Peer Revenue Publish or profit

Papers

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

All of arXiv

Friday 4 September

Optimization and Control

Sparse plus low-rank matrix embedding with applications in cancer radiotherapy optimization

Decomposing a matrix into sparse and low-rank components is central to robust principal component analysis and has broad applications in machine learning, signal processing, and computer vision. Classical formulations seek to recover the underlying sparse and low-rank structure. We instead introduce sparse-plus-low-rank matrix embedding (SLME), whose goal is to construct a computationally efficient surrogate for a large dense matrix, without requiring its components to be interpretable. Given $A\in\mathbb{R}^{m\times n}$, SLME approximates $A \approx S+HW$, where $S$ is sparse, $H\in\mathbb{R}^{m\times r}$, $W\in\mathbb{R}^{r\times n}$, and $r\ll \min\{m,n\}$. The resulting matrix-vector product can be evaluated as $Sx+H(Wx)$ in $\mathrm{nnz}(S)+r(m+n)$ operations, rather than the $mn$ operations required by $Ax$. Our primary motivation arises from optimization problems in cancer radiotherapy treatment planning, where a large dense dose-influence matrix is a major computational bottleneck. We formulate SLME as a bi-objective nonconvex optimization problem that balances approximation error against the computational cost of the downstream tasks. We then develop R3-Trust, an efficient trust-region algorithm that approximates the Pareto frontier in a single parameter-free run. Each point on the resulting frontier provides a sparse-plus-low-rank representation with a different balance between accuracy and downstream computational cost. Experiments on clinical radiotherapy matrices show that decompositions obtained from existing recovery-oriented formulations can be suboptimal for the embedding objective. Conversely, experiments on synthetic instances demonstrate that SLME and R3-Trust can also be applied to sparse-plus-low-rank recovery, where they compare favorably with state-of-the-art recovery methods in both reconstruction accuracy and computational time.
Read the full abstract

Machine Learning

GraphNOSE: A Graph Transformer in Olfaction

Predicting olfactory qualities from molecular structure is an open problem in chemoinformatics. Although linear models can link molecular features to odor descriptors, they often fail when extrapolating to novel chemical scaffolds, extreme molecular weights, or complex odor mixtures. To address this, we introduce GraphNOSE, an open-source graph transformer framework that predicts multi-label odor descriptors from simplified molecular-input line-entry system (SMILES) strings for single molecules and binary mixtures. By integrating positional and structural encodings within a transformer-based graph architecture, GraphNOSE achieves strong performance with six times fewer parameters than standard graph neural network (GNN) baseline while consistently outperforming linear models, molecular language model embeddings, molecular fingerprints, and baseline GNNs by an average area under the ROC curve (AUROC) margin of 4.52% (p < 0.01). GraphNOSE achieves an AUROC of 84% on out-of-distribution compounds (OODs). This exceeds the current state-of-the-art GNN for OOD in olfaction (Open-POM: 81%, p < 0.001), and identifies conditions under which linear models empirically fail. Finally, we apply XAI (explainable AI) methods to identify which substructures and molecular features drive odor predictions, yielding insights consistent with chemical intuition and grounded in the model's learned representations. Together, these results establish GraphNOSE as a scalable and interpretable architecture for olfactory prediction that generalizes to structurally distinct compounds underrepresented in current perceptual databases.
Read the full abstract

All authors Mrityunjay Sharma, Sarabeshwar Balaji, Valentina Parma, Ritesh Kumar

Number Theory

On the Diophantine Equations Arising from Three-Term $b$-Concatenations in Pell-Type Sequences

We study the Diophantine problem of representing a Pell or Pell–Lucas number as the base-$b$ concatenation of three other Pell or Pell–Lucas numbers. Under the condition that the index of the middle block does not exceed that of the leading block, we obtain an explicit upper bound for the sequence indices in terms of $b$. For $2\leq b\leq15$, we determine exactly $51$ solutions up to the identification $Q_0=Q_1=2$, while there are $74$ solutions when distinct index choices are counted separately.
Read the full abstract

Software Engineering

Regret Dominates Surprise: Design-Time Requirements Engineering for Agentic-AI Safety

Requirements engineers for agentic-AI domains face challenges in evaluating, specifying, and operationalizing safe autonomy. Mainstream frameworks, such as Goal-Oriented Requirements Engineering (GORE), lack mechanisms to systematically address these challenges under epistemic uncertainty. We contribute an approach that builds on GORE to model and simulate safe autonomy in agentic-AI systems. We introduce a novel Regret-Dominance Mechanism (MS-RGR) to operationalize safe autonomy. MS-RGR uses two signals: epistemic surprise (novelty detection) and cognitive regret (evaluative risk) to address the trilemma problem: should the agent operate in routine autonomy, undergo reflective reasoning, or escalate to human? We instantiate MS-RGR in elderly care monitoring and autonomous driving. A 100-seed stochastic simulation shows MS-RGR reduces silent failures to near-zero and detects risk approximately 17.5 times faster than a sensor-only baseline, remaining formally traceable via LTL safety properties. A retrospective proxy instantiation applying the DRI gate post-hoc over execution traces from 208 AGENTHARM scenarios across seven LLMs shows the gate improves harmful-task refusal only for models with strong baseline safety (over 80% pre-gate refusal, e.g., 84.1% to 90.9%), indicating MS-RGR amplifies rather than substitutes for model-level safety training. We discuss threats to validity, positioning MS-RGR as initial feasibility evidence for design-time safety constraints in agentic-AI requirements engineering.
Read the full abstract

