Peer Revenue Past citations are no guarantee of future returns

Papers

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

All of arXiv

Friday 11 September

Distributed Computing

PEAT: Pseudo-Error Assessment for GPU Kernel Validation in DNN Training

Deep neural networks (DNNs) are widely adopted in various fields, driving an emerging trend in developing software stacks associated with DNN training systems. For example, many codes have been ported across different frameworks or developed to leverage the computing power of GPUs or domain-specific accelerators. However, validating a kernel implementation in DNN training is time-consuming and generally requires massive storage. Specifically, this poses a fundamental question: how to characterize the behavior of a new implementation when it is integrated into a DNN training flow. Unfortunately, this problem is not well investigated in the literature, to the best of our knowledge. To address this shortcoming, we present PEAT - a lightweight inspection framework for \underline{P}seudo-\underline{E}rror \underline{A}ssessment associated with GPU kernel validation in DNN \underline{T}raining. Firstly, inspired by conventional fault injection (FI), PEAT's Profiler invokes an operation-wise kernel in a training flow to collect a DNN model's states (e.g., checkpoints and activations). More importantly, the Profiler introduces two simple yet effective techniques, playback FI and frequency-based runtime FI, leveraging persistent kernel calling during the training process. Secondly, PEAT's Analyzer characterizes profiled errors, revealing some signatures from the error distribution of a kernel compared to the golden one. Lastly, PEAT's Detector provides some guidelines as a sufficient condition, which enables associating several well-known error models with signature patterns. We demonstrate the applicability of our approach by presenting the results and analysis using GPUs from the two most popular vendors, NVIDIA V100 and AMD MI250, on various AI models, from vision tasks to language models, for both pretraining and finetuning scenarios.
Read the full abstract

All authors Xuan Truong Nguyen, Hong Quan Tran, Tuan Duc Chu, Thanh Tuan Dao

Artificial Intelligence

Asclepius: An Adaptive Harness for Long-Horizon Clinical Agents

LLM agents are predominantly benchmarked on short, single-task trajectories, yet real deployments run for hours under contention, surfacing a different class of failures. We use the Clinical Environment Simulator (CES), in which an agent manages an entire emergency-department shift under continuous time and resource pressure, as a testbed: long-horizon execution failures manifest measurably in a single rollout under structured, multi-dimensional grading. On CES, current agents reach the correct diagnosis in most cases yet fail to deliver complete and timely critical actions, revealing an execution gap. We attribute this gap to three long-horizon failure modes, each operationalized as a per-trace counter: instruction-adherence drift, treatment incompleteness, and a severity-equity gap in timeliness. We then introduce Asclepius, an adaptive agent scaffolding with a self-evolving harness that rewrites the operating manual between shifts from trace-level feedback, an externalized clinical skills library for high-stakes regimen knowledge, and three isolated subagents that partition per-turn decisions across the patient queue. On held-out batches never observed during harness evolution, Asclepius improves critical-action correctness by 22% (p = 0.024) over a strong baseline agent framework while preserving diagnostic accuracy, with consistent gains across five LLM judges from three model families; on the full ten-batch set, improvements reach 25% on critical actions and 13% on timeliness. The three failure modes form a coupled bottleneck: decisive reductions appear only when all three components act together.
Read the full abstract

All authors Grace Chang Yuan, Xiaoman Zhang, Sung Eun Kim, Luyang Luo, Pranav Rajpurkar

Materials Science

Local Response Theory of Electrified Interfaces from Screened Effective Charges

First-principles, electronic-structure simulations provide direct access to ground-state, global system properties. In the context of computational electrochemistry these are, e.g., bias-dependent energies, free energies, work functions, and capacitances. However, a comparably systematic local description of how these quantities respond to biasing via interfacial charging is still missing. Here, we introduce the screened electrochemical Born effective charge (SEBEC), a mixed total-energy derivative that measures the force response of an atomic degree of freedom to interfacial charging under electrochemical boundary conditions. SEBECs play the role for electrochemistry that Born effective charges play in the modern theory of polarization: they provide an unambiguous decomposition of global response and generate the same formal structures for structural relaxation, capacitance, and electrochemical Stark tuning that are familiar from solid-state response theory. The broader implication is a local response framework that brings electrochemical interfaces closer to the conceptual rigor long established for insulating solids.
Read the full abstract

Materials Science

Slow Thermalization and Long-Lived Coherence across Acoustic Phonon Branches in BAs

The exceptionally high thermal conductivity of cubic boron arsenide (BAs) arises from suppressed three-phonon scattering associated with its large acoustic–optical gap. In this regime, four-phonon processes become non-negligible, creating an unusual scattering hierarchy with unexplored consequences for nonequilibrium phonon dynamics. Here, time-resolved x-ray diffuse scattering reveals exceptionally slow, branch-dependent acoustic phonon thermalization and long-lived coherence. Exploiting branch sensitivity in one-phonon diffuse scattering, we resolve transverse acoustic (TA) and longitudinal acoustic (LA) thermalization times of 29.2 and 13.9~ps, respectively; even the LA timescale is at least three times that in common semiconductors. Phonon dispersion calculations assign coherent oscillations at 0.19 and 0.27~THz to the TA and LA modes, respectively; both persist with little decay over 50~ps, with comparable coherence found only in diamond. These findings provide a direct dynamical manifestation of the weak phonon scattering underlying the exceptional thermal transport of BAs.
Read the full abstract

