Complex Variables
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Monday 9 November
Let $S_{f,n}$ and $K_{f,n}$ denote Schröder's methods of the first and second kind with given order $ n $ ($\geq 2 $) applied to an entire function $ f $. This article investigates the existence of the entire solutions $f$ whenever $ S_{f,n} $ and $ K_{f,n} $ are rational. Utilizing the Nevanlinna theory, we obtain the general form of entire solutions for such equations. It is proved that either of $ S_{f,n} $ and $ K_{f,n} $ is a non-identity rational map if and only if $f = p e^{q}$, for some polynomials $p$ and $q$. Additionally, we provide one recurrence relation of $K_{f,n}$ depending only upon $K_{f,n-1}$.
Computation and Language
Fairness in Explainable AI Narratives: Decomposing Social Disparities in LLM-Generated Explanations
Large language models (LLMs) are increasingly used to translate technical explainable AI (XAI) artefacts such as SHAP into explanatory narratives (XAINs) for lay users. In high-risk domains, these explanations are often the main resource through which individuals can understand and contest adverse decisions. Yet LLMs are known to reproduce social bias, even from proxy variables alone, and fairness in XAINs remains unstudied. Because XAINs are bound to a ground-truth artefact, fairness concerns not only what a narrative says and how, but also whether it is equally faithful across social groups. We introduce fairness in XAINs as a new field of research and propose a framework that generates narratives under three conditions (protected attribute disclosed, withheld, or counterfactually swapped), allowing five comparisons that pinpoint the cause of disparities. We apply it to adversely classified instances from four datasets in credit scoring, education and employment, generating narratives with five LLMs, and evaluate faithfulness, feature coverage and sentiment. We find significant disparities across all datasets, metric families and models, arising in two ways. Faithfulness disparities result from feature values: LLMs make plausible but false statements, triggered by values that are unequally distributed across social groups. Feature coverage and sentiment disparities stem from the protected attribute itself: once it is disclosed, LLMs give less explanation to, or frame more negatively the narratives of, groups that have historically been discriminated against or are stereotypically associated with worse outcomes. A more faithful model removes the faithfulness disparities, and withholding protected attributes mostly removes the others, but neither suffices on its own. Fairness in XAINs should therefore be audited separately from the predictive model and its XAI artefact.
Read the full abstract
Machine Learning
TRACE: Adaptive Trust-Region Bayesian Optimization for Open-Ended Molecular Design
Optimizing molecules under small evaluation budgets requires deciding both where in chemical space to search and which candidate within that region to evaluate. We introduce Trust Region Adaptive Chemical Exploration (TRACE), a framework for open-ended molecular Bayesian optimization (BO), where candidates are generated on demand rather than selected from a fixed library. TRACE combines an objective-free graph-perturbation mechanism with a fixed, chemistry-aware Gaussian process (GP) surrogate using fragprints, a representation combining molecular fingerprints and fragment descriptors. Rather than learning the search geometry itself, TRACE makes the scale of chemical exploration explicit through a Tanimoto-similarity threshold controlling chemical extent and a cardinality budget controlling candidate-search breadth. A simple success/failure controller expands or contracts both quantities during optimization, while the GP and acquisition function select which admissible molecules to evaluate. Across six challenging molecular optimization benchmarks under matched 1,000-evaluation budgets, TRACE achieves the strongest overall sample efficiency as measured by the best molecule found as evaluations accumulate. Crossed ablations on two objectives further show that adaptive region control contributes beyond the proposal mechanism, while its benefit depends on the proposal distribution. We further apply TRACE to the stability of iron phthalocyanine (FePc), a non-precious molecular electrocatalyst whose degradation can involve loss of the Fe center. Using a first-principles density functional theory (DFT) workflow designed for accurate energy evaluation, TRACE increases the computed thermodynamic resistance to Fe dissolution from 0.42 to 2.59 eV over 80 designed evaluations.
Read the full abstract
Commutative Algebra
Rigidity does not imply substitution rigidity
We obtain a negative answer to a question of Freudenburg concerning rigidity conditions for a polynomial. We construct a polynomial that does not lie in the kernel of any nonzero locally nilpotent derivation on the polynomial ring, but there exist an affine domain, algebraically independent elements and a locally nilpotent derivation such that the result of substituting elements into the polynomial lies in the kernel of that locally nilpotent derivation, but at least one of the substituted elements does not lie in that kernel.