Soft Condensed Matter

Hidden amplitude and dynamical information in phase-separation spectra

Framewise variance normalization removes fluctuation amplitude from an observation, but not from the dynamics that generate later patterns. The current normalized spectrum therefore need not determine its own future. We study this effect in three two-dimensional phase-separation models that change transport, add active driving, or suppress large-scale separation. For paired initial fields, their normalized spectra evolve identically in the linear regime. The restoring model provides an exact example: damping does not directly change how the normalized spectrum evolves when both equations are applied to the same field. It nevertheless changes the hidden amplitude, which alters the later nonlinear shape. An exact comparison separates the immediate effect of changing the equation from the effect of reaching a different state through earlier evolution. It also reveals strong cancellation between these effects in the mobility and active models. As a practical test, spectral histories improve classification by about 4-6 percentage points at several later observation windows, although the gain depends on quench depth and spectral construction. Temporal spectra can therefore reveal dynamical consequences of information absent from every individual normalized frame.
Read the full abstract

Logic in Computer Science

Trace-Tree Magmas: Proof-Producing Infinite Countermodels and 28 New Order-Five Austin Classifications

Finite model finders cannot witness an Austin law: an identity whose finite models are all trivial but which has a nontrivial infinite model. We introduce rank-decreasing sparse trace-tree magmas, finitely presented total operations on a countably infinite constructor-tree carrier. The default product pairs its arguments; finitely many positive Horn clauses define exceptions. Our main procedure derives clauses from symbolic evaluation traces. For every model found, it proves functionality of the exceptional relation by descent on constructor size, proves the identity by exhaustive symbolic case analysis, and emits a self-contained Lean 4 certificate. A least simultaneous fixed point gives an implementation-independent semantics, so bounded search may miss models but cannot invalidate certified results. On ETP's 96 order-five Austin candidates, we discover and Lean-verify infinite countermodels for 28 identities with no prior public classification in our audit. They form 14 duality classes and establish 28 new Austin classifications. Four ALPS-known cases bring the total to 32 certified candidates. On Canonical-4187, the deduplicated union of Order5-130 and the 4,141-row ALPS pool, a fresh trace run produces 636 certificates, all accepted by Judge v3. At equal resource limits, Vampire 5.0.1, E 3.5.1, and complete Twee 2.6.1 jointly prove implications in 94 canonical classes. Only Twee returns trusted counter-satisfiable outcomes, for 18 classes; independent finite-side certificates force 16 to be infinite. None of these ATPs emits an explicit model or Lean certificate, and none decides the 28 new classifications. To the best of our audit, this is the first automated system to synthesize this trace-tree model family, generate well-founded inversion proofs, and emit self-contained Lean 4 certificates.
Read the full abstract

All authors Jiaming Zhao, Bing Wu, Tong Yang, Xu Miao

High Energy Physics (Phenomenology)

Agnostic search for a vector-like top partner at a lepton collider via the recoil mass technique

A study of the single production of a vector-like top partner $T$ at a $\sqrt{s}=3$ TeV $e^{+}e^{-}$ collider is presented, using a decay-mode and mass-agnostic event selection built around the recoil mass observable. The analysis relies on fully simulated samples processed through a common reconstruction and event selection chain. The baseline selection is validated without beam polarization or initial-state radiation (ISR). The impact of more realistic running conditions, including ISR together with an $e^{-}$ beam polarization of $-80\%$, is then evaluated and the discovery reach is mapped as a function of the top partner mass $m_{T}$ and coupling strength $\kappa_{T}$. In the baseline scan, the estimated statistical significance rises above $3\sigma$ for the lowest benchmark masses once $\kappa_{T}$ exceeds approximately 0.3.
Read the full abstract

Machine Learning

Connecting Score Matching, Maximum Likelihood, and Expectation-Maximization in Mixed Linear Regression

We study variance-preserving diffusion of the response in mixed linear regression (MLR) with unknown mixing weights. Our analysis separates the statistical guarantees of score matching from the loss geometry and optimization signal at a fixed diffusion noise level. The KL divergence links the denoising score matching objective integrated over the diffusion path with the likelihood and a terminal discrepancy. Under mild regularity conditions and terminal schedule, the resulting estimator converges up to the ground truth parameters of MLR, and its scaled error converges to the Gaussian limit of the maximum-likelihood estimator. At a fixed scale of the diffusion noise level, we derive a decomposition linking the score matching loss to cross-entropy and Expectation-Maximization (EM) operators. This decomposition yields an EM-related low-noise gradient expansion with additional correction terms of latent variance. In the high-noise limit, we further characterize gradient descent on this limiting loss under isotropic covariance. Along fixed high signal-to-noise ratio rays, the score matching imbalance gradient and the latent-variance term tend to zero pointwise. Numerical experiments illustrate our theoretical findings and statistical guarantees.
Read the full abstract

