Instrumentation and Methods for Astrophysics
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Sunday 6 September
Near-Earth Objects (NEOs) can appear as faint streaks in long-exposure astronomical images. Detecting these streaks across diverse observatories requires methods that remain reliable despite differences in image quality, orientation, sky background, and noise. However, existing detectors are commonly evaluated using data from only one source, providing limited evidence of cross-source generalization. We introduce NEO-Bench, a multi-source benchmark containing 8,376 images from five astronomical-image datasets. The sources include the Hubble Space Telescope, a Stellina smart telescope, the United Arab Emirates Meteor Monitoring Network, a TETRA1 telescope using a Celestron C14 with Fastar, and the Roboflow Asteroid dataset. We converted the data to a common YOLO format, audited a sample of labels, and defined within-source and leave-one-source-out evaluation protocols. We evaluated four approaches: Hough, Radon, Gaussian PSF, and YOLO26L. Leave-one-source-out F1 decreased in 14 of 20 image-level method-source pairs and 13 of 20 IoU@0.50 localization pairs. Across the datasets categorized as medium or hard, F1 decreased in 11 of 12 image-level pairs and 9 of 12 localization pairs. These results show that cross-source performance remains inconsistent and that reliable generalization across astronomical imaging sources remains an open challenge. The benchmark, code, and data are publicly available.
Cryptography and Security
Robustness-Aware Evaluation and Enhancement of Mutation-Based Fuzzing for Bug Discovery
Mutation-based fuzzing is widely used to discover software vulnerabilities, but its randomness complicates rigorous evaluation and reliable bug detection. Prior work measures this variability empirically but lacks a theory with computable convergence and sample-complexity guarantees. We address both problems. First, we estimate robustness from independent campaigns by measuring variation in bug-trigger rates after accounting for compute. For $M$ campaigns of length $T$, finite-trial error decreases at the standard $M^{-1/2}$ rate, while finite-length error is bounded when temporal bug correlations decay. We then introduce splitting, a black-box wrapper that copies a fuzzer's queue state after a bug trigger and continues from that state in multiple branches, directing more effort toward the discovered region. For any realized split tree, branching cannot reduce the raw number of bug-triggering events relative to a single continuation path. Expected detection also improves when states receiving more branches tend to yield more bugs later. In a simplified model, splitting reduces variance per unit compute when $p<\sqrt{2}-1$, where $p$ is the fraction of time spent in the bug region. At matched compute, splitting finds more real bugs per CPU-hour than the baseline in 38 of 40 Magma ground-truth cells (median +52%; 34 of 38 individually significant) and never finds fewer distinct bugs. It changes CVE-2019-19926 from undetected (0/20 trials) to reliably detected (20/20; Fisher $p<10^{-4}$), with six additional detection improvements, five involving CVEs. On FuzzBench, splitting finds more unique bugs in 53 of 70 pairs and reduces cross-campaign variation in 66 of 70, with a median reduction of about $10\times$. Every branch counts toward the compute budget. With about 0.14% overhead, splitting provides a practical way to measure and improve fuzzers.
Read the full abstract
Software Engineering
CAROL: Context-Aware Online Learning for Fuzzer Scheduling
Ensemble fuzzing runs multiple fuzzers on a target while a scheduler allocates CPU time among them. Existing schedulers base these decisions on compact summaries of past performance and rules fixed before a campaign. Our measurements reveal two limitations. First, past-reward summaries do not reliably capture performance evolution: after accounting for estimation noise, agreement between consecutive-window rankings is statistically indistinguishable from within-window self-agreement. Second, predictive signals vary across targets: on eight of nine targets, a weighting learned from the other eight predicts reward worse than one learned on the current target. We introduce CAROL, an online scheduler that uses each fuzzer's current context. Already available to the dispatch loop, this context describes reward trends, waiting and plateau time, reached code, and estimation uncertainty. CAROL uses context in two ways: a domain-guided method detects whether a fuzzer is rising or rotting and applies a phase-specific learning rule, while a learned method predicts reward from 15 context signals and uses predictive uncertainty for online selection. Across nine Magma targets, CAROL triggers more unique bugs than each of three ensemble-scheduling baselines whenever their results differ, and fewer on none. Compared with the strongest baseline for each target, CAROL gains 11.8% and surpasses an oracle that retrospectively selects the best single fuzzer per target. Removing context eliminates the gain, and the additional bugs are concentrated among those the baselines trigger rarely or never. Run unchanged on five widely used C++ programs, CAROL finds 120 previously unknown crashing defects, deduplicated by site, fault, and entry point; all were reported to maintainers through the projects' stated disclosure channels.