All authors Zeyu Xiang, Ying Peng, Ange Benise Niyikiza, Fanghao Zhang, Haoyuan Li, Takahiro Sato, Thomas Linker, Yanwen Sun, Meredith Henstridge, Vincent Esposito, James F. Cotter, Diling Zhu, Zhifeng Ren, Bolin Liao

High Energy Astrophysical Phenomena

Observable Signatures of Supernova Shock Breakout in Confined Circumstellar Medium

Supernova shock breakout encodes rich information about stellar explosions, including the properties of the progenitor, the explosion mechanism, and the circumstellar environment. We present the first two-dimensional multigroup radiation-hydrodynamic simulations of shock breakout from Red Supergiant (RSG) progenitor stars embedded in confined circumstellar medium (CSM) produced shortly before core collapse. Our simulations reveal that radiation escaping ahead of the shock forms a radiative precursor that modifies the shock dynamics, drives multidimensional mixing within the radiating gas layers, and alters the emergent radiation from the CSM. For the CSM models explored here, the resulting bolometric light curves reach peak luminosities of ${\sim}1\text{-}3 \times 10^{44}\ \mathrm{erg\ s^{-1}}$ with durations of ${\sim}4\text{-}35$~hr, with these signatures controlled by the CSM mass and the confined radius. Our multi-wavelength calculations further reveal a pronounced extreme-ultraviolet emission, where ionizing photons of energies ${\ge}54.4$~eV can sustain flash-ionized He II and power strong emission from its recombination line of $\lambda 4686\ \mathring{A}$. Furthermore, we identify diagnostic signatures of shock breakout with dense CSM and assess their detectability with Einstein Probe and ULTRASAT. Our models and predictions provide promising diagnostics of shock breakout in CSM that may be detectable by current and forthcoming high-cadence ultraviolet and X-ray space telescopes.
Read the full abstract

All authors Wun-Yi Chen, Ke-Jung Chen, Keiichi Maeda, Friedrich K. Röpke

Optimization and Control

Assortment Control Unlocks the Value of Dynamic Pricing in Mixed Last-Mile Delivery

The growth of e-commerce has intensified the need for last-mile delivery systems that can jointly manage customer choice and operational efficiency. We study the Dynamic Offering and Pricing of Mixed Delivery Options problem, in which a logistics service provider dynamically selects and prices attended home-delivery time slots and out-of-home pickup options for sequentially arriving customers. Each decision affects immediate revenue, customer acceptance, and the route-dependent fulfillment cost realized at the end of the booking horizon. We formulate DOPMDO as a finite-horizon Markov decision process and propose State-Value Anchored Pricing via Approximate Dynamic Programming. The method learns a continuation-value approximation on an aggregate state representation of the mixed-delivery system and uses accepted-versus-rejected value differences to estimate option-level opportunity costs. These opportunity costs are embedded in an anchored trust-region pricing problem around a calibrated fine-static benchmark. Computational experiments on the real-world Seattle instance show that SVAP–ADP increases mean episode profit by 7.0% relative to current practice (95% CI: 6.7–7.2%), primarily by reducing terminal fulfillment cost while maintaining a stable home–locker–opt-out mix. Assortment-control experiments show that dynamic pricing is most effective when the menu exposes operationally valuable locker alternatives, with richer candidate pools delivering substantially larger gains than restricted nearest-locker menus. These results indicate that anticipatory pricing and assortment control are complementary: pricing steers customers toward lower-cost options, but the menu determines whether high-value consolidation opportunities are available in the first place.
Read the full abstract

Biological Physics

Protein eXplosion Imaging (PXI): Protein Structures from Laser-Driven Explosions

We investigate the structural information retained in the distribution of explosion ion trajectories from proteins subjected to strong ionization. Using molecular-dynamics simulations and machine-learning analysis benchmarked against an analytical approach, we show that low-resolution structural information can be retrieved from the explosion distributions alone. An ensemble of convolutional neural networks trained on simulated spherical ion maps recovers the radius of gyration and the three semi-axes of an ellipsoidal molecular envelope with prediction errors of 1.2 Å and 1.5 Å, respectively, while an analytical ellipsoid charge model returns similar estimates and performs best for globular structures. We further investigate how higher-level structural information and symmetries can be extrapolated from ion measurements. This study supports the viability for structural determination of single proteins without large-scale X-ray facilities.
Read the full abstract