Plasma Physics

A Minimal Self-Consistent Model for the Nonlinear Dynamics of Asymmetric Capacitively Coupled Radio-Frequency Plasmas

We develop a minimal self-consistent lumped-element model for the nonlinear radio-frequency dynamics of geometrically asymmetric capacitively coupled plasmas. The fast electrical response is described by three dynamical variables: the powered-sheath charge, the blocking-capacitor voltage, and the discharge current. The periodic RF solution is coupled to stationary particle and energy balances, which determine the electron temperature and plasma density from the operating conditions and the absorbed power. For an argon discharge, the model produces nonlinear current oscillations and a broad higher-harmonic response associated with the plasma series resonance. Variation of the RF voltage leads to correlated changes in the harmonic spectrum, absorbed power, and plasma density. Linearization about the periodic state relates the high-frequency response to the time-dependent differential sheath elastance. A moment-based effective elastance provides a compact estimate of the corresponding characteristic PSR frequency. The model thereby connects nonlinear RF dynamics and plasma sustainment within a minimal self-consistent description.
Read the full abstract

Mesoscale and Nanoscale Physics

Spectroscopy With Intelligent Feature Tracking for automated characterization of excited-state structure in semiconductor quantum dot devices

The characterization of excited-state structure in semiconductor quantum dot (QD) devices is an important component of tuning them for spin-qubit operation. We present Spectroscopy With Intelligent Feature Tracking (SWIFT), a framework that combines machine-learning (ML)-assisted feature identification with physics-informed geometric processing to extract energy-level splittings from pulsed-gate spectroscopy data. SWIFT isolates the relevant spectral features and exploits their characteristic geometry to reduce the two-dimensional spectroscopy analysis to a one-dimensional peak-detection problem. It further combines an ensemble-based confidence metric with sequential accumulation of rapid, low-SNR scans, allowing the inferred spectrum to be reevaluated as experimental evidence accumulates. Using Si/SiGe QD devices, we demonstrate SWIFT both offline and in real time, including automated tracking of QD excited states and lead resonances. Benchmarking on 255 manually labeled scans shows that SWIFT reduces the median splitting error to 0.05 mV from 0.11 mV for a classical baseline, with the largest improvement on lower-quality measurements. These results provide a path toward incorporating excited-state spectroscopy into autonomous QD characterization, tuning, and optimization, which will be essential in large-scale quantum dot devices.
Read the full abstract

All authors A. R. Huffman, Daniel Schug, Merritt Losert, Sanghyeok Park, Daniel King, Yuna Chun, Giordano Scappucci, M. A. Eriksson, Justyna P. Zwolak

High Energy Physics (Theory)

Searching for Bell-CHSH-Violating Bose Operators

We construct bounded Bell observables for a free scalar field in complementary spacetime wedges using measurable involutions of the spectral variable of a local canonical quadrature. These involutions yield wedge-local Hermitian unitary operators that anticommute with a chosen bounded observable. For a cosine observable, we characterize the admissible translations and reflections and derive exact correlations at finite modular bandwidth. The affine-reflection construction yields a violation of the Bell-CHSH inequality of approximately 2.06.
Read the full abstract

Information Retrieval

Overview of ROMCIR 2026: The 6th Workshop on Reducing Online Misinformation through Credible Information Retrieval

In the digital online ecosystem, we are surrounded by distinct forms of information pollution, posing significant threats to both individuals and society. Fake news, for instance, wields power to sway public opinion on matters of politics and finance. Deceptive reviews can either bolster or tarnish the reputation of businesses, while unverified medical advice may steer people toward harmful health practices. In light of this challenging landscape, it has become imperative to ensure that users have access to both topically relevant and factually accurate information that does not warp their perception of reality, and there has been a surge of interest in various strategies to combat misinformation through different contexts and multiple tasks. The purpose of the ROMCIR Workshop, for some years now, is precisely that of engaging the Information Retrieval community to explore potential solutions that extend beyond conventional misinformation detection approaches. Key objectives include identifying subjective and objective factors associated with information credibility and truthfulness, respectively, and integrating such factors as fundamental dimensions of relevance within IR Systems (IRSs), achieving early detection of misinformation, and ensuring that the search results retrieved are not only truthful but also explainable to the users of IRSs. Moreover, it is essential to evaluate the role of generative models such as Large Language Models (LLMs) in inadvertently amplifying misinformation problems, and how they can be used to support IRSs, together with the contribution that the human-in-the-loop paradigm can have in this context.
Read the full abstract

All authors Marcos Fernández-Pichel, Marinella Petrocchi, Kevin Roitero, Marco Viviani

Statistical Methodology

Nonparametric Hypothesis Testing of High-dimensional Clustering With Application to Single-cell RNA Data