Read the full abstract
Computation and Language
AuthBench: A Large-Scale Multilingual Benchmark for Authorship Representation across Genres and Lengths
Authorship signals matter in settings where writing style carries identity: digital forensics, plagiarism analysis, account linking, misinformation investigation, and machine-generated text detection. Yet current authorship benchmarks remain fragmented, usually covering only a narrow language set, a single genre, or a limited document-length regime, which makes it difficult to assess whether modern representations truly generalize. We introduce AuthBench, a large-scale multilingual benchmark for authorship representation that is designed to make this evaluation broad, standardized, and realistic. AuthBench contains 428,150 documents written by 153,825 individuals across ten widely used languages, 9 primary genres, 66 fine-grained genres, and four document-length buckets. It supports two complementary tasks: authorship attribution, formulated as same-author retrieval and authorship verification, formulated as same-author binary decision. We benchmark 47 neural models and three non-neural baselines under a unified zero-shot protocol. Results show that authorship representation remains far from solved: the best retrieval model reaches only 0.258 Success@5, while the best verification model achieves 0.076 EER and 0.968 ROC-AUC. The leaderboard also reveals a meaningful task split, with different model families leading retrieval and verification, and large performance differences across languages, genres, and lengths. These findings position AuthBench not only as a new benchmark, but as a diagnostic resource for studying when and why authorship representations succeed or fail. We release AuthBench, its evaluation toolkit, and benchmark data at https://github.com/mao-code/AuthBench and https://huggingface.co/datasets/MaoXun/AuthBench.
Read the full abstract
Human-Computer Interaction
The Interface of Theseus: The Rise of Just-In-Time Interfaces
Our community creates interfaces—pre-constructed, static artifacts. These interfaces are effortful to create, so we invest great effort and care into each of them. Yet the equilibrium is shifting: we can now construct nearly any software, bespoke, within moments. When construction is cheap, we may soon live in a future in which interfaces are constructed in real time. This is a world of just-in-time interfaces: rapid, discardable, on-demand interfaces tailored to our needs in the moment. Will our community still create traditional interactive systems in this world? Or will we create interface generators, “interface.md” guides for AI consumption? I argue that we ought to lean into a future of rapid, bespoke, on-demand interfaces.
Read the full abstract
Computers and Society
Ordinary, Reasonable Chatbots: Do AI Models Track Human Legal Judgments?
As people increasingly rely on artificial intelligence (AI) for guidance in their own lives, scholars, lawyers, and even judges have begun to consider the role of AI in legal decision-making. As "silicon sampling" – the use of generative AI models in social science research – is now impacting academia, "silicon jurors" could make an appearance in courtrooms. This study joins an emerging line of research on generative AI models' ability to simulate human legal judgments. In particular, we study how large language model (LLM)-powered chatbots respond to series of questions about legal reasonableness. When the law needs to judge the appropriateness of a behavior, it most often asks whether the behavior was "reasonable." Yet despite the ubiquity of reasonableness judgments, they are the site of constant vexation for lawyers, judges, and lay people. Reasonableness seems inherently vague and unpredictable, since it relies on variable context and implicit conceptual schemas. Moreover, many scholars caution that reasonableness judgments may vary along demographic lines. We compare the answers of human participants to those of twenty-six LLMs across twenty-five different legally relevant reasonableness judgments. Overall, our findings suggest that chatbot responses generally track those of human participants. Nonetheless, we find some suggestive – and potentially concerning – results. Compared to humans, LLMs generate more homogeneous responses and occasionally treat a variable standard as an invariant rule. And, compared to humans, LLMs tend to generate answers that are more favorable to the government and to corporations. Finally, our results indicate that LLMs' responses tend to align more closely with those of respondents who are white, male, older, and more educated. More systematic research is needed to confirm or reject these initial findings.
Read the full abstract
Representation Theory
Koszul duality for finite-dimensional Koszul algebras
Let $\La$ be a finite-dimensional Koszul algebra with Koszul dual $\Ld$. Using the Koszul functors and the homological perturbation lemma, we describe the minimal projective resolution of every object of the bounded derived category of finite-dimensional graded $\La$-modules in terms of the cohomology of the corresponding complex of coperfect graded $\Ld$-modules. In particular, the linear part of the minimal graded projective resolution of a finite-dimensional graded $\La$-module is a direct sum of Koszul complexes of these cohomology modules. We deduce that $\La$ is absolutely Koszul if and only if $\Ld$ is graded left co-coherent, if and only if the category of coperfect graded $\Ld$-modules is abelian. As applications, we show that absolutely Koszul algebras of finite global linearity defect have finite graded finitistic dimension; we refine the graded and ungraded Koszul dualities and construct nonstandard $t$-structures on the bounded derived and singularity categories. Finally, we apply these results to finite-dimensional quadratic monomial algebras.
