Materials Science
Large-Area SnS Crystals by Controlled Sn-S Chemical Vapor Deposition: Growth Optimization, Morphology, and Raman Characterization
Bilal Ahmed
Tin monosulfide (SnS) is a layered IV-VI
semiconductor with strong in-plane anisotropy and promising properties for optoelectronic and ferroic applications. However, obtaining large, continuous, and morphologically well-defined SnS
crystals remains challenging. Here, we report the growth of large-area SnS
crystals using chemical vapor deposition with separate elemental Sn and S precursors. By controlling the precursor temperatures, source positions, substrate temperature, carrier-gas flow, and hydrogen flow, we obtain SnS
crystals with lateral dimensions approaching 350 micrometers. The optimized growth occurs with the Sn source at approximately 750 degrees C, the S source at approximately 230 degrees C, and the substrate maintained near 650-680 degrees C, using Ar/H2 flow rates of 120/10 sccm. X-ray diffraction shows a dominant SnS (111) reflection with no detectable secondary crystalline tin-sulfide phase within the measurement sensitivity. Raman spectroscopy reveals six characteristic SnS
phonon modes at approximately 39.3, 48.8, 95.2, 165.0, 193.1, and 219.1 cm^-1. In particular, the low-frequency modes near 39 and 49 cm^-1 are clearly resolved with FWHM values of approximately 2.52 and 2.37 cm^-1, respectively. SEM and AFM measurements further show large continuous
crystals and a relatively smooth nanoscale surface. Growth on mica produces more regular square and rectangular flakes, with lateral dimensions approaching 300 x 300 micrometers, while
crystals grown on SiO2/Si are more frequently fractured or irregular at their edges. These results demonstrate a practical growth window for obtaining large-area SnS
crystals suitable for further optical, electrical, and device studies.
Read the full abstract ▾
Machine Learning
Beyond Static Graph World Models: Learning Stochastic Latent Dynamics over Evolving Topologies
Alex Schutz, Nick Hawes and Victor-Alexandru Darvariu
Graph-based world models have recently emerged as a means of learning transitions over relational state representations. However, existing approaches are largely limited to fixed-
topology graphs or deterministic, fully observable environments. We propose the Graph Dynamics Model (GDM), a world model for graph-structured observations that is designed to handle the more general setting of evolving
topologies in
stochastic and partially observable environments. The GDM uses a sparse recurrent adjacency
matrix to model
topology updates and perform message passing, together with a recurrent state-space architecture for modelling
stochastic transitions. Furthermore, we identify a gap in the evaluation of graph-based world models, as existing methods do not provide a means of comparing predicted and true distributions over the joint graph state comprising the interdependent
topology, node features, and graph features. We therefore introduce the Graph Distribution Distance (GDD) metric, which uses maximum mean discrepancy with a graph kernel to comprehensively compare joint next-state distributions. We evaluate the GDM across several environments, including
stochastic and partially observable settings. We demonstrate that GDM outperforms baseline models and displays zero-shot generalisation on large graphs.
Read the full abstract ▾
Logic
Combinatorics of translations of meager and closed measure zero sets
Aleksander Cieślak
We study combinatorial properties of translations of meager and closed sets of measure zero. We discuss constellations of Borel
conjectures for related classes of small sets and show that there are no uncountable null-additive sets in the Miller model. We also study cardinal invariants of those classes. In particular, we investigate the cardinal invariants of $\sigma$-ideals $\mathcal{H}_{F}$ related to meager-additive sets. We obtain a new characterization of additivity of meager ideal and answer some questions of Cardona. We also show that $\mathrm{non}(\mathcal{E}^{*})$, that is equal to the translation version of the covering number of the ideal $\mathcal{E}$, is close to the cardinal invariant related to the Laver property. This strengthens a result of Bartoszyński and Judah and a result of Elekes and Steprāns. Finally, we show that every $\mathcal{E}$-Luzin set is in $\mathcal{E}^{*}$.
