based methods across generators. We study two lightweight visual-only cues, rPPG-derived waveforms extracted by RhythmFormer and lip-region discrete cosine transform (DCT) coefficients, on the seven TF methods of Celeb-DF++ under a subject-independent protocol. In-domain, lip-region DCT matches or exceeds the rPPG-derived 1D ResNet on every method except SadTalker, and Concat fusion reaches AUC 0.891 against 0.824 and 0.827 for the unimodal baselines. Under leave-one-generator-out evaluation the cues split: each transfers clearly better to three held-out methods, and IP-LAP is near chance for both. Concat averages 0.798 but falls below rPPG alone where DCT transfers poorly, so static fusion only partly exploits this complementarity. Lip-region DCT outperforms full-face DCT on six of seven methods. We treat the rPPG-derived signal as an empirical cue and do not claim it is cardiac in origin.
Read the full abstract
Computer Vision
Rethinking Streaming Video Diffusion Model: Context, Execution, and Training
Understanding the design space of streaming video diffusion is essential to exploring its potential for generation quality and computational efficiency. We develop a unified analytical framework that relates model and sampler choices, historical conditioning, execution scheduling, and training strategies. The framework accommodates a broad family of causal context-selection policies and makes their computational dependencies and training-inference alignment explicit. Within this design space, we study three representative policies: clean, same-level, and progressive history. On the full VBench prompt set, same-level and progressive history achieve aggregate scores of 85.24 and 85.60, respectively, compared with 84.45 for the clean-history reference. Long-video comparisons further show improved subject consistency and more coherent motion with progressive history. By allowing multiple denoising nodes to be processed together, progressive-history pipelining achieves $1.57$-$2.83\times$ steady-state DiT speedups under our evaluated conditions. We additionally find that LoRA adaptation of the DMD fake-score network improves generation quality using only 2.15% as many trainable fake-score parameters as full-parameter adaptation. Together, these findings show that fully denoised history is not a prerequisite for high-quality streaming generation and motivate the joint design of historical conditioning, execution, and training.
Read the full abstract
Computer Vision
Brain-to-Image Generation: Reconstructing Visual Stimuli from EEG using Generative Adversarial Networks
Reconstructing visual stimuli from electroencephalography (EEG) is difficult because scalp measurements have high temporal but limited spatial resolution, and paired EEG-image datasets remain small relative to modern generative-model training corpora. We present a reproducible single-subject baseline on THINGS-EEG2 that first tests the more defensible question of whether EEG can retrieve the viewed stimulus in a visual embedding space. A compact temporal-spatial convolutional encoder maps repetition-averaged EEG (63 by 250) to provided 512-dimensional ViT-B/32 image features. Model selection uses a concept-disjoint validation split, and final evaluation uses the official 200-image, 200-concept test gallery. Across three training seeds, the model obtains 12.83 +/- 0.58%, 39.17 +/- 1.76%, and 58.00 +/- 1.73% image recall at 1, 5, and 10 (mean +/- sample standard deviation), compared with analytical chance levels of 0.5%, 2.5%, and 5.0%. A session-balanced ablation shows that averaging more test repetitions generally improves ranking. Applying the Subject 01 model to the other nine subjects without adaptation causes a sharp performance drop, exposing subject specificity. We further report exploratory stress tests of direct conditional generators trained without external visual weights: single-subject and ten-subject variants produce noise-dominated outputs, with early validation improvements reversing after one to four epochs. Finally, we distinguish direct reconstruction from semantic rendering with a pretrained diffusion prior. The results support above-chance coarse semantic decoding under a closed-set, repetition-averaged protocol, but do not support faithful recovery of stimulus pixels.