All authors Alfredo Bellisario, Tomas André, Carl Caleman, Nicuşor Tîmneanu

Distributed Computing

Dynamic HBM Repartitioning for Multi-Turn MoE Serving

Long-running multi-turn requests accumulate reusable key-value (KV) state. Once this state exceeds a fixed GPU KV-cache allocation, serving systems evict reusable prefixes, repeat prefill work, and may preempt requests. This pressure is particularly acute for Mixture-of-Experts (MoE) models: their expert weights occupy most GPU high-bandwidth memory (HBM), even though each token activates only a sparse subset of experts. A static boundary between weights and the KV cache prevents serving systems from using expert memory to preserve reusable state as conversations grow. We present VAMP, an MoE serving framework that changes this HBM boundary at runtime. When a KV allocation cannot be satisfied, VAMP compares the estimated future work of three alternatives: staging expert weights from host memory, evicting cached prefixes that may require re-prefill, or preempting and rescheduling requests. It then converts a bounded expert-weight region into KV-cache capacity when that action has the lowest estimated penalty. CUDA Virtual Memory Management page remapping performs this conversion without copying resident KV data. We implement VAMP in the vLLM serving engine and evaluate Qwen3-Next-80B on recorded and controlled multi-turn workloads. Across five replays of a recorded 2,103-turn SWE-bench agent workload, VAMP with a 15% maximum expert-offloading ratio reduces time-to-first-token (TTFT) p90 from 26.1 s to 1.10 s (23.6 times) and increases request throughput by 20.7% relative to unmodified vLLM, while increasing time per output token (TPOT) by 31.1%.
Read the full abstract

All authors Jinpyo Kim, Mingi Kwon, Younghoon Min, Jongryool Kim, Jishen Zhao

Applied Physics

Identifiability of the three-dimensional anisotropic thermal-conductivity tensor in thermoreflectance measurements: Single-orientation limits and full-tensor recover

Materials with full anisotropic thermal conductivity tensors are widely used in advanced applications, creating a need for dedicated characterization. Such tensors contain six independent components: three diagonal conductivities and three non-zero off-diagonal couplings. However, whether thermoreflectance measurements on a single crystalline orientation can uniquely determine an arbitrary tensor remains unclear. Here, we show that every fully anisotropic thermal-conductivity tensor admits a Schur-equivalent tensor with zero cross-plane couplings ($k_{xz}=k_{yz}=0$). The two tensors produce identical surface thermal responses, and this equivalence persists as frequency, delay time, beam geometry, and spatial offset vary. We further demonstrate that parameter-identifiability assessments based on heuristic interpretations of sensitivity trends or singular-value decomposition (SVD) of the sensitivity matrix cannot, by themselves, rule out globally equivalent solutions in fully anisotropic systems. Combining systematic analysis with numerical validation, we provide guidance on selecting signals and measurement orientations for full-tensor characterization. Specifically, measurements at three mutually orthogonal crystallographic orientations provide complementary information for full-tensor recovery. These findings clarify the limits of single-surface measurements and establish a systematic approach to characterizing fully anisotropic thermal-conductivity tensors.
Read the full abstract

Artificial Intelligence

From Legal Text to AI-specific Risk Sources: A Systematic Analysis of the EU AI Act's High-Risk Requirements

The EU AI Act introduces mandatory requirements for high-risk AI systems with the explicit goal of ensuring the development and operation of trustworthy AI. At the same time, AI risk management practices rely on structured risk taxonomies to systematically identify and treat AI-specific risk sources. As both the AI Act and established risk taxonomies aim to address AI-induced risks, a natural question is whether they align in the risk sources they cover. However, no clear mapping exists between the risks implicitly addressed by the Act's high-risk requirements and established taxonomies, leaving practitioners without a structured basis for aligning regulatory obligations with AI risk management practice. This paper presents a systematic classification of the requirements extracted from the EU AI Act Section 2 (Requirements for high-risk AI systems), revealing that only a minority directly address AI-specific risk sources, while the majority impose organizational process and documentation obligations. From the AI risk-related requirements, a consolidated list of distinct AI-specific risk sources is derived. The resulting EU AI Act Risk Source List takes an important step towards bridging the gap between legal obligation and AI risk management practice, providing a structured reference for explicit comparison between existing AI risk taxonomies and the risk sources implicitly addressed by the EU AI Act. Important Note: This is the authors' preprint. The paper was presented at the 4th International Conference on Frontiers of Artificial Intelligence, Ethics, and Multidisciplinary Applications. A link to the conference's official proceedings will be provided upon publication.
Read the full abstract

All authors Ronald Schnitzer, Mike Auer, Rumpa Choudhury, Andreas Hapfelmeier, Maximilian Hoeving, Isabelle Painter, Josiane Xavier Parreira, Sonja Zillner

Computation and Language

Harmfulness Propagation Dynamics: Layer-wise Trajectories of Adversarial Intent in Large Language Models