Read the full abstract ▾
High Energy Astrophysical Phenomena
Composition-dependent acceleration of UHE Cosmic Rays in AGN electric fields
E. Koutsoumpou, I. Contopoulos, A. Nathanail et al. (+2 more)
The origin of ultra-high-energy
cosmic rays (UHECRs) remains elusive, as proposed acceleration mechanisms are rarely connected directly to composition-sensitive air-shower observables. We establish such a connection for voltage-drop acceleration in near-horizon current sheets of magnetically arrested disks around supermassive black holes. Fragmentation is represented by a log-normal distribution of effective potential-sampling factors $\eta$, with $\mu_\eta=\langle\ln\eta\rangle$, $\sigma_\eta=[\mathrm{Var}(\ln\eta)]^{1/2}$, and median $\widetilde{\eta}=e^{\mu_\eta}$, while the energy gain scales with
nuclear charge $Z$ and maximum potential $\mathcal{V}_{\max}$. We propagate the accelerated particles through Bethe–Heitler and adiabatic losses,
proton photopion attenuation, and a mass-evolving photodisintegration cascade, and predict $\langle X_{\max}\rangle$ and $\sigma(X_{\max})$. We perform
Bayesian fits to the Pierre Auger Observatory 2025 DNN measurements for two fixed injection templates and propagation distances of 10, 30, and 100 Mpc. An Auger-inspired,
proton-poor injection reproduces both shower moments at 10 and 30 Mpc, with $\chi^2/\nu=0.33$ and $0.47$. At 30 Mpc, $\sigma_\eta=0.59^{+0.13}_{-0.10}$, corresponding to a multiplicative dispersion of approximately 1.8. The median sampling factor and maximum potential remain strongly degenerate, with $\widetilde{\eta}\mathcal{V}_{\max}$ more directly constrained. A metal-enriched injection containing $\sim53\%$ hydrogen gives $\chi^2$ larger by 17.9 and 25.9 at 10 and 30 Mpc. Both templates remain viable at 10 and 30 Mpc but fit poorly at 100 Mpc. Within the scenarios tested, the results favour a relatively
proton-poor accelerated population and show that the shower-width
evolution can constrain the
stochastic structure of electric-field acceleration in nearby AGN.
Read the full abstract ▾
All authors E. Koutsoumpou, I. Contopoulos, A. Nathanail, A. Loules, D. Kazanas
Solar and Stellar Astrophysics
The physics of deep-interior convective shell boundaries in a massive pre-supernova star: Internal gravity wave mixing at the $N^2$ peak
Praneet Pathak, Falk Herwig, Simon Blouin et al. (+2 more)
The physics of convective boundary mixing involves concepts such as penetration and overshooting and has been mostly characterized for envelope and core convection. For the burning shells deep inside massive
stars in their final stages, the convective boundary remains poorly understood. 1D
stellar-
evolution codes typically place a steep
entropy gradient and the corresponding narrow \brunt{} ($N^2$) peak at a lower convective shell boundary. Based on high-resolution 3D hydrodynamic simulations of a $25 \rm{M_\odot}$
star 11 days before core collapse, we find that convection excites high-frequency internal
gravity waves at the $N^2$ peak representing the C-shell convection bottom. These waves efficiently mix
entropy and species and the $N^2$ peak erodes by ${\sim}40\%$ within hours, correspondingly reducing the
entropy gradient. The wave-driven
diffusion coefficient is $D\approx 5\text{--}6\times 10^{9} \rm{cm}^2 \rm{second}^{-1}$ at the $N^2$ peak in numerically converged simulations. Our simulations show that the dominant physics at this boundary is mixing due to internal
gravity waves. Extrapolated with the time-dependent $D$ coefficient from the 3D simulations, the
entropy step across the peak decreases by 40–85% within half a day, thereby reducing the density, temperature and
entropy gradients that set the
accretion-rate drop a stalled
supernova shock encounters. This physics is not included in present-day 1D models and enters the structural parameters that predict explodability. Flattening the gradients days before collapse favors an O–C shell merger, which could reshape pre-collapse nucleosynthesis. Without this physics 1D models carry too sharp gradients that introduce previously unaccounted systematic errors into the explodability surveys, yield tables and chemical-
evolution studies built on them.