Read the full abstract
Computer Vision
Performance vs Consistency: Evaluating a Foundation Model in Lung-RADS Screening
Foundation models have recently demonstrated strong capabilities across a wide range of medical imaging tasks. However, their performance in structured clinical interpretation settings remains insufficiently explored. In lung cancer screening, interpretative variability persists despite standardized frameworks such as Lung-RADS. In this study, we evaluate MedGemma, a medical general-purpose foundation model derived from Gemini and its fine-tuned version adapted for lung cancer detection and diagnosis, compared against radiologists performing Lung-RADS v2022 assessment on the NLST dataset. Twelve radiologists independently evaluated each case in a multi-reader design, enabling quantification of inter-reader variability. Radiologists achieved a mean AUC of 0.90, with substantial variability across readers (range: 0.80-0.94). The native foundation model achieved an AUC of 0.70, failing to reach clinically relevant performance. In contrast, fine-tuning significantly improved performance to an AUC of 0.83, placing the model within the lower range of individual radiologists performance. These findings highlight a trade-off between peak accuracy and prediction consistency. Unlike radiologists, under fixed conditions, the model produces deterministic outputs, removing inter-run variability under identical inputs, in contrast to inter-reader variability observed among radiologists. This supports the role of fine-tuned foundation models potential complementary tools for clinical decision support, particularly in settings with limited expertise. However, evaluation is performed on a case-enriched cohort from NLST and does not account for real-world prevalence or external validation, limiting direct clinical generalization.
Read the full abstract
Logic
Every introenumerable set contains a uniformly introreducible subset
We prove that every infinite introenumerable set contains an infinite uniformly introreducible subset, answering both parts of Question 1.7 of Greenberg, Harrison-Trainor, Patey, and Turetsky affirmatively. Moreover, a single procedure computes the original set from every infinite subset of the selected set. The proof combines the uniformization and refinement results of these authors with Ramsey-theoretic methods and a canonical reconstruction argument. We also obtain refinements controlling the ordinals and hyperjumps of the selected subsets.
Read the full abstract
General Relativity and Quantum Cosmology
Gravitating sources with zero energy density in cosmology
A cosmological source can have identically zero comoving energy density and still gravitate through its pressure. This differs from an energy density passing through zero at an isolated time. Maintaining zero density with nonzero pressure requires energy exchange in a nonstatic FLRW universe. We construct proportional-pressure and constant-pressure backgrounds, including steady-state, bouncing, recollapsing, and $\Lambda$CDM expansion histories, and relate them to creation pressure, bulk viscosity, and time-dependent vacuum descriptions. Scalar fields can realize zero density along a trajectory; imposing it as an identity on a timelike $P(X,\phi)$ domain selects $P=A(\phi)\sqrt{X}$, locally a signed potential-free cuscuton for nonzero $A$. We formulate linear perturbations without dividing by the vanishing background density. For a nondegenerate $P(X,\phi)$ scalar with a timelike field gradient, zero density and negative pressure cannot coexist with positive kinetic and gradient coefficients when the interaction leaves its principal kinetic structure unchanged. Two models couple the scalar to number-conserving dust through a field-dependent mass. The proportional-pressure trajectory is a homogeneous saddle and has no growing power-law dust mode on its accelerating branch in the matter-sourced quasistatic subhorizon approximation. A distinct constant-pressure model reproduces the $\Lambda$CDM expansion history, but in the same approximation the dust mode that initially grows during matter domination reaches a maximum at low redshift and then decays. These restrictions apply to the stated scalar class and models. Exact Bianchi I solutions demonstrate separately conserved zero-density sources supported by anisotropic stress. An inverse-pressure construction distinguishes the density of an interacting constituent from an inferred dark-energy density that may cross zero.
Read the full abstract
Geometric Topology
$\mathbb{Z}$-Torus Exteriors and Small Knot-Surgery Four-Manifolds
We study a cut-and-paste operation in which torus neighborhoods in the author's symplectic building blocks $Y_K$ and $X_K$, associated to a genus-one fibered knot $K$, are replaced by marked exteriors of tori in $S^4$ with infinite cyclic complement group and two null peripheral slopes. We also recall the exact Luttinger-surgery realization of $M_K\times S^1$ from $\Sigma_g\times T^2$, keeping the knot-surgery/fiber-sum and Luttinger-surgery viewpoints in the same framework. For the trefoil block $Y_K$, two marked replacements give a simply connected manifold with intersection form $H$, hence a manifold homeomorphic to $S^2\times S^2$. A one-exterior gluing gives a smooth homotopy $4$-sphere. For the rank-six construction we use the identity double of two copies of $Y_K\setminus\nu\Sigma_2$, rather than the involutive gluing defining the original $X_K$. Two marked replacements along the surviving rim tori give a simply connected manifold with $e=8$ and $\sigma=0$. An explicit geometric basis has intersection form $3H$, so the resulting manifold is homeomorphic to $\#_3(S^2\times S^2)$. The Case II small-perturbation Seiberg–Witten invariant vanishes. The Seiberg–Witten invariant of the identity-glued rank-six family also vanishes; in particular, these manifolds are nonsymplectic.