Single-cell RNA sequencing studies routinely use clustering to define putative cell types and cell states, yet the observed separation may arise from sampling variability rather than genuine biological heterogeneity. This paper studies formal significance testing of such clustering structure in high-dimensional data. Existing SigClust methods assess clustering significance through Monte Carlo simulation under a Gaussian single-cluster null, but this assumption can be unreliable for normalized gene expression data and other non-Gaussian settings. We propose SigClust-LCP, a nonparametric extension that models a single cluster by a log-concave distribution. To make this approach computationally feasible in moderate to high dimensions, we develop a score-matching estimator for log-concave projection inspired by recent generative modeling ideas. We establish theoretical guarantees for the estimator and for its use in clustering significance testing. Simulations show that SigClust-LCP controls Type-I error more reliably than existing methods across a range of unimodal and mixture distributions while retaining competitive power. In a single-cell RNA sequencing analysis of Hydra cells, the method avoids spurious subclusters within annotated cell populations and supports biologically meaningful separation across lineages and body-axis regions.
Read the full abstract

Computation and Language

A Rubric-Guided Large Language Model Solution for Opioid Use Disorder Computable Phenotyping

Opioid use disorder (OUD) remains a public health crisis in the United States, yet it is difficult to identify from electronic health records (EHRs) because missing diagnosis codes and supporting evidence are buried in clinical narratives. Accurate OUD identification is critical to support interventions and improve health outcomes. This study developed a rubric-guided large language model (LLM) that incorporated Optimization by PROmpting (OPRO) for OUD computable phenotyping (CP). The framework used an 18-item, expert-identified rubric to instruct LLMs to automatically extract critical text with supporting evidence to determine OUD flags. Two UF Health physicians (GMR and WMG) chart-reviewed 253 patients, including 68 OUD-positive cases. Our LLM-based computable phenotype (CP) achieved the best F1 score of 0.774 and an AUROC of 0.934, outperforming the machine learning-based CP using EHR and natural language processing-extracted variables, and zero-shot LLMs by relative F1 improvements of 12.8% and 44.4%, respectively. The proposed LLM-based CP could link LLM-extracted evidence to OUD phenotyping for better explainability.
Read the full abstract

All authors Mengxian Lyu, Paredes Pardo, Cheng Peng, Ziyi Chen, Mengyuan Zhang, Jieting Li Lu, Gary M Reisfield, William M Greene, Jenny Lo-Ciganic, Yonghui Wu

Combinatorics

Stack Sorting on Words with Bounded-Repetition

We consider the family of operators of stack sorting (s_m)_{m >= 1} acting on words according to the following rule: s_m(w) is the output produced by the usual West algorithm of stack sorting on the input word w, but allowing for at most m repetitions of the same letter stacked in succession. Our first main theorem gives an explicit formula for the action of the operator s_m on binary words w = a^p b a^q (with distinct letters a > b): s_m(a^p b a^q) = a^max(p-m,0) b a^min(p,m)+q, d_m(a^k b) = ceil(k/m). From the above formulas it follows that for every m < m' we have d_m(w) >= d_{m'}(w) for binary words w = a^k b, and moreover, the ratio d_m(w)/d_{m'}(w) is always possible to be chosen exactly m'/m, so the ratios d_m/d_{m'} are unbounded for this family on the specified class – this is an explicit form of the expected separation-speedup phenomenon. Additionally, we provide explicit constructions of words of length 4 for which s_m and s_{m'} do not commute, and words of length 7 for which the speed d_m is not monotonic in m. Further, we provide a full recursive characterization – generalizing the classical 231-avoidance result of Knuth and West – of words which can be sorted using s_m only once, and then we employ it to show that there cannot exist an m-independent classical pattern avoidance condition for one-pass sortable words. Finally, we give a complete proof of the inequality d_1(w) >= d_m(w) >= d_infinity(w) for all m for two-letter words, together with a structural result establishing it for a single pass in general; nevertheless, we show that the family of operators (s_m) does not, after all, lie entirely between s_1 and s_infinity, exhibiting for every n >= 7 an explicit word w_n with d_1(w_n) = n-4 < n-3 = d_m(w_n) for every m >= 2, including m = infinity, so that the inequality above is false in general.
Read the full abstract

Astrophysics of Galaxies

Tracing the Hidden Molecular Gas in the Small Magellanic Cloud with SOFIA and APEX

In low-metallicity environments, reduced dust abundance weakens molecular cloud shielding against far-ultraviolet radiation, enhancing CO photodissociation and producing extended CO-faint or CO-dark molecular gas. The [CII] 158 $\mu$m line is a key tracer of these CO-faint phases. We characterize the multiphase origin of [CII] emission and quantify the CO-faint molecular gas fraction across several regions of the Small Magellanic Cloud (SMC), the nearest low-metallicity ($\sim 0.2 Z_\odot$) galaxy. We combine high spectral resolution SOFIA/GREAT and upGREAT [CII], APEX CO(2-1), and ASKAP+Parkes H I 21 cm data to decompose [CII] into its atomic and molecular contributions. On average, $(78\pm 1)\%$ of [CII] intensity originates from molecular gas and $(18\pm 1)\%$ from atomic gas. The [CII]-traced H$_2$ accounts for $(77\pm4)\%$ of the total molecular gas column density, indicating that the SMC molecular reservoir is dominated by CO-faint gas. We derive a H$_2$-to-CO conversion factor $X_{CO}\approx 8.9\times10^{20}$ cm$^{-2}$ (K km s$^{-1}$)$^{-1}$ on 4 pc scales in CO-emitting regions, 4.5 times the canonical Galactic value. We find no significant correlation between the CO-dark fraction and either total column density or $\Sigma_{SFR}$, suggesting that the CO-faint fraction is primarily governed by dust and gas shielding rather than current star formation activity. Our results demonstrate that [CII] is the dominant tracer of molecular gas in the SMC and confirm the prevalence of an extended, stable CO-faint molecular phase in low-metallicity environments.
Read the full abstract