Read the full abstract
Quantum Physics
Long-Range Ergotropy Transfer with Weakened Boundary-Coupling Spin Chains
Ergotropy is a key quantity of a quantum battery that measures the maximum extractable energy. When the extraction process is not local, a faithful transfer of ergotropy across distance is necessary. In this work, we investigate such transfer by using spin chains with weakened coupling at the boundaries as mediator. Weakening the coupling at the two boundaries of the chain gives rise to localized edge states allowing high fidelity transfer within given time duration. We found that, for long spin chains, the transferred ergotropy alternates between even and odd number of spins. We explain this parity effect by investigating the chain spectral structure. Furthermore, we consider both the nearest-neighbor and long-range interactions between spin. In both model of interactions, weakening the boundary couplings improves the maximum transferred ergotropy for long spin chain, compared with in the case of conventional uniform chain. Although the long-range interaction reduces the transferred ergotropy, the transfer time is reduced, too, in particular for longer chain. In addition to the total transferred ergotropy, we also investigate the effect of initial states of the sender spin to the transport of the coherent and incoherent components of the total ergotropy, and the average transport efficiency.
Read the full abstract
Mesoscale and Nanoscale Physics
Spin-valve effect through electrostatic coupling
We propose a spin-valve effect in which an insulating magnetic van der Waals multilayer controls the conductance of an adjacent non-magnetic conductor by electrostatic coupling, without transport-electrons entering the magnetic layers. The coupling is induced by the variations of the electronic structure when the relative orientation of the magnetization of adjacent monolayers changes from parallel to antiparallel, modifying both the dielectric constant and band alignment. We consider two types of coupling and device concepts: charge transfer to a graphene monolayer and capacitive coupling in a single-electron transistor. We use density functional calculations to assess the magnitude of the effect in various materials, including CrI$_3$, CrBr$_3$, CrCl$_3$ and NiBr$_2$. We predict magnetoresistance ratios in the range of 1.14-1.20 for the charge-transfer device and up to $10^3$ in the single-electron transistor.
Read the full abstract
Computation and Language
AgentOPSD: Turn-Level Credit Assignment via Recursive Belief Revision
Reinforcement learning with verifiable rewards ordinarily spreads a single advantage over a whole trajectory, a granularity too coarse to isolate the few decisions on which success in a long, multi-turn interaction actually turns. Privileged self-distillation supplies denser supervision, but it leaves unanswered how a set of local teacher-student discrepancies should be converted into credit along a sequence. We introduce AgentOPSD, a critic-free scheme that assigns credit recursively, one interaction turn at a time. Inside a turn, the token-level log-probability gaps between a skill-conditioned self-teacher and the student are summed, and that sum acts as evidence in a recursive Bayesian update carried out in log-odds space; the resulting revisions reallocate the outcome-derived advantage across turns through a multiplier that is bounded and preserves its sign. With Qwen2.5-3B and Qwen2.5-7B as policy backbones on ALFWorld, the method improves consistently on GRPO and on self-distillation baselines, and attains 89.1% success at the 7B scale. Ablations attribute the gain to aggregating evidence at turn boundaries and to making the belief revision depend on history, rather than to the local gap by itself.
Read the full abstract
Atomic Physics
Effects of pulse durations in atom interferometry with double Raman diffraction
Double Raman diffraction is a key enabling technique for future space missions relying on atom interferometry that offers enhanced sensitivity and intrinsic suppression of systematic effects. This diffraction mechanism naturally arises when employing a retro-reflection configuration in microgravity, but then accelerations can no longer be determined through a frequency-chirp scan. As a consequence, measurements become much more sensitive to systematic effects associated with pulse timing and duration. Indeed, the finite pulse duration can introduce additional phase shifts and small modifications of the arm trajectories, which are critical sources of error in high-precision measurements. In this work, we present the first complete analytical results of these effects for double Raman diffraction, building upon previous work for single diffraction and extending the standard literature. Our analysis determines the impact of finite pulse durations on the full interferometer sequence, quantifying the shifts of the atomic wave packets' trajectories and the acquired phases. The description relies on a comprehensive framework describing matter-wave propagation in terms of central trajectories and centered wave packets, and can be easily incorporated into relativistic treatments. Furthermore, we make use of group theoretic methods that are particularly useful when accounting for residual accelerations and frequency chirps, which are also relevant for ground-based applications. Our study includes time-dependent accelerations and frequency chirps, gravity gradients, retardation effects, AC Stark shifts induced by the Raman pulses, and external forces such those due to magnetic fields or temperature gradients in the black-body radiation.