Read the full abstract
Computation and Language
Making Political Text Scaling Comparable: Infrastructure and Hyperparameter Sensitivity for 17 Algorithms
Computational text-based ideal point estimation (CT-IPE) methods are usually compared as named algorithms, yet applying them involves numerous researcher choices that configure how political text is turned into position estimates. This paper argues that CT-IPE methods are better understood as configurable measurement pipelines than as fixed estimators. Building on a large-scale comparative experiment spanning 17 CT-IPE algorithms, 5,537 experimental runs, and approximately 4.25 million left-right position estimates, I describe the shared infrastructure that makes these heterogeneous methods jointly executable and quantify how sensitive their estimates are to alternative hyperparameter choices. Variance-partitioning and SHAP-based sensitivity analyses show that, for most algorithms, hyperparameter profiles explain little residual variance through a shared shift: 13 of the 17 algorithms exhibit ICC values below .10. Where this profile-level sensitivity is present, it is concentrated in a small number of consequential researcher choices, most notably the selection of the underlying language or embedding model, the seed keyword lists that anchor the construct, and the number of topics.
Read the full abstract
Game Theory
Decentralized Optimal Equilibrium Learning Over Dynamic Networks
This paper studies decentralized learning of socially optimal equilibria in finite normal-form games over dynamic communication networks. Each agent observes only its own realized payoffs, does not know the game a priori, and can communicate only with time-varying neighbors using low-bandwidth messages. We propose networked decentralized optimal equilibrium learning dynamics in which agents generate randomized semantic content/discontent signals from local payoff comparisons and exchange time-stamped time-stacked tables rather than raw actions, payoff information or local estimates/parameters. The method combines table fusion with temporal majority reconstruction to mitigate dynamic communication while preserving fully decentralized operation. We establish finite-time logarithmic regret guarantees, with an in-phase exploration perturbation, for optimal equilibrium selection under utilitarian and proportional-fair social welfare objectives. Simulation results further show that the proposed approach can effectively select socially desirable equilibria over dynamic communication networks.
Read the full abstract
Fluid Dynamics
Flame Bifurcation, Flashback Margins, and NOx Scaling in a Swirl-Stabilized Hydrogen Aero-Engine Combustor
Burning neat hydrogen in aircraft turbines eliminates carbon emissions, yet rapid reaction rates trigger nozzle flashback hazards and high nitrogen oxide emissions across flight throttles. This study investigates aerothermal holding mechanisms, flashback safety margins, and emission pathways in a dual-swirl combustor across equivalence ratios from 0.55 to 1.00. Three-dimensional simulations combine curvature-corrected shear-stress transport turbulence closure, dual-rate finite-rate and eddy-dissipation chemical kinetics, and discrete ordinates radiation, validated against experimental laser benchmarks using ASME grid standards. Advancing engine throttle triggers a topological flame bifurcation from a faceplate-attached M-flame at lean idle to a lifted V-flame above equivalence ratio 0.895, while wall flashback safety indices consistently exceed 3.42. Nitric oxide emissions transition from water-chaperoned nitrous oxide intermediate reactions at lean idle (28.01 ppm, EINOx = 1.85 g/kg), scaling to 319.21 ppm (EINOx = 35.40 g/kg) at takeoff under thermal Zeldovich dominance, governed by a power-law exponent of 4.92. These findings deliver validated operability limits and establish an accessible workstation-based screening methodology for practical zero-carbon aero-engine combustor development.