All authors K. Rodriguez-Ambler, R. Herrera-Camus, M. Rúbio, A. D. Bolatto, D. Riquelme-Vásquez, K. Jameson, C. Muñoz-López, E. J. Tarantino, J. Stutzki

Quantum Physics

Quantum lower bounds for convex optimization and real matrix-vector query problems

We (the author and the AI systems that did the heavy lifting) show that the quantum query complexity of minimizing a convex function over a convex subset of $\mathbb{R}^n$ with evaluation and membership queries is $\tilde\Omega(n)$, nearly matching the best known upper bound. In particular, we show this even for quadratic minimization, which is equivalent to inverting an $n \times n$ real matrix using matrix-vector queries. We also show linear or nearly linear lower bounds on the quantum query complexity of computing the trace, the sign of the determinant, and the magnitude of the determinant of a real matrix in the matrix-vector query model. We use a novel quantum lower bound technique, the determinantal witness method, based on identifying a witness whose Fourier transform vanishes on low-rank matrices and that correlates well with the function being computed.
Read the full abstract

Statistical Methodology

Adaptive Regularization via Extreme Value Distributions for Gaussian Graphical Models

Edge selection in Gaussian graphical models is fundamentally a variable selection problem where pairwise relationships determine construct validity and variable importance in psychological networks. In psychology, network estimation relies predominantly on \(\ell_1\) regularization where uniform shrinkage systematically underestimates edge and centrality parameters. Alternative penalties overcome this bias but rely on fixed hyperparameters that do not adapt to the signal in the data. We develop a family of data-adaptive regularization penalties grounded in extreme value theory. Across 290 empirical psychological datasets, we show that absolute partial correlations are well-described by the Weibull distribution. Using this empirical regularity, we derive Weibull, Gumbel, and Exponential penalties that approximate \(\ell_0\) penalization and calibrate their hyperparameters to each dataset's noise floor. We formally prove the asymptotic properties of their static forms and conduct a large-scale simulation spanning two network topologies and various sample sizes (\(N\) = 100–10,000), demonstrating that their adaptive forms maintain high specificity while accumulating sensitivity as sample size increases with low parameter bias and high rank-order centrality congruence relative to field standards. Empirically, method choice alone determined centrality rankings at sample sizes typical in psychology. Of the three adaptive penalties, Weibull is recommended given the interpretability of its parameters.
Read the full abstract

All authors Alexander P. Christensen, Jeongwon Choi, Haoyi Yang, Lingzhou Xue

Computation and Language

Who Maintains Agent Skills? A Longitudinal Study of Human-Governed, AI-Assisted Skill Maintenance

Lifelong LLM agents increasingly rely on external skill artifacts as one element for preserving and reusing capabilities over time. These skills (usually portable Markdown files such as SKILL.md) describe when and how to apply a capability and must be corrected, expanded, and consolidated as tools and usage patterns shift over deployment. Recent work seeks to automate skill curation, but it largely evaluates against automated baselines and treats human maintenance as an unmeasured bottleneck. We study that missing process directly. We mine the full commit histories of five public AI-skill repositories, a purposive sample of AI-tooling organizations, covering 873 commits, 143 skill files, and 254 substantive post-creation edits from October 2025 to June 2026. We code each edit with pre-registered governance, operation, and trigger-evidence codebooks. Three findings emerge. First, every substantive edit is authored or merged through a named human account, while 62% carry an AI co-author trailer, with large repository-level variation. Second, these edits are genuine curation: an audited sample shows that most change skill content, and the coded operations are dominated by additions and corrections. Third, a pre-registered rule-likeness axis fails its reliability gate; reliably coding rule-likeness from commit artifacts remains an open measurement problem. We release the corpus, codebooks, mining scripts, and a replay protocol for automated skill curators. For self-evolving agents, public skill maintenance currently looks less like an autonomous pipeline than a human-governed, AI-assisted loop that future curators must measure against and operate within.
Read the full abstract

Machine Learning

PAC-Private Autoregressive Generation: Calibrating Noise to Ensemble Disagreement