We identify Harmfulness Propagation Dynamics (HPD): for harmful prompts, the projection of the last-token hidden state onto a learned harm direction rises monotonically with transformer depth, whereas benign prompts remain flat or oscillatory. This cross-layer signature reflects harmful intent as a progressively resolved semantic property: surface form appears early, while pragmatic intent consolidates later, making the trajectory shape more informative than any single-layer snapshot. Moreover, LDA-based harm directions, learned per layer, remain stable across random splits (pairwise cosine similarity $>0.97$), supporting the projection sequence as a reproducible structured signal. Building on HPD, we introduce \textbf{\herald{}} (Harmful Encoding Recognition via Activation Layer Dynamics). This lightweight input moderator extracts a seven-dimensional feature record, slope, curvature, monotonicity, onset layer, and related statistics from the cross-layer projection sequence and classifies it with a 288-parameter MLP. \herald{} stores one $d$-dimensional direction per layer ($262$\,KB for a 32-layer, $d{=}4096$ model), requires no gradient computation during training, and adds only $2.6{\times}10^{-6}$ prefill FLOPs at inference. Across eight prompt-harmfulness benchmarks and four model families, \herald{} achieves an average F1 of $89.3$ on OLMo2-7B, surpassing all tested guard models on adversarial jailbreak detection ($98.4$ vs.\ $96.9$ F1) and outperforming prior latent-based methods by $2.3$-$4.1$ F1 points on every backbone. Per-instance trajectories provide machine-readable audit records that reveal when and how harmfulness emerges, offering an interpretability advantage over single-layer approaches.
Read the full abstract

Neural and Evolutionary Computing

Early-Stopping Thresholds for ES-HyperNEAT: A Data-Driven Approach from Fitness Dynamics

Most hyperparameter configurations for Evolvable-Substrate HyperNEAT (ES-HyperNEAT) produce networks that stagnate at random-guessing performance, wasting computational resources. We frame early stopping as binary classification on early fitness trajectories: for each trial, we compute the cumulative median of best-per-generation fitness and test it against a threshold derived by maximizing the F1 score on an initial 90-trial dataset. The resulting rule (generation G* = 3, threshold T* = 0.140) achieves F1 = 0.872 on 180 independent validation trials, retaining over 90% of successful trials while cutting computational cost by 41.6%. Compared to Hyperband, our domain-specific rule is 64% more efficient with higher mean fitness, though Hyperband occasionally discovers higher peak solutions. On a converged search population the rule becomes too aggressive (recall 31.1%), motivating adaptive thresholds. The specific thresholds are ES-HyperNEAT-specific, but the methodology, deriving stopping criteria from fitness dynamics classification, is applicable to other evolutionary algorithms with stagnation-prone hyperparameter spaces.
Read the full abstract

All authors Romain Claret, Arthur Gygax, Michael O'Neill, Paul Cotofrei, Pascal Felber

Materials Science

Non-reactive sintering enhances density and ionic conductivity of NASICON solid electrolytes

NASICON materials are promising solid electrolytes for room-temperature sodium solid-state batteries and are typically synthesized via solid-state reaction. While sintering has been extensively studied, the effect of calcination on electrolyte properties remains poorly understood. In this work, the temperatures at which the NASICON phase forms in Na3Zr2Si2PO12 and Na3.4Zr2Si2.4P0.6O12 are identified. Calcination temperature is then varied between 900 {\deg}C and 1200 {\deg}C to obtain powders with different degrees of reaction prior to sintering. Under identical sintering conditions, higher NASICON phase content in the calcined powder is shown to yield denser electrolytes. Non-reactive sintering also improves grain boundary conductivity, increasing it by 130% for Na3.4Zr2Si2.4P0.6O12 and raising total conductivity from 1.60 to 2.95 mS/cm for powders calcined at 900 {\deg}C and 1200 {\deg}C, respectively. Finally, it is shown that phosphorus loss during processing compromises cycling stability against Na metal electrodes, and that adding off-stoichiometric phosphorus resolves this issue, while reaching a critical current density of 5.0 mA/cm^2 and a room-temperature conductivity of 3.82 mS/cm with over 400 hours of stable cycling. Overall, these findings directly relate synthesis and processing conditions to the final material properties and electrochemical performance of NASICON solid electrolytes.
Read the full abstract

All authors Andrea Cornelio, Björn Mieller, Johannes Baller, Andrea Fantin, Janina Roik, Jessica Kindt, Jonas Krug von Nidda, Tim-Patrick Fellinger, Gustav Graeber

Machine Learning

Attention Is All You Need (to Avoid Spurious Oscillations)