Read the full abstract
Numerical Analysis
A Large Gap between Galerkin Errors with and without Mesh-Uniform Regularity for the KdV Equation with Boundary Dissipation
We compare two cubic Hermite Galerkin semidiscretizations of the Korteweg-de Vries equation on a bounded interval with boundary dissipation. They use the same trial space. The first is the weighted scheme considered in our previous paper (arXiv:2610.39269), called the W-scheme. It has mesh-uniform regularity: the discrete semigroups of the linearized equation inherit Crandall-Pazy regularity and Kato smoothing, with bounds uniform in the mesh size $h$. The second is the standard Galerkin method in $L^2$, called the S-scheme. It has no mesh-uniform regularity. For the W-scheme, the full error and the ordering error converge at an algebraic rate in $h$, uniformly on $[t_0,\infty)$ times small balls of $L^2$ for any fixed $t_0>0$. They are bounded by $Ch^p\varepsilon$ and $Ch^p\varepsilon^2$, respectively, where $\varepsilon$ is the size of the datum and $p>0$. We prove that neither bound holds for the S-scheme by using exact high-frequency spurious eigenpairs of the S-scheme. The existence of such eigenpairs for meshes with at least 512 elements and the failure of both bounds have proofs checked by Lean 4. The only premise is the mesh-uniform Crandall-Pazy estimate of the W-scheme (arXiv:2610.39269). Computations compare the two schemes on unit-norm data built from these eigenfunctions, on 512, 1024 and 2048 elements. The full error of the S-scheme stays at $4.6\times10^{-2}$. That of the W-scheme is at most $2.2\times10^{-9}$. The ordering error of the S-scheme stays at $9.6\times10^{-6}$. That of the W-scheme falls from $2.3\times10^{-9}$ to $1.4\times10^{-10}$. Their ratio grows from $4.2\times10^3$ to $6.6\times10^4$.
Read the full abstract
Medical Physics
Vendor-Agnostic Pseudocontinuous Arterial Spin Labeling Using Pulseq: Cross-Vendor Quantitative Evaluation
Purpose: Inter-scanner variability in cerebral blood flow (CBF) quantification limits reproducibility in multi-center arterial spin labeling (ASL) studies. We developed and validated a vendor-agnostic pseudo-continuous ASL (PCASL) sequence using the open-source Pulseq framework, and assessed its extensibility to advanced perfusion imaging. Methods: A vendor-independent PCASL sequence was implemented in Pulseq and deployed on 3T scanners from Siemens, GE, and United Imaging. Seven healthy volunteers underwent single-post-labeling-delay (PLD) PCASL on all platforms, including test-retest scans. An in-house adapted Siemens vendor-based PCASL sequence served as reference. Reproducibility was assessed using within-subject coefficient of variation (CVwithin), root mean square error (RMSE), regional CBF correlation, and voxel-wise variation ratio maps. Multi-PLD PCASL and planning-free VE-PCASL were implemented as proof-of-concept extensions. Results: Mean gray matter (GM) CBF was similar across implementations. Test-retest GM CVwithin was below 4% on all platforms; Siemens-Pulseq had the lowest value (2.65%), versus 3.64% for the native sequence. GM RMSE between Siemens-Vendor and Pulseq implementations was comparable to test-retest RMSE, and all inter-scanner CVwithin values were below 3%. Voxel-wise variation ratio distributions were centered near 100%, suggesting cross-vendor variability comparable to physiological baseline fluctuations. Multi-PLD and VE-PCASL ran successfully on all platforms, yielding consistent CBF/ATT maps and vascular territory delineations. Conclusion: Pulseq-based PCASL enables vendor-agnostic perfusion imaging with low inter-scanner variability. This open-source approach provides a reproducible, standardized solution for multi-center ASL studies and facilitates deployment of advanced ASL techniques in clinical research.
Read the full abstract
Optimization and Control
Interface-Preserving ${H}_\infty$ Model Reduction of Network Systems
We study projection-based reduced-order models that exactly retain specified input and output interfaces. Admissible projections are characterized geometrically: the retained subspace must contain every input and output direction, which fixes the minimum order. The reduction error factors into input-coupling, internal-dynamics, and output-coupling systems, giving a submultiplicative ${H}_\infty$ bound; for positively dominated factors the ${H}_\infty$ error equals a dc gain computable from linear solves. Strict dissipativity certifies the stability of every orthogonal reduction, and a Hankel singular value bounds the optimum a posteriori. These results motivate sequential graph-coarsening heuristics for networks. On a 65,566-state chip thermal network, interface-preserving coarsening yields sparse, positive lumped models that are more accurate than Kron reduction of the same interface at every order, with far fewer edges; balanced truncation is more accurate but destroys the network structure.