Language models adapted on private text are often served through APIs, so privacy leakage occurs through generated outputs rather than exposed weights. Private prediction protects these releases. Methods such as PMixED incur privacy cost at each release and increasingly rely on the public model over long horizons. PAC privacy instead calibrates noise to output variability across possible secrets, adding less noise when predictions are stable. To our knowledge, PAC-private prediction has not previously been extended from classification to autoregressive generation. We construct $m=128$ overlapping worlds from the private corpus, with each record appearing in exactly $m/2$ worlds, and train one adapter per world over a frozen public model. The realized world is the secret. At each token, the public model defines a candidate set, the worlds vote, and their posterior-weighted disagreement determines the PAC noise; unanimity requires no calibration noise. We prove $I(S;Y_{1:T}) \leq I(S;H_T) \leq bT$. Our contributions are extending PAC privacy to autoregressive generation, handling adaptive self-generated contexts, and introducing coupled decoding that preserves privacy accounting while avoiding greedy degeneration. On WikiText-103 with GPT-2-small, we retain 74% of the fine-tuning gain at a per-token budget of $2^{-32}$, while membership-inference success is bounded by 51.08% after $10^6$ tokens; posterior-entropy estimates of leakage are roughly 17% of the charged budget. Inference privacy is not content protection: even when membership advantage on a memorized canary is indistinguishable from zero, the canary is emitted at the same rate. Against PMixED under matched membership-inference bounds on the same data universe and test set, we retain 98% of non-private headroom from $10^2$ to $10^6$ tokens, versus at most 56%, with no crossover.
Read the full abstract

Astrophysics of Galaxies

Reigniting the FUSE I: Star Formation in Massive Elliptical Galaxies

In this inaugural paper of our "Reigniting the FUSE" series, we present the data collection & reduction, methodology, and preliminary results for a comprehensive and homogeneous reanalysis of all 30 massive elliptical galaxies that have archival far-ultraviolet (FUV) spectra with the Far Ultraviolet Spectroscopic Explorer (FUSE). We perform SED fitting to these spectra, with aperture-matched photometry from GALEX, SDSS, DES/DECaLS, 2MASS, and WISE, and construct a catalogue of high quality constraints on their stellar masses, star formation histories, stellar and gas-phase metallicities, and dust opacities. As a part of this analysis, we have also compiled a comprehensive modern set of stellar template libraries, with high resolution in the FUV, and broad coverage from $90\,\mathring{\rm A}$ - $1000\,{\rm \mu m}$ at lower resolution, which is made available through modified versions of FSPS and Python FSPS. We find strong evidence to suggest that the youngest stars in these systems do not form out of recycled gas from the old stellar populations, but instead are fueled either by gas cooling out of the hot circumgalactic/intracluster medium, or cool gas stripped from nearby or merging dwarf systems. Future papers in this series will reexamine the O VI $\lambda\lambda$1032,1038 $\mathring{\rm A}$ emission in the context of the cooling flow problem.
Read the full abstract

All authors Michael Reefe, Michael McDonald, Eric Miller, Thomas Brown

Instrumentation and Detectors

The DarkLight Experiment at TRIUMF

This paper reports on the commissioning of the DarkLight experiment at the ARIEL superconducting electron linear accelerator located at TRIUMF in Vancouver, Canada. DarkLight was designed to utilize the ARIEL electron beam to search for a new boson, $A^\prime$, which preferentially couples to leptons, with a mass in the range 13-17 MeV/$c^2$. Such a boson could be produced via the $e^-~X\rightarrow e^-~X~A^\prime$ process, and the resulting $A^\prime\rightarrow e^+~e^-$ decay used to reconstruct the invariant mass of the $A^\prime$. The $e^+e^-$ pair is detected using a pair of magnetic spectrometers. Each spectrometer is instrumented with two triple GEM detectors and a scintillating strip hodoscope to trigger the event readout. Electron beam energies between 10 and 30 MeV were used to commission the experiment with a 1$~\mu$m thick carbon target. With both spectrometer polarities set to select electrons, elastic and Møller scattering were used to examine detector performance and compare with predicted distributions from Monte Carlo simulations. Planned upgrades to the e-linac will allow searches to be made with beam energies up to 50 MeV using a 1$~\mu$m thick tantalum target.
Read the full abstract

All authors R. Alarcon, D. Almonte, M.I. Alstaty, J. Azzi, R. Baartman, J.C. Bernauer, X. Braun, D. Ciarniello, C. Clabeaux, E.W. Cline, E. Corey, R. Corliss, W. Deconinck, A. Deshpande, J. Dilling, J. Dodge, D.H. Dongwi, C. Ekeman, E.A. Fielding, S. Frantzen, I. Friščić, G. Gelinas, S. Gelinas, K. Gill, A. Sabzevari Gonzalez, S. Gupte, D.K. Hasell, M. Hasinoff, B. Humphries, E. Ihloff, N. Jackson, D. Jayakodige, R. Kanungo J.E. Kelsey, O. Kester, M. Kohl, H.W. Koay, R. Laxdal, L. Levack, X. Li, W. Lin, C. Ma, A. Mahon, H. Merkel, J.W. Martin, L. Miller, R. Milner, K. Olchanski, K. Pachal, T. Planche, D. Preddy, S.D. Rädel, G. Randall, J. Redinger, D.A. Reiter, R. Richards, B. Scully, S. Shapiro, M. Suresh, C. Vidal, S. Wang, H. Witte, H. Yang, S. Yen