Read the full abstract
Commutative Algebra
Frobenius dilation of Hilbert function, Demailly's conjecture, and varieties of powers
We prove a Frobenius-dilation inequality for Hilbert functions of ideals generated by powers of fixed homogeneous forms in characteristic zero. Via Emsalem–Iarrobino inverse system duality, it yields quantitative fat-point postulation estimates and a new proof of the recent theorem of Ha–Sivakumar resolving Demailly's conjecture for arbitrary finite point sets. A fixed-characteristic version controls tangent spaces to varieties of powers and, together with plane interpolation and secant-theoretic results, proves complete nondefectivity of $V^k_{2,d}$ for all $k\ge3$ and $d\ge2$; in particular, it settles the ternary Waring problem for generic $k$-rank conjectured by Lundqvist–Oneto–Reznick–Shapiro (suggested by Ottaviani). We also show that, for a prime power exponent $q$, any failure of Nicklasson's conjecture in three variables is confined to $2q-2$ consecutive degrees.
Read the full abstract
History and Overview of Mathematics
Illustrating Hyperbolic Surfaces with Mesh Embeddings
Hyperbolic geometry exhibits geometric phenomena, such as fast area growth, that are difficult to visualize faithfully in Euclidean space, and which standard models like the Poincaré disk can obscure. To bring hyperbolic geometry to life, we embed hyperbolic surfaces in Euclidean space by discretizing the surfaces into meshes, and minimizing a distortion energy so that the edge lengths in the embeddings match those in the hyperbolic plane. The resulting surfaces buckle and ruffle to accommodate the extra area, making visible what flat models hide. We present exemplary illustrations, such as embedded disks, equidistant strips, diverging geodesics, and also artistic organic-like renders. We discuss our use of these models, as renders and 3D prints, in research talks, public engagement, outreach, and education.
Read the full abstract
Differential Geometry
Uniform doubling on compact homogeneous spaces with cyclic Lie brackets
We establish the uniform doubling property for $G$-invariant metrics on a wide class of compact Riemannian homogeneous spaces $(G/K,g)$, describing explicitly the volume growth of the metric balls $B_g(p,r)$. This property was previously known only for specific cases, including abelian Lie groups, the Lie group $SU(2)$ and quotients of $SU(2)\times \mathbb R^n$. Building on the approach of Eldredge, Gordina and Saloff-Coste for $SU(2)$, we refine and develop geometric and Lie-theoretic tools that allow us to replace the explicit identities of the Milnor basis in $SU(2)$ by a general structural assumption on the metric eigenspace decomposition. In particular, we show that the uniform doubling of $(G/K,g)$ is a consequence of the cyclic bracket condition $[\mathfrak{m}_i,\mathfrak{m}_j]=\mathfrak{m}_k$, for $i,j,k$ pairwise distinct, a property that naturally generalizes the Milnor structure of $SU(2)$. We apply our results to $\mathbb Z_2\times \mathbb Z_2$-symmetric spaces, establishing the uniform doubling property for the complete family of $G$-invariant metrics on several classes of generalized Wallach spaces. For compact homogeneous spaces, we also derive a global Poincaré inequality which holds uniformly for the spaces $(G/K,g)$ under consideration, with a constant controlled by the volume doubling constant of the space.
Read the full abstract
Solar and Stellar Astrophysics
Elusive helium stars in the gap between subdwarfs and Wolf-Rayet stars III. Formation of the Galactic population of stripped helium stars
Population synthesis based on the grid of pre-computed evolutionary sequences for close binaries is applied to the problem of formation of the Galactic population of stripped helium stars, the remnants of $5 \lessapprox M/M_\odot \lessapprox 24$ primary components of close binaries after RLOF. Stellar rotation is taken into account. For the current star formation rate in the Galaxy $2\,M_\odot yr^{-1}$, the estimate of the number of 1 to $7\,M_\odot$ objects with $T_{eff} \gtrapprox $25000K ranges from 14700 to 28500, while for the number of objects with masses from $2\,M_\odot$ to $7\,M_\odot$ it is 3200 to 5500, depending on the assumptions about formation of common envelopes and their outcomes. Other factors influencing the estimate are adopted star formation rate, distributions of close binaries over initial masses of primary components, mass ratios of components, orbital periods. Observational selection may severely limit the number of detectable objects. Angular velocities of rotation of companions to stripped helium stars are 60 to 80 per cent of the critical one. Using the models of initially homogeneous helium stars as approximations for the remnants after mass loss results in the underestimate of pre-supernovae masses.