Read the full abstract
Combinatorics
Exact codegree Turán densities of even zycles
We determine the exact codegree Turán density of the $3$-uniform zycle $Z_\ell$ for every even integer $\ell\ge 8$ with $3\nmid\ell$,showing that $\gamma(Z_\ell)=1/3$. More generally, we prove that $\gamma(Z_\ell)\le 1/3$ for all even $\ell\ge 8$. Our results partially answer a question of Piga and Schülke [Bull. Lond. Math. Soc. 58 (2026), e70348] concerning the sharpness of their upper bound of $1/2$.
Read the full abstract
Earth and Planetary Astrophysics
The compositional link between rocky exoplanets and their host stars is consistent with Earth-like devolatilization
Previous work found a higher bulk iron-to-silicon (Fe/Si) ratio in rocky exoplanets relative to their host star, traditionally explained by iron enrichment. An alternative explanation is devolatilization of the planetary abundance ratios with respect to the host star, where volatile elements such as O and S are progressively depleted compared to refractory elements such as Fe and Mg, as in Earth. We explore whether these elemental molar ratios are consistent with a devolatilization trend. We revisit the core mass fraction of 14 rocky exoplanets with masses between 0.5 and 10 Earth masses whose host stars have spectroscopic abundances of Fe, Si, Mg, O, and S in the literature. We use interior structure retrievals with two independent forward models, the Marseille Super-Earth Interior model and pyExoInt, to estimate compositional parameters, including elemental molar ratios, from mass and radius measurements. We find that 8 out of 14 exoplanets in our sample present an Earth-like devolatilization trend, while only one planet shows a 1:1 relationship between planetary and stellar abundances. This suggests that devolatilization offers a plausible explanation for the observed non-1:1 relation. The two interior models agree on the devolatilization trend, demonstrating that our derived trends are robust. The main limiting factor in constraining refractory compositions of rocky planets relative to their host star is the small sample with constrained stellar abundances and the lack of variety in orbital distances. The PLATO mission will help by extending the sample of characterized rocky planets to orbital periods of up to one year. The variety in refractory elemental ratio trends may indicate a wide diversity in bulk composition, outgassing and other mantle properties relevant for habitability.
Read the full abstract
High Energy Astrophysical Phenomena
A Many-Shapes framework for Galactic Cosmic Rays
Thanks to recent measurements by space and ground-level detectors, the energy spectra of cosmic ray protons and helium nuclei are now known with good precision over a broad energy range extending for more than six orders of magnitude, from a few GeV to approximately 10PeV. These measurements reveal that both spectra do not have a simple power–law form, and exhibit several features. The proton to helium ratio exhibits a very unexpected behavior: it decreases for rigidities $\lesssim 10^5$~GV, and increases for higher rigidities. Several authors have interpreted the observed spectral features as revealing the fact the CR spectra are formed by the contributions of a few components (in some cases as few as two) produced by sources that belong to distinct classes of astrophysical accelerators, with sources in the same class producing spectra with the same characteristic shape. Most of these authors also consider it necessary to include the contribution of one (or perhaps more than one) “local” sources, recently active in the vicinity of the solar system. However, this approach has several limitations, and in this work we discuss an alternative scenario in which the cosmic ray sources, even belonging to the same class, produce spectra with a broad range of shapes. The power–law form of the total source spectrum emerges as the average of these contributions. In this framework observable spectral features can be produced by statistical fluctuations, and different spectral slopes for protons and helium nuclei can be observed even if one assumes that acceleration is rigidity-dependent.
Read the full abstract
Systems and Control
Covert System Identification Under Timing Constraints: A Static Linear Model and Its Detectability Implications
This paper studies covert system identification in cyber-physical systems, where an attacker infers plant dynamics while attempting to remain stealthy. We introduce a lightweight static linear model in which the time to execute a setpoint change is proportional to the commanded speed difference. On a Siemens PROFINET testbed this gain captures the drive's ramp response accurately but not its terminal overshoot. That omitted transient becomes the attacker's main exposure: on real traces, the residual left by the linear masking procedure is detected by two anomaly monitors, Process-Aware Stealthy-Attack Detection (PASAD) and the CUSUM change detector. Across the cascade identification and staircase attack strategies defined in the paper, the stealth boundary relates to the plant's own settling transients, a structural limit for this class of models under the adopted detector-calibration regime. Comparison against dynamic models obtained by N4SID, trained at multiple excitation amplitudes, shows that the evaluated N4SID models do not improve stealth over the structure-informed ramp model. Since per-cycle computation is negligible for both procedures, the binding trade-off is not inference cost but whether the attacker can acquire and refresh a model without injecting detectable perturbations.