Can attention move a shock across several cells in one update without breaking it? We develop a conservative, fixed grid finite-volume scheme in which a CFL-conditioned attention flux selects upstream information according to the transport required by the current time step. One-dimensional inviscid Burgers transport is used as the central mechanism test: the same learned flux remains reliable in the conventional small-step regime and, with a time step four times larger, preserves sharp shocks while using one stage per update. A standard fifth-order WENO scheme with third-order strong-stability-preserving Runge-Kutta time integration (WENO-5+SSP-RK3) is included alongside controlled Forward Euler comparisons to separate flux selection from time integration. The learned attention shifts upstream with the local transport reach and becomes more selective near shocks; inference-time interventions and retrained ablations show that transport-scale information and state-dependent selection contribute directly to performance. Directional two-dimensional scalar Burgers transport and the one-dimensional shallow-water system then test whether the conservation-scale-selection principle transfers beyond the original scalar setting. The results support attention as a learnable information stencil for conservative large-step shock transport, while identifying finite candidate reach and problem-dependent robustness as the present limits.
Read the full abstract

All authors Jinyoung Jeong, Joseph B. Choi, Xinlun Cheng, H. S. Udaykumar, Sanghun Choi, Stephen S. Baek

Medical Physics

The Electrodynamic Basis of Dichroism-Mediated Polarization Perception

Humans see the polarization of light through entoptic percepts arising from the macula's Henle fiber layer, where xanthophyll pigments absorb preferentially across the radiating fibers. From the layer's complex dielectric tensor alone, Maxwell's equations in Berreman $4\times4$ form yield its Mueller matrix, set by four scalars $\{A,B,C,D\}$. Its intensity channel depends only on the dichroic pair $A$ and $B$, which fix the percepts at a maximum contrast $|B|/A\approx0.05$. One relation generates the whole dichroism-mediated family: Haidinger's brushes under a uniform field, their dark arms perpendicular to the $\mathbf{E}$-vector; fractured brushes under spatially varying fields; and $N$-fold brushes under vector-vortex illumination.
Read the full abstract

All authors Gary P. Misson, Dusan Sarenac, Dmitry. A. Pushin, Shelby E. Temple

Machine Learning

Generative Interpretability via Scalable Neuro-Symbolic Models

As the use of Large Language Models moves from chatbots into agentic systems, where outputs become actions with irreversible consequences on reality, the existing paradigm on AI Interpretability research, post-hoc interpretability, is structurally inadequate for safe and trustworthy model deployment: it explains behavior after the fact but cannot audit or intervene in an inference computation before it commits to an output. We therefore argue for a shift toward generative interpretability, an architectural property under which a model's inference pass natively exposes semantically meaningful checkpoints that are human-understandable and amenable to causal intervention. We show the merits of generative interpretability as comparison to other interpretability research paradigms, and propose Neuro-Symbolic Models as a concrete instantiation.
Read the full abstract

Optics

Dynamically-scaled local patch method for numerically simulating the scattering effects of small particles into large beams

Simulating the propagation of a large beam with high spatial resolution is a common challenge in optics due to both computational time and memory demands. While computational time sets a practical limit on what one can simulate, memory resources are an absolute hard limit. In this paper, we report a novel method to overcome a gap in spatial scale between the beam and the relatively small optical disturbance that needs to be modeled, such as those due to particulate contamination. The method splits the entire propagation chain into to two paths: a specular field propagation on coarse grids and scattered fields on fine grids. Especially for the latter, we dynamically apply high-resolution local patches for individual particles. Not limited to particulate contamination, this method is applicable to any tiny discontinuous structures on the beam propagation path without memory overflow triggered.
Read the full abstract

Earth and Planetary Astrophysics

Incompatibility of iron in post-perovskite and the stability of basal magma oceans in super-Earths

Post-perovskite is expected to dominate much of the solid mantles of rocky exoplanets, yet iron partitioning between post-perovskite and silicate melt, which controls the compositional evolution and buoyancy of crystallizing magma oceans, is unconstrained at these pressures. We use first-principles molecular dynamics and thermodynamic integration to compute the Fe–Mg distribution coefficient $K_D$ between post-perovskite and (Mg,Fe)SiO$_3$ liquid at 150–600~GPa and 6000–10000~K. Iron is strongly incompatible in post-perovskite and becomes increasingly so with pressure. Combining $K_D$ with equations of state, we find that iron enrichment of residual liquid reverses the solid–liquid density contrast, causing post-perovskite to become buoyant. Basal magma oceans are therefore gravitationally stable in super-Earth exoplanets up to 4 M$_{\oplus}$.
Read the full abstract

Optimization and Control

A family of spectral conjugate gradient algorithms derived by least-squares approximations based on a modified quasi–Newton update with application to a revised robust binary classification model

We develop a spectral three-term modification of the classic Hestenes–Stiefel conjugate gradient algorithm, preserving its anti-jamming characteristic and, simultaneously, taking care of the sufficient descent property. We discuss how a modified secant equation can be extracted from our modification scheme, yielding a memoryless BFGS updating formula. Then, the spectral parameter of our method is obtained by steering its direction toward the given BFGS direction within a least-squares context. Using our technical improvements, we outline the general framework of our algorithm and discuss its theoretical features, including the descent and convergence properties, without the convexity assumption. We put our algorithm to the test in comparison with the three other conjugate gradient algorithms on a set of CUTEr unconstrained optimization test models, comparing the outputs using the Dolan–More measure. Next, we provide a concise evaluation of the results, highlighting the practical advantages of our algorithm. As a real-world case study, we introduce a reduced quadratic surface SVM with the rescaled loss for robust binary classification and apply the proposed algorithm to assess its accuracy and training time against several other SVMs.
Read the full abstract