Read the full abstract
Astrophysics of Galaxies
Fractional Bose-Einstein Condensate Dark Matter Cores: Numerical GPP Solutions and SPARC Rotation Curves
We investigate a fractional Bose Einstein condensate dark matter scenario in which an ultralight bosonic component forms a self gravitating core embedded in a predominantly cold dark matter halo. The stationary GPP system solved numerically beyond the TF approximation leads to energetically bound equilibrium configurations and good phenomenological fits to the selected rotation curves from 30 SPARC galaxies.
Read the full abstract
Analysis of PDEs
Higher order KdV-type equations with Robin boundary conditions on $\mathbb{R}^+$
We study the initial boundary value problem (IBVP) for the higher order Korteweg–de Vries type equation \[\partial_tu+(-1)^{j+1}\partial_x^{2j+1}u+\frac12\partial_x(u^2)=0,\qquad j\in\mathbb{N},\] on the right half line, subject to the Robin boundary conditions \[(\partial_x+\gamma)\partial_x^{\ell-1}u(t,0)=\varphi_\ell(t),\qquad 1\le\ell\le j,\] where $\gamma\in\mathbb{R}$ is common to all boundary conditions. We prove local well posedness for \[u_0\in H^s(\mathbb{R}^+),\qquad \varphi_\ell\in H^{\frac{s+j-\ell}{2j+1}}(0,T),\qquad -j+\frac14<s<\frac32.\] In particular, the regularity range extends the low regularity theory for the higher order Dirichlet IBVP to Robin boundary conditions. The main ingredient is an explicit unified transform representation for the higher order IBVP with Robin boundary conditions. The Robin hierarchy couples adjacent boundary traces and leads, in each spectral sector, to a nontrivial system for the unknown boundary transforms. Rather than computing the inverse of this system, we resolve precisely the linear combination required by the unified transform representation through Lagrange interpolation at the rotated spectral points. This yields an exact factorization in which the dependence on the Robin parameter is concentrated in the single factor $(k-i\gamma)^{-1}$. Consequently, the only possible Robin pole is $k=i\gamma$, which contributes the residue mode \[e^{-\gamma x+\gamma^{2j+1}t}\] exactly when $j$ is odd and $\gamma>0$. Combining this representation with linear estimates in modified Fourier restriction spaces and higher order KdV bilinear estimates yields the low regularity well posedness result. The representation also recovers the classical KdV formulas with Robin and Neumann boundary data when $j=1$ and is formally consistent with the higher order Dirichlet representation under the reciprocal Robin limit.
Read the full abstract
Machine Learning
LATS: Levy Adaptive Tree Sampling for Feedback-Driven Diverse Target Discovery
While diffusion models excel at capturing complex data distributions, scientific discovery often requires steering generation toward specific, uncharacterized regions that maximize a target objective. These high-utility modes frequently reside in low-likelihood tail regions and are only revealed sequentially through interactive feedback. Existing diffusion samplers fail in this regime: they inherit the pre-trained model's bias toward high-density regions, leaving rare yet promising phenomena underexplored. Conversely, exploration-heavy samplers ensure broad coverage but fail to efficiently exploit high-utility modes when constrained by a strict sampling budget. To resolve this dilemma, we introduce Levy Adaptive Tree Search (LATS), a principled sampling framework for online feedback-driven search. LATS leverages heavy-tailed exploration coupled with tree-based value backpropagation to progressively uncover preferred modes. By maintaining broad distributional coverage, LATS successfully discovers low-likelihood, high-utility regions while preserving sample fidelity and structural diversity. Experiments across diverse benchmarks, including materials science, demonstrate that LATS significantly outperforms baselines in target discovery efficiency.