Read the full abstract
Soft Condensed Matter
Nonlinear integral equations for stress-controlled rheology: Collocation methods for Volterra equations with functional kernels
We present a class of numerical collocation methods to solve a typical inverse problem that arises in the stress-controlled rheology of complex fluids with pronounced history-dependent response, where the relation between stress and flow rate is given by an integral constitutive equation as the material law. We demonstrate the method by calculating the creep response of a nontrivial model of a shear-thinning viscoelastic fluid where a stress overshoot in rate-controlled startup flow leads to “sudden yielding” in stress-controlled creep. Our method is able to address the long-time behavior of the model, distinguishing the amorphous solid from the stress-induced yielded fluid, and from the in-between case of infinite deformation that never achieves a stationary shear rate (proper creep). To this end, we implement a method with flexible time-stepping, and prove uniqueness of the solution and convergence of the methods for a broad class of highly viscoelastic non-linear rheological laws in integral form.
Read the full abstract
Optimization and Control
Monetisation and Cost Function Scalarisation Cannot Guarantee Valid Multi-Criteria Decision-Making
Monetary quantities are not preferences. Cost is an object-related system performance; value-for-money is an actor's subjective evaluation of it. Monetisation embeds trade-off information: its conversion coefficients express valuations and determine how heterogeneous performances compensate one another. Cost-benefit analysis, even net present value, yields quantities of money, not the preference value actors attribute to them. This creates a monetisation paradox. Cost enters a multi-criteria problem as one performance among several, but money is then elevated into the universal representation of all others, creating representational dominance. The problem collapses into a single-criterion one in which every performance acts only through its imposed monetary equivalent. Cost-function scalarisation does not resolve this by substituting points, normalised scores, losses, or penalties: these transformations and their weights fix preference directions, intensities, and trade-offs, so they are implicit preference models, yet are treated as technical constructs rather than measured and validated. A well-defined scalar optimum is therefore not necessarily a decision-valid multi-criteria outcome. Numerical commensurability does not establish preference commensurability. Decision validity requires heterogeneous performances to be transformed into actor preferences in a unified preference domain. Preference is relational and contextual, yet can be represented on an affine scale. Preference Function Modelling and the Linear Preference Space provide the commensurability required for valid aggregation. The decision criterion should thus maximise aggregated actor preference over the feasible and acceptable design-decision space, not monetised quantities or scalarised scores.
Read the full abstract
Quantum Physics
All-to-all connectivity via adiabatic ZZ interactions in a superconducting flux qubit processor
Scalable superconducting quantum processors would benefit from high qubit connectivity, yet existing architectures connect neighbouring qubits via tunable couplers, which introduce substantial wiring and control overhead. Here, we propose a baseband-driven adiabatic ZZ gate that entangles any selected pair in a register of flux qubits coupled via a common cavity. The qubits stay dc-biased at their flux sweet spot. Strong drives applied selectively through local flux lines, at frequencies far below the qubit transitions, adiabatically tilt the quantization axes and generate an effective longitudinal ZZ interaction. The strength of the coupling is tunable and controlled by the amplitudes and relative phase of the drives rather than by any resonance condition. Numerical simulations give coherent gate errors below $10^{-4}$ over broad ranges of drive frequency, with comparable fidelities for two simultaneous gates in a four-qubit register. The protocol is robust against frequency fluctuations and remains effective even when many qubits are coupled to the cavity.
Read the full abstract
Machine Learning
From Likelihood Convergence to Parameter Convergence in POMDPs
Likelihood-based methods are widely used to learn the parameters of partially observable Markov decision processes (POMDPs). While finite-sample guarantees for policy suboptimality exist for certain POMDP subclasses, analogous guarantees for parameter recovery remain less well understood. For any candidate POMDP parameter, we bound its parameter recovery error in terms of its likelihood gap to the ground truth, the structural conditioning of the observation and transition kernels, and the behavior policy used to collect data. The bound applies in both the undercomplete and overcomplete regimes, with the latter following naturally from the undercomplete analysis. Because the result applies to any sufficiently good candidate, it yields finite-sample guarantees for the maximum-likelihood estimator and any procedure producing a parameter with controlled likelihood gap. This is particularly useful for near-optimal learning in infinite-horizon POMDPs, where existing approaches largely rely on model-based planning with accurate transition and observation kernel estimates. The bound separates the contributions of the behavior policy and model, showing how data-collection choices, including action coverage, and their effects on mixing can accelerate parameter learning.