Cosmology and Nongalactic Astrophysics

Supernova Cousins: The Intrinsic Dispersion and Environmental Dependence of Type Ia Supernovae in Galaxy Groups with TITAN

At low redshift, scatter in the Type Ia supernova (SN Ia) Hubble diagram is increased by peculiar velocities (PVs) and correlations between SN distances and their host-galaxy environments, limiting the precision of cosmological distance measurements. We investigate these effects using 1086 spectroscopically-confirmed SNe Ia from the Type Ia Supernova Trove from ATLAS in the Nearby Universe (TITAN) sample in the redshift range $0.01<z<0.05$. We identify 314 group-associated SNe, including 80 SN "cousins" occurring in different galaxies within 32 common groups. We first examine whether using the group-averaged redshifts of SNe in galaxy groups can mitigate PV-induced scatter; however, we find that the Hubble residual scatter increases slightly from $0.184^{+0.017}_{-0.009}$ mag using individual host-galaxy redshifts to $0.200^{+0.016}_{-0.015}$ mag using group-averaged redshifts. For the sample of SN cousins, we next analyze pairwise differences in their distances and compare their scatter to the Hubble residual scatter of the full SN sample. For analysis variants that limit line-of-sight differences in distance modulus between cousins, we find $\sim$30-40% lower scatter from SN cousins at significances of 2.3-2.8$\sigma$. We also find tentative evidence that these differences in scatter may be driven by the youngest groups, with a 2$\sigma$ reduction in scatter for groups with high fractions of spiral galaxies. These results suggest that SN cousins may provide a novel route for increasing SN Ia distance precision and understanding the role of host-galaxy and group properties on SN Ia distance measurements.
Read the full abstract

All authors Allison Blum, Erik R. Peterson, David O. Jones, R. Brent Tully, Elijah G. Marlin, Yukei S. Murakami, Jack W. Tweddle, Dillon Brout, Mitchell Dixon, Stephen J. Smartt

Machine Learning

Rolling Day-Wise Mortality Prediction in Critically Ill Patients With AKI on CRRT Utilizing Machine Pressure Waveforms

Critically ill patients with acute kidney injury (AKI) on continuous renal replacement therapy (CRRT) face high mortality, yet current risk assessment relies primarily on clinical parameters from electronic health records (EHR) and ignores minute-level circuit pressure waveforms generated by CRRT machines that track the extracorporeal circuit's interaction with the patient. Clinicians therefore cannot see deterioration as it develops. Risk is reassessed only when labs are drawn, while this continuous record is discarded because it is contaminated by shared-device records, non-physiological minutes, and sensor artifacts. To make the stream usable, we aligned machine records to charted therapy intervals to prevent cross-patient leakage, removed priming and downtime minutes, tuned denoising on a synthetic spike-injection benchmark, and masked unobserved intervals rather than imputing them. On this cleaned stream, we define a rolling day-wise task and a transformer-based stacked ensemble that late-fuses a window-reduced sequence transformer with classical models using circuit-instability features and clinical EHR variables. In a leak-safe benchmark on the multi-center CRRTnet cohort (976 patients, 4,585 treatment days), the machine-only model had the lowest standalone prognostic value (AUROC 0.625), followed by the EHR-only model (0.717). Integrating EHR and machine streams reached a one-day mortality AUROC of 0.766. SHAP attribution showed that circuit-instability descriptors raised the machine share of the top 15 combined-model features from 3 to 7 (20.0% to 46.7%), highlighting filter pressure, transmembrane pressure (TMP), and access-to-return difference (ARD). To our knowledge, this is the first patient-level mortality prediction incorporating CRRT machine data, turning a discarded bedside stream into a continuous risk signal.
Read the full abstract

All authors Shehan Irteza Pranto, Joanna Yang, Joshua Lambert, Stuart L. Goldstein, Lili Chan, Girish N. Nadkarni, Tiago K. Colicchio, Javier A. Neyra, Jin Chen

Logic in Computer Science

Positive Topology and Feasible Refinement: Forcing Matrices, Positivity, and Information