Read the full abstract
High Energy Physics (Phenomenology)
Solar Capture Tests of Inelastic Dark Matter after the LZ High-Recoil Event
The LUX-ZEPLIN (LZ) Collaboration has reported a $248$ keV nuclear-recoil candidate for which endothermic dark matter (DM) gives some of the largest local significances. We study Solar-capture constraints on three interpretations: a thermal Higgsino, a thermal pseudo-Dirac fermion with off-diagonal vector interactions, and neutron-philic endothermic spin-dependent scattering through $\mathcal{O}_4$. The canonical full-density thermal Higgsino is excluded: its LZ-preferred splitting near $0.38\,\mathrm{MeV}$ lies well below the splitting required to suppress Solar capture sufficiently to satisfy the IceCube upper limits on DM annihilation in the Sun, $\delta\gtrsim0.51$-$0.56\,\mathrm{MeV}$. For the pseudo-Dirac benchmark parameters that fit the LZ event, we find $C_\odot=1.53\times10^{20}\,\mathrm{s}^{-1}$, while two-state kinetics limits the fixed-orbit annihilation rate to $\Gamma_A\lesssim4.4\times10^{14}\,\mathrm{s}^{-1}$, about $1.7\times10^5$ below the IceCube upper limit for the $b\bar b$ channel. A semi-analytic treatment indicates that re-excitation cycles further cool the captured population, although a full phase-space calculation is required for its final distribution. For neutron-philic $\mathcal{O}_4$ scattering at $m_\chi=1$ TeV and $\delta=300$ keV, finite-temperature capture gives $C_\odot=3.98\times10^{17}\,\mathrm{s}^{-1}$, with nuclear-structure uncertainties giving an envelope $(1.77$-$9.72)\times10^{17}\,\mathrm{s}^{-1}$. Under the equilibrium assumption, the upper edge remains about a factor $154$ below the IceCube upper limit for the $b\bar b$ channel. We use $b\bar b$ only as a soft-hadronic proxy because an exact constraint requires the model-specific annihilation spectrum. Solar capture therefore excludes the thermal-Higgsino interpretation but not endothermic explanations generically.
Read the full abstract
Algebraic Geometry
Hasse Obstructions to Rationality for Special Fourfolds
For every nonempty Hassett divisor $\mathcal C_d$ parameterizing special cubic fourfolds, we consider the quaternion class $\beta_d=(d/2,-3)$ in the two-torsion Brauer group. We prove that if a Hodge-general member of ${C}_d$ is rational then $\beta_d=0$ – or, equivalently, that Huybrechts' twisted-K3 condition $(**')$ holds. Consequently, a very general member of $\mathcal C_d$ is irrational if $\beta_d\ne0$. Thus, a very general cubic fourfold containing a smooth cubic scroll or a Veronese surface is irrational, and hence so is a very general Küchle fourfold of type $(\mathrm{c7})$. We also obtain an analogous obstruction for Hodge–special Gushel–Mukai fourfolds: a very general member in the discriminant–$d$ locus can be rational only if $d$ is a sum of two squares. Consequently, a very general Gushel–Mukai fourfold containing a cubic scroll is irrational.
Read the full abstract
Computer Vision
Agentic Visual Generation: From Generative Models to Agentic Control
Visual generation is evolving from generative models used through a single invocation into agentic control processes that can plan, select tools, inspect intermediate synthesized outputs, revise failures, and reuse prior experience. In most existing systems, the controller is an LLM or VLM, while visual generation models serve as tools or executors. However, existing work lacks a consistent criterion for determining when a generation system becomes agentic. Planning depth, tool use, multi-role collaboration, and reinforcement learning are often treated as evidence of agenticity, even though none of them necessarily determines which generation decisions the controller can make. We organize the field according to what the controller can directly control in the generation process. At L1 Conditioning Control, the controller prepares the input to a predetermined generator but does not control which visual operation is executed. At L2 Execution Control, it selects and invokes actual generation, editing, rendering, or other content-modifying operations. At L3 Outcome-Adaptive Control, it observes an intermediate outcome and uses that observation to change a subsequent operation within the current task. At L4 Experience-Adaptive Control, it retains experience from completed tasks and uses that experience to change decisions on future tasks. L0 Fixed Support separately denotes generators, editors, evaluators, reward models, benchmarks, and fixed pipelines without a deployed controller that makes generation-level decisions. These levels describe a progressively broader decision-making scope rather than model size, system complexity, output quality, tool or role count, or training method. Applying this framework across image, video, editing, 3D, world, slide, and user-interface generation reveals how controller capabilities have evolved and how their mechanisms are distributed across levels.
Read the full abstract
Optimization and Control
An Exact Dual for Second-Order Cone Programming Using Only Lorentz-Cone Constraints
Without a constraint qualification, second-order cone programs can have a duality gap or an unattained dual value. For every such program with arbitrary real coefficients, we construct an exact dual using only affine equations and memberships in products of Lorentz cones. Every feasible dual point gives a valid bound. Whenever the primal is feasible with finite value, the dual attains that value. This answers a question raised by Pólik and Terlaky and reiterated in subsequent work: whether an exact dual for second-order cone programming can use only Lorentz-cone constraints. The same construction produces an affine second-order cone system that is feasible exactly when a given affine second-order cone system is infeasible, including in the weakly infeasible case. Both formulations have polynomial size and are obtained uniformly from the input coefficients using only fixed rational constants. These are formulation-size results over exact real data; they do not imply polynomial-time solvability or polynomial bounds on the bit length of rational certificates.