Read the full abstract
Genomics
Structure is not mechanism: high-gain gated-FFN rows across text and genomic foundation models
A small number of unusually high-gain parameters can exert disproportionate effects in transformer language models, but whether analogous structures recur in genomic foundation models and whether structural geometry determines functional importance remains unknown. We analyzed high-gain rows in gated feed-forward networks across text and genomic foundation models, including a frozen 22-model causal census. Computing an associated bilinear weight operator exactly, without a diagonal approximation, we tested whether structural extremeness is a transferable mechanism. Activation-derived candidates were functionally enriched relative to random and top-norm same-layer controls, yet neither spectral concentration nor operator magnitude predicted causal effect size, and these associations vanished within the endpoint-homogeneous text-decoder subset. A within-layer sweep of 36 rows in one genomic and one text decoder resolved this into two regimes: below the detector's acceptance threshold the ratio carried no positive information about causal damage, whereas above it the ratio ordered rows strongly but did not grade severity as a dose-response. The same sweep revealed a second individually catastrophic row invisible to a one-candidate-per-model census, and non-additive damage among co-located critical rows. Case studies showed divergent causal organizations: a robust super-additive pair interaction in DNABERT-2, and in GENERator a sharply position-localized dependence in which preserving or restoring the row's beginning-of-sequence contribution rescued essentially all native-loss damage. High-gain gated-FFN rows are therefore a recurrent architectural phenotype whose structural prominence acts as an enrichment signal, not a calibrated measure of functional criticality or a specification of causal organization. Enrichment is general, but the mechanism is model-specific.
Read the full abstract
General Mathematics
Largest Circle Enclosing Exactly $n$ Interior Lattice Points. II
In a previous article doi.org/10.3390/geometry2030012, the author investigated a class of elementary plane geometry problems closely related to the theme of this work. Here, we prove a weaker version of a previous conjecture by demonstrating that there are infinitely many maximally circlable (MAC) numbers – positive integers $n$ for which there exists a largest circle enclosing exactly $n$ interior lattice points. Furthermore, by extending numerical computations to $n \le 2700$, we identify two counterexamples to a conjecture in loc cit. regarding the symmetry of the largest circle enclosing a strong MAC number (a MAC number $n$ where $n+1$ is non-MAC). We also propose a potential infinite family of strong MAC numbers derived from Pythagorean triples; the existence of this family would imply the infinity of non-MAC numbers, as conjectured by Zhao. Throughout this paper, we provide extensive data characterizing both MAC and strong MAC numbers alongside their corresponding largest enclosing circles.
Read the full abstract
High Energy Astrophysical Phenomena
Bayesian Inferences on Analytical Equations of State Approximations of Neutron Stars
Equations of state (EoS) for dense matter are commonly provided in tabulated pressure-energy density relations, complicating numerical implementation and potentially compromising thermodynamic consistency. In this work, we propose a piecewise-continuous analytical form with a common parametrization capable of representing a broad class of dense matter EoSs. We validate this parametrization against tabulated EoSs from the CompOSE and LALSimulation repositories. Following the initial deterministic fitting, we employed two distinct Bayesian approaches: one based on synthetic EoSs generated by adding noise to the fitted EoS and another based on multi-messenger measurements of neutron-star masses and tidal deformabilities. In both approaches, the initial best-fit parameters are used as reference values. For the considered EoSs, spanning from very soft to very stiff, the resulting fits reproduce the tabulated EoSs and the associated macroscopic neutron-star observables in the repositories with sufficient accuracy. The inferred parameters exhibit strong correlations arising from the continuity conditions imposed at the segment boundaries. These correlations persist in the low- and intermediate-density segments under multi-messenger inference but become weak in the core, reflecting the limited constraining power of current observations at high densities. The ability of a single functional form to describe EoSs derived from different microscopic frameworks, together with the recurrence of similar parameter correlations, suggests that diverse dense matter EoSs may share a common underlying structure.
Read the full abstract
Programming Languages
QuickerChick
Property-based testing (PBT) with QuickChick relies on extracting Rocq programs to OCaml. However, this extraction mechanism, while crucial for QuickChick to function, has significant performance implications. In this work, we describe how we optimized QuickChick, exploiting various opportunities offered by program extraction to significantly improve each test's extraction, compilation, and running time. We also evaluate these improvements using the ETNA benchmarking platform for PBT to assess how individual improvements impacted the overall test run time.