Read the full abstract ▾
All authors Praneet Pathak, Falk Herwig, Simon Blouin, Paul R. Woodward, Joshua Issa
Differential Geometry
An upper bound for the mean Lorentz force of magnetic flows without conjugate points in terms of the geodesic curvature of horocycles
Anthony J. García and Rafael O. Ruggiero
We study
magnetic flows on closed Riemannian surfaces of genus at least two. Let $(M,g)$ be a surface without focal points and let $\Omega$ be a
magnetic field. We prove that, above the Mañé critical value, the absence of conjugate points imposes
geometric restrictions on the
magnetic flow. First, we obtain a bound for the integral of the Lorentz force in terms of the geodesic
curvature of the horocycles of the underlying Riemannian metric. As a consequence, a lower bound on the Gaussian
curvature yields an explicit bound in terms of the area of the surface. The proofs combine the global geometry of
magnetic geodesics with Liouville's formula for the geodesic
curvature in the orthogonal coordinates given by the level sets and geodesics of a Busemann function on the universal cover.
Read the full abstract ▾
Machine Learning
OPDiv: Optimal Selection of Top-K High-Scoring, Diverse Compounds
Miroslav Lžičař
A virtual screening campaign may produce thousands of promising candidates, but only a small number can be purchased, synthesized, or tested. The practical question is how to select a set of compounds that both rank well and are diverse enough: this poses a genuine tradeoff, where selecting the highest-scoring molecules yields limited diversity, while diversity selection sacrifices some well-scoring molecules. We introduce OPDiv, a diversity selection and evaluation
algorithm solving this tradeoff by finding an optimal subset of molecules using integer optimization. We demonstrate the selection
algorithm in practice with fingerprint distance, shape and electrostatic diversity and compare the resulting diversity spectra. We argue that virtual screening is not merely a ranking problem, but also an implicit constrained optimization task: when redundant chemotypes are undesirable, pipelines should be compared based on the top-k compound selections satisfying the desired diversity constraints. OPDiv makes it possible to find the optimal compound set under a given diversity threshold efficiently and serves as a fair
benchmark of the best diverse selection achievable by a given structure-based or ligand-based virtual screening pipeline, molecular search or generative model.
Read the full abstract ▾
Dynamical Systems
Repulsive Mean-Field Couplings and Self-Consistent Transfer Operators
Roberto Castorrini, Stefano Galatolo and Matteo Tanzi
We study a one-dimensional self-consistent transfer operator generated by a repulsive mean-field coupling. The interaction admits a natural interpretation as a myopic relocation dynamics balancing congestion and uniformly distributed resources. Using the
quantile representation of probability measures, combined with self-consistent transfer operators methods, we prove the existence and uniqueness of a physically relevant invariant measure for the system and characterize it explicitly through a nonlinear fixed-point equation. We then show that the corresponding finite-particle
equilibrium converges to the continuum
equilibrium with optimal order in Wasserstein distance and establish a quantitative thermodynamic limit.
Read the full abstract ▾
Accelerator Physics
Vector Potential-based Beam Current Sensors: Applications for Mu2e, and Picosecond Bunch Substructure
P. W. Gorham, M. T. Hedges, R. Kim et al. (+8 more)
We report on a new beam current monitor system based on work originally done to develop timing planes for future
collider detectors. In contrast to sensors based on wall currents or toroid inductance, we make use of electric field induced by the electromagnetic vector potential of the beam current as it passes near or through a transmission line (TL), arranged with its dominant TEM or TE10 fields aligned with the current, thus directly coupling to TL modes. For resonantly extracted
proton pulses currently being commissioned at the Fermilab Delivery Ring, we utilized a TEM parallel-plate TL geometry, since the requirements included measuring the structure of the bunch over $\sim{}200~\rm{ns}$ time scales. We present the design and results of simulated and realized sensors now deployed at Fermilab for future use as pulse-to-pulse
proton intensity monitors at the Mu2e experiment. In addition, inspired by this and prior work, we investigate the use of broadband mm-wave dual-ridge waveguides as a means of measuring substructure of nanosecond bunches down to the tens of picosecond level for future use at the Long Baseline
Neutrino Facility (LBNF).