Read the full abstract
High Energy Physics (Phenomenology)
LZ Nuclear-Recoil Excess from Boosted Light Magnetic Dipole-dipole Dark Matter
The LZ collaboration has reported a nuclear-recoil excess near 248 keV with a global significance of $2.6\sigma$. Although halo dark matter with a magnetic dipole-dipole interaction and a TeV-scale mass provides the best fit to the excess among the interactions considered by LZ, it predicts a considerable number of events at lower recoil energies, where no excess is observed. We show that a boosted velocity distribution can alleviate this tension and provide a better fit to the LZ recoil spectrum. Moreover, the boost opens up the possibility of explaining the excess with much lighter dark matter, with masses down to the GeV scale. We demonstrate these features first in a model-independent analysis and then realize them in a concrete dark matter model, in which halo dark matter annihilates into on-shell mediators that subsequently decay into boosted dark-sector particles. Our results demonstrate that boosted dark sector particles provide a viable interpretation of the LZ excess.
Read the full abstract
Group Theory
Subgroups of free groups with the same derived subgroup
We prove that if H and K are subgroups of a free group F of arbitrary rank and $H'=K'\ne 1$, then $H=K$. As a consequence, if R is a noncyclic subgroup of a free group F and $R'$ is normal in F, then R is normal in F. This gives a positive answer to a problem of Shpilrain, originally posed as Problem 11.124 in the Kourovka Notebook and later appearing as Problem F15 in Kapovich-Myasnikov-Shpilrain.
Read the full abstract
Probability
Central limit theorem for real zeros of random Weyl polynomials with general coefficients
For a random polynomial, the number of real zeros $N_{\mathbb R}$ is a highly nonlinear function of its coefficients, and its statistical properties have been studied extensively. A natural question is whether $N_{\mathbb R}$ satisfies a central limit theorem. For various ensembles with iid standard Gaussian coefficients, such central limit theorems have been established; see, for instance, arXiv:1911.12182, arXiv:1801.06331, arXiv:1401.5745; Azais and Leon, Electron. J. Probab. 18 (2013), no. 68; arXiv:1504.05355, arXiv:2111.09015, arXiv:1707.09276, arXiv:1005.4113. These results rely on a rich range of tools, including Kac-Rice formulas, moment methods, and Wiener chaos decompositions. In the non-Gaussian setting, however, many of these tools are unavailable. To the best of our knowledge, prior central limit theorems beyond the Gaussian setting were limited to Kac-type polynomials, including hyperbolic polynomials; see the works of Maslova (1974), O. Nguyen and Vu (arXiv:1904.04347), and, more recently, Do, N. Nguyen, and O'Rourke (arXiv:2605.26402). In this paper, we prove a central limit theorem for the total number of real zeros of Weyl polynomials whose coefficients are iid copies of a symmetric, mean-zero, variance-one subgaussian random variable $\xi$. This substantially extends one of the main results of Do and Vu (arXiv:1707.09276) to a broad class of non-Gaussian distributions, including the Rademacher distribution. Without the symmetry assumption, we prove central limit theorems for the number of real zeros for positive bulk intervals, as well as for $[0,\infty)$. Our proof combines the uniform one-point anti-concentration estimates from our recent work (arXiv:2511.07735) with the localization of Weyl polynomials around the coefficient index $i\approx x^2$. While our proofs use comparison to compute the variances, the CLT deduction is rather direct.
Read the full abstract
Representation Theory
Derived Reconstruction of Semisimple Tensor and Module Categories
We study bounded derived categories of semisimple tensor categories and their semisimple module categories. These categories may have infinitely many simple isomorphism classes. We classify bounded monoidal $t$-structures by integer-valued characters of the universal grading group and an integer deviation. We also classify compatible module $t$-structures using stabilizer subgroups of the module components. The corresponding hearts are tensor and module equivalent to the original categories. It follows that a monoidal derived equivalence and a coherent derived module equivalence recover both the tensor category and its module category.