Read the full abstract
Computer Vision
Evaluating Cross-Signal Coherence in Mammography Vision-Language Models
As vision-language models (VLMs) move toward interpreting medical images directly, their outputs increasingly include not just a label but a stated confidence and a written justification. Yet the field still predominantly evaluates VLMs using just accuracy. We evaluate the coherence between answer, confidence, and explanation across four VLMs on breast density and BI-RADS categorization, using two mammography datasets. We perform controlled perturbations of the question, multiple choice option orders, and the images and assess changes in the three outputs. Our results suggest confidence does not decrease even as the mammography image is changed. Shuffling the options also changes most VLMs' answers. Notably, there are cases where a model's own stated reasoning contradicts its final answer, a distinction accuracy alone cannot capture but one that governs whether an AI-generated answer can be trusted.
Read the full abstract
Chemical Physics
Ab Initio Molecular Dynamics-Guided Machine-Learning Potentials for Mapping Nitric Oxide Interactions with Oxygen-Covered Graphite
Exploratory ab initio molecular dynamics is used to construct machine-learning interatomic potentials for NO interactions with highly oriented pyrolytic graphite at 33% bridge-site oxygen coverage. Representative configurations sampled across surface temperatures and incidence energies provide reference energies and forces for training and validation. The selected potential enables relaxation of 300 adsorption configurations and construction of spatial and angular interaction maps. These maps reveal increasingly restricted attractive regions during vertical approach and reduced short-range repulsion upon molecular reorientation. The resulting adsorption geometries and interaction landscapes guide further first-principles sampling toward refining potentials for NO oxidation dynamics.
Read the full abstract
Cryptography and Security
Text2NGAC: a multi-LLM pipeline for NGAC policy extraction with mutation-based verification
Access-control policies are written in natural language but enforced as formal models, and translating them into NIST Next Generation Access Control (NGAC) graphs remains unsolved: prior systems cover only subsets of the NGAC model, and none verify the graphs they extract. We present Text2NGAC, a pipeline of cooperating large language models that disambiguates each sentence before extraction, escalates uncertain sentences to a multi-model ensemble, and checks the resulting graph with a five-strategy verifier whose own effectiveness is measured by controlled mutation testing. To our knowledge it is the first system to extract all four NGAC relationship types - associations, prohibitions, event-response obligations, and assignments - from real-world policy documents. Three results anchor the evaluation. The verifier detects 99.7% of 918 injected mutants and every structural fail-open fault, and the fault classes it misses are characterized and weighted by security consequence rather than left unexamined. The extracted graphs are enforceable: one corpus's graph loads unmodified into the NIST reference implementation, which reproduces all 55 access decisions the graph entails and fires an extracted obligation end-to-end. Extraction reaches 97.3-99.6% exact-tuple-match accuracy on the 714 scored sentences of four real corpora, far above three reproduced baselines on matched metrics, and paraphrase probes with semantic re-scoring bound the pipeline's sensitivity to rewording. Reported accuracy is wording- and benchmark-specific, adversarial input is outside the verifier's guarantees, and the intended deployment is extraction plus structural verification feeding expert review. A complete two-tier reproducibility artifact accompanies the paper.
Read the full abstract
Digital Libraries
More Papers, Not More Output: Collaboration and Publication Trends among Early-Career Researchers
Concerns about publication pressure often assume that early-career researchers now produce more papers than comparable researchers did in the past. However, publication counts may give a misleading impression of individual publication output because research has become increasingly collaborative. Researchers can list more papers on their CVs not only by producing more work themselves, but also by working with more coauthors and sharing credit for the same papers. Whether the collaboration-adjusted output of early-career researchers has increased over time has rarely been examined empirically. This study addresses this gap by analysing publication records from more than 200,000 early-career researchers at Western research institutions across five major science, engineering, and health fields between 1996 and 2024. Recent cohorts appear on more papers than earlier cohorts, but they also work in much larger research teams. Once publication credit is divided among coauthors, collaboration-adjusted output remains stable or declines across all fields. These findings show that research evaluation metrics that ignore collaboration are increasingly poorly aligned with how scientific work is produced. Rising publication counts alone should therefore not be taken as evidence that early-career researchers are producing more individual publication output.