Read the full abstract ▾
All authors P. W. Gorham, M. T. Hedges, R. Kim, P. M. Lewis, K. Lynch, C. Miki, J. Morgan, N. Nyza, D. Pang, T. Polischuk, D. Still
Quantum Physics
Universal holonomic control of algebras of observables
Paolo Zanardi
We extend
geometric parallel transport from states and degenerate eigenspaces to observable
algebras defining generalized
quantum subsystems. At fixed block structure,
unitary invariance fixes a unique connection, recovering Berry–Wilczek–Zee transport on bands. On positive-dimensional strata, its holonomy is the special-
unitary part of the normalizer: all inner logical gates and equal-type sector permutations are attainable. Given a compatible gapped
Hamiltonian and access to the required paths, inner gates can be implemented adiabatically by isospectral control of a band refinement.
Read the full abstract ▾
Instrumentation and Detectors
Cross-geometry transfer and model collapse in point cloud calorimeter shower generation
Thorsten Buss, Frank Gaede, Gregor Kasieczka et al. (+2 more)
Particle shower simulation is a major computational cost in high-energy physics. Monte Carlo methods such as Geant4 are accurate but expensive, while most machine learning surrogates are tied to specific detector geometries and require retraining for each design change. We study cross-geometry transfer learning with CaloClouds II, a generative model that produces point clouds rather than voxels and can project onto arbitrary detector readouts. We pre-train on
photon showers in the International Large Detector (ILD) and adapt to electron showers in the cylindrical CaloChallenge
Dataset 3. With only 100 target showers, fine-tuning reduces the
geometric mean Wasserstein distance to Geant4 by about 51% over training from scratch. Bias-only fine-tuning (BitFit) stays within 5% of full fine-tuning while updating only 17% of the
diffusion network parameters. We also examine model collapse in CaloClouds II by retraining its normalising flow and
diffusion model on its own generated showers across successive generations.
Read the full abstract ▾
All authors Thorsten Buss, Frank Gaede, Gregor Kasieczka, Lorenzo Valente, Duncan Weber
Robotics
Morphometric Imitation: From Morphology and Contact Aware Hand Retargeting to Sim-to-Real Visuomotor Policy
Tara Sadjadpour, Siming He, C.K. Wolfe et al. (+5 more)
Human hand-object interactions (HOIs) provide a rich source of demonstrations for dexterous manipulation, but learning directly from them presents challenges in bridging morphology gaps, ensuring dynamical feasibility, and sim-to-real deployment. We present Morphometric Imitation, a three-stage framework that transforms reconstructed HOIs into zero-shot sim-to-real visuomotor policies. First, morphometric optimization (MMO) kinematically retargets human motion across hand morphologies while preserving demonstrated contacts. Second, residual reinforcement learning (RL) refines the kinematic reference using object pose and contact information from the human motion to produce dynamically feasible
robot demonstrations. Third, these demonstrations are distilled into visuomotor policies. Across three
robot hands and ten HOIs, MMO outperforms five baselines in contact F1, improving on the strongest ones by 8 to 28 points, while improving the success rate of downstream dynamic retargeting by as much as 35 points. On a Sharpa hand, the visuomotor policies achieve 89.3% zero-shot success in 300 real-world trials on 30 objects spanning 10 categories. Project page: https://morphometricimitation.github.io
Read the full abstract ▾
All authors Tara Sadjadpour, Siming He, C.K. Wolfe, Haozhi Qi, Lea Wilken, S. Shankar Sastry, Claire Tomlin, Jitendra Malik
Cryptography and Security
"What I See is What I Hear": Deepfake Detection Across Diverse Hearing Abilities
Magdalena Pasternak, Malvika Jadhav, Palavi V. Bhole et al. (+8 more)
The proliferation of audiovisual deepfakes has lowered the cost of fraud, impersonation, and misinformation, but their success ultimately depends on human perception. Detection requires integrating auditory and visual cues, yet security and
privacy research has largely overlooked d/Deaf and hard-of-hearing (DHH) populations. We address this gap with an in-person, mixed-methods study of 80 participants: 31 hearing persons (HPs), 15 hard-of-hearing (HoH) participants, 17 d/Deaf participants, and 17 cochlear implant (CI) users. Each participant judged the authenticity of 30 clips, where manipulations spanned text-to-
speech, voice conversion, lip-sync, or face-swap. DHH participants were less accurate than HPs overall (76.4% vs. 88.0%, p<.001), primarily because they more often classified authentic clips as manipulated (FPR: 29.7% vs. 11.2%). Differences depended strongly on the manipulated channel. For audio-only manipulations, HoH participants matched HPs (90.0% vs. 90.3%), followed by CI users (79.4%) and d/Deaf participants (41.2%). When clips contained an audiovisual manipulation, accuracy clustered between 84% and 87%, although performance still varied by manipulation method. Our work systematically characterizes how deepfakes affect DHH populations, highlighting the asymmetric risks audiovisual manipulations may pose to groups with different hearing abilities and the need for accessible, tailored defenses that support all users.