Read the full abstract
Superconductivity
Symmetry-selective field-induced triplet superconductivity in Ising-superconductor monolayer NbSe$_2$
We investigate superconductivity in monolayer NbSe$_2$, an Ising superconductor, under an in-plane Zeeman field $h_x$, focusing on the emergence of symmetry selected equal-spin triplet pairing. Using a self-consistent Bogoliubov–de Gennes approach with realistic hopping parameters for monolayer NbSe$_2$ on a triangular lattice, we determine the energetically favored singlet pairing states in a range of chemical potentials $\mu$ for onsite, nearest-neighbor, and next-nearest-neighbor pairing channels, and map the resulting phase diagram in $(h_x,\mu)$ plane. A momentum-resolved analysis reveals distinct dominant contributions in order parameters arise from the surroundings of $\Gamma$, $K$, and $K'$ points of the Brillouin zone. We find that an in-plane magnetic field helps to induce a triplet pairing whose symmetry is determined by the parent singlet state, upon opening the triplet interaction channel. For the nonlocal pairing channels, chiral-$d$ pairing is energetically favored and it induces chiral-$p$ equal-spin triplet component with opposite chirality. We further find that Rashba spin-orbit coupling, relevant to substrate coupling and electrostatic gating, suppresses the superconducting orders and reduces the critical field. Our results establish a direct link between the symmetry of the parent singlet condensate and the emergent triplet superconductivity, highlighting monolayer NbSe$_2$ as a promising platform for field-tunable mixed-parity superconducting states.
Read the full abstract
Strongly Correlated Electrons
Ginzburg-Landau Theory for Non-Invertible Symmetry-Breaking Transitions
Recently, a generalized Landau paradigm has been proposed, in which a broad class of "unconventional" phase transitions lying beyond the original symmetry-breaking framework of Landau are understood as the symmetry-breaking transitions of generalized symmetries. In particular, in (1+1)D, such "topological" or "deconfined" transitions can be mapped to the symmetry-breaking transitions of non-invertible symmetries. How to extract the universal dynamical properties of the transitions from this understanding? In this paper, we follow the Ginzburg-Landau philosophy and develop a field-theoretic description of such critical points in terms of the fluctuations of local order parameters, now of non-invertible symmetries. The Symmetry Topological Field Theory formalism plays a central role in our analysis, allowing us to identify the algebra of local order parameters from the Lagrangian algebra of a (2+1)D topological order. We illustrate the key ideas and the full steps of this generalized Ginzburg-Landau procedure using the transitions between gapped phases with a simple non-invertible symmetry — the Rep($S_3$) symmetry.
Read the full abstract
High Energy Physics (Phenomenology)
GUT-induced collider signatures of the LUX-ZEPLIN event
The recent observation of an event with large recoil energy at LUX-ZEPLIN (LZ) has found a possible explanation in terms of the inelastic scattering of a pseudo-Dirac particle, for example, the Higgsino of the MSSM. For pseudo-Dirac dark matter, the mass splitting $\delta m$ is correlated with the mass of some heavier states. While in the MSSM rough dimensional considerations seem to indicate mass ranges of these heavy particles hopelessly out of reach even at future colliders, we point out that this is not the case in GUT-induced scenarios, with or without supersymmetry. Working in a non-supersymmetric model based on SU(6), we show that the experimentally favored $\delta m$ implies the presence of a $Z'$ gauge boson with a mass of several TeVs. We derive the predicted $Z'$ signal in the dilepton channel and we confront it with the sensitivity of the HL-LHC and the FCC-$hh$ to $Z'$ resonances. We show that the $Z'$ falls squarely in reach of the discovery potential of both experiments.
Read the full abstract
Cryptography and Security
A Novel Semantic Manifold Alignment Attack against Embedding-to-Embedding Obfuscation in Privacy-Preserving LLMs
With the widespread applications of large language models (LLMs), privacy-preserving inference has become increasingly essential for sensitive queries. To balance privacy and utility, a series of lightweight obfuscation approaches has recently been proposed, where users locally transform plaintext embeddings into the fixed ciphertext ones. While such Embedding-to-Embedding Obfuscation (E2EO) schemes demonstrate considerable resilience against traditional token frequency and embedding inversion attacks, the core mechanism behind remains to be the large-scale one-to-one substitution, which provides no cryptographic guarantees. In this paper, we propose Proxy Manifold Alignment (PMA), a novel attack against E2EO in privacy-preserving LLMs. Our key observation is that E2EO schemes keep the original semantic structure, so that the obfuscated vector stream can be regarded as an unknown tokenizer-language whose symbols are the vectors themselves. Therefore, the proposed ciphertext to plaintext reconstruction attack can be formulated as a translation task from the unknown tokenizer-language to plaintext. Specifically, by only accessing the obfuscated vector stream, the target tokenizer and a public corpus, the PMA attack first employs Word2Vec to model the co-occurrence patterns within the obfuscated stream and the public corpus independently, and constructs two proxy vector embeddings. Then, the attack aligns the underlying manifolds of these two embeddings based on structural similarity. Finally, it maps the obfuscated vectors back to plaintext. Experimental results demonstrate that PMA consistently achieves higher plaintext recovery than other state-of-the-art attack methods.