Read the full abstract
Hardware Architecture
Hidden in Plain Sight: Privacy-Enhanced Computing Through Register Renaming
Privacy-Enhanced Computing (PEC) is a mechanism to hide program secrets from the outside world without paying the cost of fully-homomorphic encryption. Previous work has developed hardware mechanisms to achieve this goal but they require a dedicated processor enclave where all computations are done, requiring duplicate ALU resources and limiting the performance of PEC-utilizing applications due to not being able to access the full compute resources of the processor (e.g., multiple execution units and vector units), or not being able to execute PEC instructions in parallel and out-of-order. In this work, we introduce HPS-PEC (Hidden in Plain Sight PEC), a novel method of utilizing existing infrastructure in the CPU to achieve PEC with minor modifications to the architecture. We show that, across 25 VIP-Bench benchmarks and with a 7-cycle QARMA-64 cipher, HPS-PEC runs encrypted code at 3.41x the execution time of unprotected native code (geometric mean), of which 1.03x is the cost of data-oblivious code itself, and is 1.26x faster than a Sequestered Encryption (SE) unit while utilizing under a third of its dedicated area. Our method keeps plaintext out of reach of every instruction, as an enclave does given a correctly functioning renamer and register file, which every program already depends on.
Read the full abstract
Strongly Correlated Electrons
Contrasting one- and two-particle excitations in cuprate strange metals
In unconventional superconductivity, the pairing mechanism remains one of the central unresolved problems in solid-state physics. Moreover, the normal state shows strong deviations from conventional Fermi-liquid behavior, most notably a scattering rate that is linear in energy and temperature. Here we study the momentum dependence of the energy and damping of collective excitations in optimally doped cuprates in the normal state. In particular, we analyze existing Electron Energy Loss Spectroscopy (EELS) data of Bi2Sr2CaCu2O8 [Phys. Rev. B 44, 7155 (1991)] and new analogous optical plasmon data of Bi2Sr2Ca2Cu3O10. Furthermore, we reanalyze similar Resonant Inelastic Xray Scattering (RIXS) acoustic plasmon data [Phys. Rev. R 6, 043184 (2024)] of La1.84Sr0.16CuO4 and La1.84Ce0.16CuO4. We contrast these two-particle excitation results with one-particle excitations measured by Angle-Resolved Photoemission Spectroscopy (ARPES). Most of the plasmon behavior can be described well by combined self-energies of a hole and an electron. The dispersion of optical plasmons at high energy is only weakly determined by correlation effects, indicating resilient quasiparticles. The damping of the optical plasmons at low energy, however, is strongly influenced by the reduced scattering rates described in the marginal Fermi liquid model for "strange-metal" behavior. In addition, the same model describes the linear-in-energy damping of acoustic plasmons. Contrary to some theoretical predictions, we see no overdamped optical plasmons below the classical Lindhard critical wave vector.
Read the full abstract
Robotics
Modeling and Estimation of a Towed Marine System
This paper addresses the modeling and state estimation of a towed marine system consisting of two autonomous surface vehicles (ASVs), a pair of umbilical cables, towfish, and a trailing loop cable. Inspired by the existing literature, we employ a lumped-mass modeling approach for the cables and towfish, capturing the essential dynamics while maintaining computational tractability. A Lyapunov-based analysis is conducted to verify the stability and convergence properties of the proposed model. Building on this model, we propose two state estimation methods, namely a reduced-order observer and a perfect-measurement extended Kalman filter (PMEKF), using only the measurements from the ASVs' onboard sensors. Numerical simulations and experimental results are provided to validate the effectiveness of the proposed methods.
Read the full abstract
Machine Learning
LawKAN: Group-Law Layers for Composition – When Imposing a Known Law Pays Off
Many tasks compose transformations by a known rule: chained rotations in robotics and navigation. Learned models relearn it, compounding error; classical estimators hardcode it and miss hardware deviations. We impose the known structure and learn only the remainder: an exact non-commutative group law sits in the node of a self-contained layer, its edges learn local increments, and a zero-initialized correction psi learns unknown deviations of the law. A pre-training criterion predicts exact representability. On fixed-angle SU(2) composition, a 54-parameter dense edge is the most accurate dense model, at 27x fewer parameters than the strongest fixed-matrix baseline, and lies 0.05 deg above the Bayes floor. On real robot arms, an isolated 306-parameter psi reaches 1.12-1.14 mm versus 1.29-1.33 mm for Gauss-Newton at N >= 60, and the recipe transfers across three vendors and two protocols. The recipe matches classical fits where the data obey the law and loses where the imposed law is not the task's.