Geophysics

The tetragonal-cubic transition of davemaoite: Implications for lower mantle seismic anomalies

Davemaoite (CaSiO3 perovskite) is the third most abundant mineral in Earth's lower mantle and a dominant phase in subducted oceanic crust. Its crystal structure is distorted (tetragonal or orthorhombic) at ambient conditions but is considered to transform to cubic at high temperatures. Previous experiments reported low, nearly pressure-independent transition temperatures (approximately 600 K), demonstrating a long-standing discrepancy with theoretical predictions that mostly exceed 1000 K. Here, we determine the phase stability and thermal equation of state of CaSiO3 davemaoite and its titanium-bearing solid solution [Ca(Si0.75,Ti0.25)O3] under simultaneous high-pressure and high-temperature conditions using a laser-heated diamond anvil cell combined with synchrotron X-ray diffraction. We find that the tetragonal-to-cubic transition occurs at substantially higher temperatures than previously reported, with the titanium substitution further stabilizing the tetragonal phase and shifting the transition boundary to even higher temperatures. These findings indicate that CaSiO3 davemaoite in subducted oceanic crust likely undergoes its ferroelastic transition in the mid-lower mantle, whereas Ti-rich davemaoite may remain tetragonal throughout most of the lower mantle, transforming to cubic near the core-mantle boundary. Our results demonstrate that the compositionally dependent phase behaviour of davemaoite can account for the seismic anomalies observed in both the mid-lower mantle and the lowermost mantle.
Read the full abstract

All authors Yoshiyuki Okuda, Bin Chen, Juliana Peckenpaugh, Keng-Hsien Chao, Saori Kawaguchi-Imada, Hirokazu Kadobayashi, Zhenxian Liu, Dongzhou Zhang

Quantum Physics

Kerr Induced Control of Synchronization and Quantum State Recovery in a Driven van der Pol Oscillator

We investigate how Kerr nonlinearity modifies quantum synchronization in a squeezed quantum van der Pol oscillator. We show that the Kerr interaction produces an amplitude-dependent frequency shift that drives a saddle-node bifurcation, transforming the classical phase-space structure from bistable to monostable dynamics. In the quantum regime, this transition manifests as systematic frequency pulling and spectral broadening, while the steady-state Wigner function reveals a continuous correspondence between the quantum state and the semiclassical attractor despite finite quantum fluctuations. By constructing global synchronization phase diagrams in the squeezing–Kerr parameter space, we uncover a remarkably linear dependence of the critical squeezing strength required to maintain phase locking on the Kerr nonlinearity. We further demonstrate that the synchronization boundary does not coincide with the crossover between super- and sub-Poissonian photon statistics, showing that synchronization and photon-number statistics characterize distinct aspects of the quantum steady state. These results provide quantitative design principles for controlling quantum synchronization through Kerr nonlinearity, with potential relevance to trapped-ion, superconducting-circuit, and optomechanical platforms.
Read the full abstract

Chemical Physics

Differentiable Solvation Shell Model for Rational Electrolyte Design

Tailoring Li+ solvation structures has emerged as a promising design principle across multiple Li metal battery (LMB) electrolytes, with localized high concentration electrolytes (LHCE) emerging as a leading candidate. However, rational design of these electrolytes remains limited by a poor quantitative understanding of how molecular properties govern Li+ solvation shell composition. Here we introduce a mean field modeling framework, built on an Ising model, that predicts solvation shell composition from donor number (DN), acceptor number (AN), molar ratio, and molecular size. This framework is end-to-end differentiable, which enables parameterization directly from molecular dynamics (MD) solvation structures via gradient-based optimization. Focusing on LHCE, the model achieves 10.7% RMSE and R2=0.87 for shell composition, 2.3% RMSE on free solvent ratio, and further reproduces solvation trends in real LHCE electrolytes, at a fraction of MD's cost. Our model also shows good generalizability on other electrolyte systems in addition to LHCE. Analysis of its interaction terms shows that DN dominates Li+ solvation energetics, providing a thermodynamic basis for empirical DN-based design rules. We further demonstrate the model's design utility on the LiTFSI/tetraglyme (G4) system absent from training. The model identifies fluorobenzene (FB) as a promising diluent and predicts a salt-concentration window with anion-rich solvation shells and low free solvent, favorable for stable anion-derived SEIs and high oxidative stability, respectively. This prediction is validated by higher-fidelity MD. This work establishes an interpretable differentiable framework for predicting Li+ solvation shell composition, with potential to extend to other electrolyte classes.
Read the full abstract

All authors Hancheng Zhao, Hongyi Lin, Celia Kelly, Venkatasubramanian Viswanathan