Read the full abstract
Probability
Random walk and Weber-Schafheitlin integral: generalizations and discussion
This paper presents a study of the Weber-Schafheitlin integral and its connections to random walk theory. The Weber-Schafheitlin integral, defined as the improper integral of a product of Bessel functions, arises naturally in the evaluation of probability distributions for multidimensional random walks. This work bridges pure special function theory with applied probability.
Read the full abstract
Combinatorics
Punctured adjacency-degree algebras of Cartesian products
For a connected regular graph G and a vertex a, we study the algebra generated by the adjacency and degree matrices of G-a and its cyclic module P_a generated by the all-ones vector. Our main theorem determines dim P_a for Cartesian products whose factors have equitable distance partitions at the chosen roots. A normalized logarithmic derivative of the local spectral generating function partitions the factors into boundary classes. We identify the boundary-return space exactly and express dim P_a as a sum of affine ranks on additive spectral fibres. For a distance-regular factor with distinct spectrum \Theta, this gives dim P_a(F^{\square m}) = |m\Theta| - 1. For products of powers of two distinct complete graphs, we evaluate the fibre formula in closed form. We also determine the full punctured algebras of all Hamming graphs: equality with the compressed Terwilliger algebra holds precisely in dimensions at most four for the hypercube and at most two for larger alphabets. For distance-regular graphs, adjacency moments alone determine the intersection array, with an explicit finite reconstruction. Finally, Cartesian stabilizer formulas separate metric loss from orbit splitting; on Doob graphs their distance-graded defect recovers the number of Shrikhande factors.
Read the full abstract
Artificial Intelligence
Reason Through the Latent! Making Latent Visual Reasoning Necessary
Latent visual reasoning aims to perform multimodal reasoning through hidden-state computation rather than explicit textual chains of thought. However, visual information being present in a latent state does not imply that the model actually relies on that state when producing its answer, especially when alternative image-conditioned paths remain available. We introduce Causal Visual Recurrent Reasoning (CVRR), which preserves pretrained visual competence while making recurrent computation the required image-conditioned path to prediction. CVRR initializes recurrence from the question hidden state after the pretrained vision-language model has incorporated the image, then repeatedly updates this state while re-reading the same fixed visual evidence. Before decoding, visual states and the original multimodal KV cache are removed so that only the final recurrent state carries image-conditioned information to the answer. Across the $V^*$, MMVP, BLINK, and MME-RealWorld-Lite benchmarks, CVRR retains strong performance under this strict interface, while compatible latent reasoners fail to recover comparable visual competence even when retrained under the same constraint. Causal interventions further show that predictions remain sensitive to recurrent content when the question is held fixed, and that persistent visual evidence causally revises the recurrent trajectory. These results distinguish latent informativeness from latent computation that is actually used for prediction.
Read the full abstract
Earth and Planetary Astrophysics
Calibration of CNEOS Fireball Velocities and the Robustness of Nominally Hyperbolic Events
The NASA/JPL CNEOS Fireball Data catalogue provides a near-global record of bright atmospheric entries but does not report per-event velocity covariance. We calibrate its kinematic accuracy using 19 events with independent reference trajectories and use optical, satellite, radar, and infrasound catalogues as source-specific controls. For nine modern high-quality comparisons, the CNEOS-minus-reference speed residual has mean -1.00 km s$^{-1}$ and standard deviation 0.86 km s$^{-1}$; variance decomposition assigns 0.77 km s$^{-1}$ to the CNEOS component. The radiant residuals have a 0.74 degree per-axis core plus one 16.3 degree outlier, so the extreme tail remains poorly determined. Seven of 354 complete CNEOS states are nominally hyperbolic without correction. A +0.9935 km s$^{-1}$ atmospheric-stage shift produces five additional crossings, but these are treated only as sensitivity cases. Polar-IM (2026 April 1) is the strongest post-2018 CNEOS anomaly, whereas independent GLM-LI stereo solutions yield 57.3 +/- 2.0 and 56.7 +/- 3.0 km s$^{-1}$, close to their direction-specific parabolic thresholds. We conclude that CNEOS can identify priority candidates, but secure interstellar classification requires an independent, uncertainty-bearing three-dimensional trajectory for the same event.