We develop a conceptual and operational account of Positive Topology starting from a basic relation between points or models and observable properties. From this relation, two complementary structures emerge. The first captures universal refinement and cover: what must hold across all relevant cases and how information can be systematically refined. The second captures positivity and witnessed existence: what can be positively realized and sustained without relying on classical complements. A central result shows that the underlying relation between points and observables can be reconstructed from either of these induced structures. We also clarify the distinction between point-based and pointfree formulations: when points are available, positivity can be derived from the underlying forcing relation, while in the formal pointfree setting positivity is taken as primitive and its compatibility with cover is imposed axiomatically. We develop two complementary interpretations of the framework. The first is information-theoretic, viewing cover as refinement of partial information and positivity as witnessed feasibility. The second is game-theoretic, viewing positivity as the ability of a witness or hypothesis to survive successive legitimate refinements. The final part of the paper is deliberately programmatic. We outline how resource constraints, verification costs, and finite budgets can be incorporated into the framework. This leads to resource-sensitive notions of forcing, cover, positivity, and refinement, and raises new questions about how these structures behave as available resources change. Examples from medical diagnosis, legal reasoning, and AI systems illustrate the potential relevance of the approach to grounded, explainable, and resource-aware inference.
Read the full abstract

Systems and Control

A Duality Reformulation of the Companion-Matrix Lyapunov Problem

We study the relation between positive semidefiniteness and entrywise nonnegativity for solutions of continuous-time Lyapunov equations. For a real Hurwitz matrix $A$, we show that the solution of $AP+PA^\top=-Q$ is positive semidefinite for every symmetric entrywise nonnegative $Q$ if and only if the solution of $A^\top X+XA=-R$ is entrywise nonnegative for every positive semidefinite $R$. This equivalence follows from the adjointness of the two solution operators and extends to all real unmixed matrices. We then prove that both properties hold for every real Hurwitz companion matrix, settling a conjecture previously established under the additional assumption of a real spectrum. The proof combines a controllability-Gramian normalization with a pairwise positivity theorem for the coefficients of the adjugate polynomial of a real accretive matrix. The latter is obtained from a coefficient-sign property of bivariate polynomials, an auxiliary determinant that does not vanish on the product of two open right half-planes, and a rank-one perturbation argument. Covariance and energy interpretations connect these results with dissipative realizations, damped second-order systems, and comparisons between input Gramians.
Read the full abstract

Quantum Physics

Energy Is Not Enough: Leakage-Aware Ansatz Criterion for Variational Quantum Eigensolvers

In molecular variational quantum eigensolver (VQE) calculations for a fixed electron-number sector, the projected-sector energy alone does not certify that a prepared state is physically valid. We show that a state can have zero projected-sector error while almost all of its probability mass lies outside the target particle-number sector. We therefore evaluate ansatz families using three quantities: projected-sector error, leakage outside the physical sector and parameter count. Our noiseless experiments show that particle-preserving UCC-style families eliminate leakage induced by ansatze, but do not guarantee ground-state accuracy. In the parameter-budget comparison experiments, the one-shot ranking rules select identical generator sequences and yield identical errors. The adaptive hybrid procedure achieves lower mean projected-sector errors at selected budgets. Using established Kraus invariant-subspace criteria, we also express channel-induced leakage through a positive operator and discuss global and local depolarizing noise analytically. This gives a practical diagnostic framework for comparing ansatze beyond energy alone.
Read the full abstract

Computation and Language

From Token Probabilities to Semantic Constraints: Towards Declarative Probabilistic Evaluation of Language Models

While Large Language Models have improved rapidly, many fundamental questions remain about how to evaluate the knowledge and reasoning abilities they acquire, and how such evaluations relate to the learning signals used in pre-training. In this paper, we propose ModelLog, a declarative probabilistic framework for pre-training evaluation that makes the semantic structure of model behavior explicit and provides new formal tools for relating evaluation to learning. ModelLog specifies evaluation targets as symbolic constraints over token-level predictions and measures how strongly a model's distribution satisfies those constraints. We explore the framework through a new suite of tasks targeting negation, mutual exclusivity, and consistency, finding systematic failures that are difficult to characterize through token likelihood or answer accuracy alone. We further show that these evaluation scores can also be interpreted as losses, whose gradients reflect logical strength, informativeness, and variable-level sensitivity. This links evaluation and learning through a shared semantics, suggesting evaluation methods that diagnose model behavior while also helping to clarify the semantic structure of learning.
Read the full abstract

All authors Kyle Richardson, Cullen Anderson, Pranav Balakrishnan, Takuto Ban, Daksha Ladia, Ankita Gupta, Marisa Hudspeth

Artificial Intelligence

Fraglingo: Molecular Design via Attachment-Aware Autoregressive Fragment Generation

We introduce Fraglingo, an autoregressive molecular generator that constructs molecules step by step from chemically meaningful fragments connected through predefined attachment sites. At each generation step, Fraglingo jointly predicts which fragment to add and how it should attach by producing an attachment-aware fragment embedding and retrieving the nearest fragment through latent-space search. A wildcard-anchored readout represents both the growing molecule and candidate fragments relative to their attachment sites, enabling a single latent prediction to determine both fragment identity and attachment configuration. Because prediction operates in a continuous embedding space rather than over fixed fragment identifiers, larger fragment libraries can be introduced at inference time without retraining. This retrieval-based formulation provides a unified generation primitive for molecule generation, scaffold generation, scaffold decoration, and molecule optimization. Fraglingo also supports property-conditional generation, allowing desired molecular properties to guide the generation process. On controlled property-conditional benchmarks, Fraglingo achieves stronger joint property control than comparably trained baselines while maintaining competitive validity, uniqueness, and novelty. It also generalizes to fragment libraries up to four times larger than those used during training without retraining. Code is available at: https://anonymous.4open.science/r/FragLingo-3551.
Read the full abstract