Algebraic Geometry

Coarse moduli of motivic augmentations

A variety $X$ over a suitable base $Z$ gives rise to a highly structured algebra $C^*(X)$ in motives over $Z$. In turn, a $Z$-point gives rise to an augmentation $C^*(X) \to 1$. This assignment $X(Z) \to Aug(C^*(X))$ factors the so-called “unipotent Kummer map” to torsors under the unipotent fundamental group in realizations. In one direction, this suggests the possibility of “performing” Chabauty-Kim theory motivically without waiting for a motivic t-structure. In a different (largely independent) direction, we may hope to extract arithmetic information for use in bounding sets of integral points from the full rational homotopy type going beyond $\pi_1$. In both directions, the coarse space for motivic augmentations $Aug(C^*(X))$ would benefit from a structure of finite type $\mathbb{Q}$-variety, and similarly, the coarse space $Aug(C^*_{F\phi}(X))$ of augmentations in filtered $\phi$ modules would benefit from a structure of finite type $\mathbb{Q}_p$-variety. We establish two criteria for representability, compute these spaces in several examples, and construct a comparison with Selmer varieties. Finally, we demonstrate how these constructions lead to K-theoretic finiteness criteria in an example.
Read the full abstract

Analysis of PDEs

The stability conjecture for the Bernoulli problem in dimension four

We prove that every global classical stable solution to the one-phase Bernoulli problem in $\mathbb R^4$ is one-dimensional. In particular, if nonempty, its free boundary consists of one or two parallel hyperplanes. This settles the stability conjecture for the Bernoulli problem in dimension four. As consequences, we obtain universal local Hessian and free-boundary curvature estimates for classical stable solutions in $\mathbb R^4$, as well as axial symmetry of global classical solutions with finite Morse index in $\mathbb R^4$.
Read the full abstract

Mesoscale and Nanoscale Physics

A magneto-mechanical gyroscope with spintronic readout

Gyroscopes are essential elements in navigation, consumer electronics, robotics, and aerospace applications. Most micro electro-mechanical systems (MEMS) implementations rely on capacitive sensing mechanisms, which limit the dimensional scaling to the micrometer scale. In this work, we introduce a MEMS-like two-degree-of-freedom (2-DOF) gyroscope that exploits the rectification functionality of magnetic tunnel junctions (MTJs) as its readout mechanism and as the transducer of the mechanical dynamics. Experimentally characterized MTJs have been used to calibrate and perform an experiment-informed design of the magneto-mechanical model combining micromagnetic theory with 2-DOF mechanical equations. We demonstrated that the output is linear with angular rate, and that the proposed device is able to extract the angular rate in dynamic cases exploiting a homodyne demodulation approach. The results open a path towards a compact, complementary metal-oxide semiconductor (CMOS)-compatible readout pathway that relaxes reliance on tight capacitive gaps and motivates multi-physics designs of the device.
Read the full abstract

All authors Andrea Meo, Francesca Garesci, Pedro Bossi Nunez, Maddalena Fiorentino, Teresa Natale, Ludovico Dindelli, Victor Lopez-Dominguez, Banibrato Sinha, Riccardo Tomasello, Francesco Dell'Olio, Pedram Khalili Amiri, Mario Carpentieri, Giovanni Finocchio

Software Engineering

The Impact of GenAI on the Future of Requirements Engineering

Recent advances in artificial intelligence (AI), particularly large language models (LLMs), are transforming how we design and build systems by increasing access to domain knowledge and by providing automation support to software engineering (SE). As implementation becomes less expensive through generalist SE agents, engineering effort shifts away from writing correct code and toward expressing, curating, verifying, and evaluating requirements. In this paper, we survey the state of the art in AI for requirements engineering (RE) research leading up to the transformation, before reviewing advances in LLMs. We survey two subsequent research areas: prompt programming, which treats LLM instructions as a program in SE vernacular, and generalist SE agents, which combine multiple LLM advances to yield semi-autonomous processes that complete SE tasks. Finally, we explore the future of requirements engineering along two axes: matters changing how we interact with requirements through the SE process, and matters changing how requirements are experienced by software developers and stakeholders more broadly, including end-users. This article aims to inform how RE researchers can navigate this transformation in the selection of future research priorities.
Read the full abstract

General Relativity and Quantum Cosmology

Fermionic Casimir memory from an anisotropic gravitational pulse

We investigate the late-time energy deposited by a weak, spatially homogeneous anisotropic gravitational pulse in a massless Dirac field confined between two static parallel plates. The field satisfies MIT bag boundary conditions, while the prescribed geometry approaches Minkowski spacetime in both asymptotic time regions. Using a tetrad formulation, we show that the linear spin-connection contribution cancels for the traceless anisotropy, leaving a coupling to the shear component of the Dirac stress tensor. The pulse produces real fermion–antifermion pairs, whose deposited energy is evaluated through the corresponding Bogoliubov coefficients. Subtracting the production in unbounded space isolates a finite boundary-dependent contribution governed by the pulse duration relative to the characteristic spectral timescale set by the MIT gap. This contribution is negative in the adiabatic regime, where confinement suppresses excitations relative to the continuous free spectrum, and approaches zero from above in the impulsive limit. The resulting pressure correction is nonmonotonic and exhibits two sign changes, alternately reinforcing and reducing the ordinary attractive Casimir force. Unlike the bosonic vacuum-polarization memory previously obtained for the same class of backgrounds, the present effect is a dissipative fermionic memory carried by gravitationally created pairs. Here, memory denotes the persistent late-time modification of the quantum state after the pulse has vanished.
Read the full abstract