Read the full abstract
Information Theory
Further results on binary linear covering codes and saturating sets in projective spaces
The binary length function $\ell_2(r,R)$ is the smallest length of a binary linear code with codimension (redundancy) $r$ and covering radius $R$. Let $s_2(N,\rho)$ be the smallest size of a $\rho$-saturating set in the binary projective space $\mathrm{PG}(N,2)$. It is known that the binary length function satisfies $\ell_2(r,R)=s_2(r-1,R-1)$. We obtain the following new upper bounds on $\ell_2(r,R)$, improving the previously known upper bounds by $\Delta(r,R)$: $R=2,r=2t\ge28$, and $r=10,18,20$, $\ell_2(r,2)=s_2(r-1,1)\le51\cdot2^{r/2-5}-1,\Delta(r,2)=2^{r/2-5}$; $R=3,r=3t-1\ge41$ and $r=26$, $\ell_2(r,3)=s_2(r-1,2)\le818\cdot2^{(r-26)/3}-1$, $\Delta(41,3)=63,\Delta(r,3)=2^{(r-26)/3}$ if $r\ge44$. To obtain these bounds, we construct new infinite code families, using distinct versions of the $q^m$-concatenating constructions of covering codes, including new versions proposed in this paper. We also obtain new useful partitions of column sets of parity check matrices of some codes. The new code families provide the asymptotic covering-density bounds $\overline{\mu}(2)\le2601/2048<1.27002$, $\overline{\mu}(3)\le1.35935$, improving the previously known bounds and giving rise to the new upper bounds $f(2)\le2601/2048<1.27002$, $f(3)\le1.35935$ on the constants $f(2),f(3)$ of the Green's Open Problem 40.
Read the full abstract
Machine Learning
Pseudo-Label Augmentation for Affect Sensing in Small Collaborative Groups
Physiological affect sensing in naturalistic group interaction is often limited by sparse labels rather than sensor data: wearable devices produce many time windows, while self-reports are collected only a few times per session. Using GroupAffect-4, a four-person collaborative dataset with wearable physiology, eye tracking, Big Five personality, and post-task VAD labels, we study pseudo-label augmentation for affect sensing under sparse supervision. We compare no augmentation, Gaussian Process pseudo-labelling, personality-aware trust weighting, and joint personality-plus-confidence weighting within a shared target-construction pipeline. Results show that pseudo-label augmentation improves over the labelled-only baseline in the known-team setting. However, the narrow range of Big Five cosine similarities (0.91-0.99) makes fine-grained personality weighting ineffective; personality similarity functions mainly as a same-team filter rather than a calibrated trust signal. With smoothing, augmented SVM variants are effectively tied on Valence and Arousal, while the joint personality-plus-confidence variant gives the highest Dominance score. Cross-subject LOSO transfer remains encouraging, especially for Arousal, whereas strict session-isolated LOGO removes the augmentation benefit. Given only 10 groups, LOGO should be interpreted as a conservative lower bound on unseen-group transfer. Overall, the results suggest that pseudo-label augmentation can make better use of sparsely labelled collaborative affect data, while personality information is most useful as a within-team selection mechanism.
Read the full abstract
Computation and Language
The Imitation Game: When LLMs Learn to Reason Like Programs via Code-Centric Reasoning Data Synthesis
Large Language Models (LLMs) excel at programming tasks but frequently fail at deterministic, fine-grained reasoning in natural language, relying heavily on semantic approximations rather than robust symbolic execution. To bridge this gap, we propose MIMIC, a framework that leverages executable code as a rigorous medium for reasoning data synthesis. MIMIC fundamentally transforms algorithms into verifiable reasoning trajectories through narrative fusion, code-guided test synthesis, and dynamic code instrumentation. Crucially, these explicit intermediate execution states naturally form a Code-Instrumented Reward (CIR), providing dense, high-fidelity process supervision for reinforcement learning without external reward models. Extensive evaluations reveal that models trained via SFT and GRPO on our synthesized dataset achieve substantial, consistent gains. Our method significantly elevates accuracy across general reasoning, complex mathematical benchmarks, and fine-grained deterministic tasks, demonstrating that the procedural rigor of executable code can effectively unlock and enhance the generalized reasoning capabilities of LLMs. Our code and data are available at https://github.com/zjy1298/MIMIC.