Read the full abstract
Materials Science
Quantum benchmarks for magnetic energies and transition temperatures
Predicting magnetic ordering temperatures requires accurate exchange interactions and an appropriate treatment of spin thermodynamics. Here we assess these requirements through complementary quantum Monte Carlo benchmarks. For five insulating antiferromagnets, we compare ferromagnetic-antiferromagnetic energy differences from diffusion Monte Carlo (DMC) and density functional theory with references reconstructed from inelastic neutron scattering (INS) exchanges. DMC underestimates these references but achieves a mean absolute error comparable to the best-performing functionals tested. Separately, we directly test the previously proposed (S+1)/S rescaling of classical Monte Carlo temperatures against quantum stochastic series expansion simulations for 17 compounds using corresponding INS-derived Heisenberg Hamiltonians. Rescaling reduces the mean absolute classical-quantum temperature difference from 81.7 to 8.2 K, with a mean absolute percentage difference of 5.9 %. These results establish a direct quantum benchmark for the rescaling within the tested models and emphasize that improvements in electronic magnetic energy scales should be accompanied by attention to quantum spin thermodynamics.
Read the full abstract
General Economics
That Time of the Year Again: Government Pay Revisions in the Aggregate Economy
Public pay is set by law in nominal terms, so inflation erodes its real value between legal acts until the government revises it. I model each revision as the choice of a forward-looking government, shaped by two costs that push in opposite directions. A fixed cost of acting makes revisions large and rare, whereas a convex cost of financing the increase at once makes them small and frequent. Through the budget, each revision also moves debt and the risk of fiscal stress under which the next one is made. In the documented pay decisions of eleven countries, pay set at discretion erodes against prices and private earnings and is restored in discrete steps, every cut was decided during market stress or an adjustment program, and court rulings drove a wedge between the cuts legislated, the cuts paid and the cuts that lasted. Estimated on the legal acts, the model reproduces the documented paths of pay in most countries when data and simulations are measured the same way. In Portugal's estimated model, the savings that the Treasury requires once spreads rise produce the cuts, the debt overhang holds pay down after the crisis, and a court that strikes cuts after they take effect invites cuts it will strike. Review of cuts before they took effect would have lowered the social cost of Portugal's crisis relative to review after the fact, and partial indexation would have paid in Bangladesh unless its automatic increases were costly to finance.
Read the full abstract
Computer Vision
GHiML: VLM-Guided Hierarchical Multi-Label Learning with Applications to Benthic Imagery
Vision language models (VLMs) encode substantial expert knowledge within their weights, yet their direct application to specialised classification domains such as benthic imagery remains limited. Meanwhile, vision transformers trained on these tasks rely predominantly on low-level visual latents (colours, textures, and shapes) that lack the higher-level reasoning humans employ for identification. We propose Guided Hierarchical Multi-Label (GHiML) learning, a pipeline that extracts conceptual latent variables from a VLM conditioned on training data, generates sample-relevant captions grounded in those variables, and fuses the resulting textual embeddings with image embeddings via cross-attention and a per-node gated router. We also introduce the inverted HML score ($d_{\mathrm{inv}}$), a similarity score that evaluates hierarchical multi-label predictions through graph traversal distance and a cardinality penalty, providing an interpretable alternative to existing measures, which are insensitive to realistic degradations in sparse, deep hierarchies. Experiments on BenthicNet, FathomNet and iNaturalist show that the router improves macro $F_1$ over the vision-only baseline by up to 6.7 pp and $d_{\mathrm{inv}}$ by up to 21.1 pp on in-domain benthic data, with gains persisting under distribution shift. Ablations locate the contribution of the text itself: on the multi-label biotic hierarchies, removing the per-sample text signal degrades $d_{\mathrm{inv}}$ while leaving macro $F_1$ unchanged, consistent with text acting through prediction cardinality rather than node discrimination, whereas iNaturalist, whose annotations are single-label, loses 18.7% of macro $F_1$. Our results support the hypothesis that VLMs contain exploitable specialised knowledge that, when appropriately elicited, can augment classification in domains where direct visual learning is insufficient.