All authors Thao Nguyen, Jeonghwan Kim, Zhenhailong Wang, Heng Ji

Machine Learning

GeoTTER: Leveraging Local Geometry of Optimal Transport for Zero-Shot Classification

We present GeoTTER, a novel framework that redefines optimal transport in the realm of zero-shot classification. Conventional methods often suffer from miscalibration and a lack of adaptability, as they rely on fixed cost matrices derived solely from pre-trained model embeddings. In contrast, GeoTTER addresses these limitations by incorporating two key techniques. First, to alleviate high-frequency label jaggedness (sample-level manifold jitter that assigns neighboring embeddings to different classes), GeoTTER integrates local geometric structure into the optimal transport formulation via graph-Laplacian smoothing, a technique grounded in spectral graph theory that enforces neighborhood consistency. Second, to correct coherent angular drift (a low-frequency orientation bias in which large groups of samples share the same angular offset from their true label prototypes), we fuse clustering-guided cost components with a globally adjusted transport cost, achieving a multi-objective optimization that respects both global distribution constraints and latent data structure. With a median improvement of +6.82% compared to zero-shot and +2.13% compared to OTTER, GeoTTER shows robust improvements across a diverse set of benchmarks.
Read the full abstract

Robotics

Constraint-Grounded Reinforcement Learning for Variable Impedance Control in Contact-Rich Robotic Insertion

In robotic insertion under uncertain contact, the axial force limit and the appropriate controller gain vary across tasks. As a result, a single fixed gain is unlikely to remain suitable across different task conditions, making conventional impedance controllers reliant on manual retuning. To eliminate manual retuning, we propose Constraint-Grounded Reinforcement Learning (CG-RL), a variable impedance framework for online gain adaptation. Conditioned on the force limit and contact feedback, the policy outputs a residual motion, an insertion rate, and a requested gain. The controller projects this gain into the admissible range without exposing the range itself to the policy. This separation allows a single policy to operate under different force limits without retraining or manual retuning. We evaluate CG-RL on simulated oblique insertion across five training seeds. CG-RL achieves an $85.8\pm7.7\%$ (mean $\pm$ SD) success rate of insertions without violating the force limit, while keeping the applied gain within the admissible range. As a comparison, a fixed-gain baseline using the midpoint gain achieves a success rate of $50.1\%$. The policy adapts its insertion rate continuously to the specified force limit and further generalizes to more permissive force limits above the training range. In contrast, the same actor without force-limit input does not exhibit this adaptation. The applied gain is guaranteed to remain within the admissible range, while force-limit satisfaction is validated empirically rather than guaranteed formally.
Read the full abstract

Atmospheric and Oceanic Physics

Bay Assessment Model: A Python-based Salinity Projection Tool for Florida Bay, USA

Extreme salinity events in Florida Bay are a central concern of the Comprehensive Everglades Restoration Program (CERP). A large-scale seagrass die-off in the late 1980s, driven by hypersaline conditions following reduced freshwater delivery, made reliable salinity prediction essential to restoration planning. The Bay Assessment Model (BAM) is a hydrologic model that simulates salinity across 54 idealized basins representing Florida Bay. It is mass-conservative apart from where required for model stability. Interbasin fluxes are computed from hydraulic gradients across shoals, and each basin receives direct rainfall and evapotranspiration forcing. Along the shoreward boundary BAM is driven by observed water levels from the Everglades Depth Estimation Network (EDEN), or by output from upstream restoration planning models. Along the marine margins, tidal boundary conditions from NOAA subordinate station harmonic constituents are superimposed on a regional mean sea level anomaly. BAM is implemented entirely in Python and is open source. Over the extended 1999-2026 record the domain-wide mean water level bias is -0.024 m with a mean RMSE of 0.092 m, and the domain-wide mean salinity bias is 0.00 ppt with a mean RMSE of 6.21 ppt. Model bias varies systematically with hydrologic regime, increasing in magnitude from drought to wet conditions and reversing sign at one central bay basin, constraining the interpretation of anomaly-based metrics. Applying uniform offsets at the Everglades shoreward boundary yields salinity responses of 2-4 ppt in the northeastern bay, attenuating toward the marine margins; a fuller treatment of restoration and sea level scenarios is reported separately. BAM provides a transparent, efficient, and community-accessible tool for evaluating the salinity consequences of Everglades restoration actions in Florida Bay.
Read the full abstract