Read the full abstract
Robotics
AssemblyGrid v1: A Benchmark for Multi-Robot Production with Temporary Coalitions, Local Information, and Geometric Constraints
Flexible robotic production requires joint decisions on process progression, material routing, resource assignment, temporary cooperation, and simultaneous execution, since each decision can affect the feasibility of the others. The challenge is greater under decentralized control, where each robot acts from bounded local information while system progress depends on collective decisions, shared resources, material state, and workspace compatibility. These properties closely match cooperative multi-agent decision making under partial observability and resource contention. This paper introduces AssemblyGrid v1, a reproducible benchmark for repeated multi-robot production that combines explicit process progression, decentralized observations, material transfer, temporary multi-robot coalitions, productive concurrency, and geometry-dependent feasibility within one task-level formulation. The benchmark includes Flow, Coalition, and Concurrency workload families, each with three scenario levels. Task success and evaluation measures are defined independently of learning reward and solution method, allowing learning-based and non-learning methods to address the same production problem. AssemblyGrid v1 is evaluated through executable conformance checks, mechanism studies, and algorithmic experiments using a privileged centralized reference, structured decentralized controllers, and MARL methods including IPPO, MAPPO, and QMIX. Results demonstrate productive execution under centralized and decentralized control. The MARL experiments further show that decentralized policies can learn effective production behavior from local observations and actions, supporting AssemblyGrid as a controlled benchmark for studying cooperative decision making in flexible robotic production.
Read the full abstract
Robotics
World-Action Models for Robot Learning and Control: A Survey
Robots operating in open environments act under partial observability, physical constraints, and dynamic task contexts. Beyond mapping observations and language instructions to actions, they must anticipate how candidate actions may affect future states and task-relevant outcomes. Recent advances in world models, video generation, and Vision-Language-Action (VLA) policies have motivated the development of World-Action Models (WAMs), which couple future world prediction with executable action generation. This survey provides a robotics-oriented review of WAMs. We clarify their scope relative to conventional world models, model-based reinforcement learning, action-conditioned video generation, and reactive VLA policies, and organize existing methods through a unified taxonomy covering representations, transition modeling, action interfaces, architectures, training pipelines, data modalities, and scaling strategies. We further review applications of WAMs in manipulation, navigation, and autonomous driving, and we summarize the datasets, benchmarks, metrics, and protocols used to evaluate WAM systems. Finally, we discuss key challenges in action alignment, world-action factorization, spatial and multi-view consistency, long-horizon memory, neural simulation for closed-loop policy learning, and efficient inference. Taken together, this survey aims to provide a concise technical foundation for integrating predictive world modeling with action generation, toward more reliable embodied robot intelligence. Project page: https://rcl-robotics.github.io/Awesome-World-Action-Models.
Read the full abstract
Computation and Language
The Immutable Past: Formalizing State Mutability and Conflict Resolution in Mutable RAG
Retrieval-Augmented Generation (RAG) serves as the primary memory architecture for long-horizon autonomous agents. However, treating shared memory as an append-only stream introduces Semantic Shadowing, a critical failure mode where conflicting historical observations accumulate and statistically dominate valid recent updates. In dynamic environments, this results in severe state divergence as agents retrieve and act upon obsolete facts. This paper formalizes the mechanics of State Mutability to prove that standard dense retrieval suffers from Asymptotic Recall Decay. Furthermore, we formally demonstrate a Majority Vote Trap, revealing that increasing the retrieval context window paradoxically degrades generation accuracy by diluting the attention mechanism under conditions of semantic equivalence. To resolve this, we introduce GC-Mem (Garbage Collection for Memory), a strict inference-time consistency protocol. Unlike heuristic time-decay mechanisms—which indiscriminately destroy valid long-term memory—GC-Mem relies purely on a temporal dominance operator ($\Phi_{\mathcal{T}}$) paired with contradiction detection to surgically excise shadowed context. Evaluated across a rigorous, behaviorally inferred benchmark of 137,760 memory chunks and continuous accumulation sweeps, standard RAG and timestamp re-ranking baselines experience severe degradation. In contrast, GC-Mem empirically recovers $>90\%$ conflict resolution accuracy. We establish strict precision and recall deployment thresholds, ensuring state convergence where standard mutable RAG fundamentally fails.