Read the full abstract ▾
All authors Magdalena Pasternak, Malvika Jadhav, Palavi V. Bhole, Aviva Smith, Elaina Trapatsos, Vincent Bindschaedler, Roshan Peiris, Ersin Uzun, Patrick Traynor, Matthew Wright, Kevin R. B. Butler
Fluid Dynamics
General model for shear stress and surface pressure distributions for flow over spheres
Bryce Daniels and Thomas Schwartzentruber
A general physics-based model for the distribution of shear stress and surface pressure is developed for laminar flow over spheres. Although integral coefficients have been studied extensively for spheres, they do not capture the local surface distributions that govern phenomena such as heat transfer, phase change, and mass loss in multiphase and particle-laden flows. The model depends on the flow regime of the sphere ($Re_\infty$, $M_\infty$, $Kn_\infty$), local gas properties ($\gamma$, $\omega$, $Pr$), and surface conditions ($T_w/T_0$, $\sigma_t$, $\sigma_n$). The formulation incorporates boundary-layer scaling, flow separation, shock-wave physics, velocity-slip effects, and high-speed rarefaction, while recovering the known asymptotic behavior in the creeping-flow, hypersonic, and free-molecular limits. The proposed model is validated in the continuum and rarefied regimes using CFD and DSMC simulations from literature. In the continuum regime, it is compared with CFD simulations of flow over a sphere in air under subsonic, supersonic, and hypersonic conditions with varying surface temperatures. The model captures the CFD distributions very well over a wide range of $Re_\infty$, $M_\infty$, and $T_w/T_0$. In the rarefied regime, it is compared with DSMC simulations of flow over a sphere in a monatomic gas under subsonic and supersonic conditions with fully diffuse and partial surface accommodation. It shows good agreement with the simulated shear stress and surface pressure distributions over a wide range of $M_\infty$ and $Kn_\infty$, while accurately capturing the effects of surface accommodation around spherical bodies.
Read the full abstract ▾
Quantum Physics
Quantum computational advantage in random-circuit sampling on IBM superconducting quantum computers
Tigran Sedrakyan, Yuxuan Zhang, Hovnatan Karapetyan et al. (+3 more)
We report forward random-
circuit sampling (RCS) on the 120-
qubit Nighthawk r2 superconducting processor (ibm_phoenix) with square-
lattice connectivity, using 61
qubits, native CZ gates, and the standard cloud execution stack with no
benchmark-specific calibration. Two independent fidelity
estimators—mirror benchmarking and three- and four-patch cross-
entropy benchmarking (XEB)—agree with each other at every measured depth, the mirror from 4 to 40 cycles and the patched
estimators from 20 to 40 cycles, across more than two orders of magnitude of fidelity decay, and exceed the first-generation Nighthawk r1 device by more than an order of magnitude at fixed depth. The 36-cycle
circuits sit at the depth where tensor-network contraction cost saturates at system size: a contraction-cost
estimator validated against the published Sycamore and Zuchongzhi networks places the single-amplitude cost at $\sim$$10^{22}$ complex operations. At $F_{\mathrm{XEB}}(36)=2.3\times10^{-3}$ under favorable
memory assumptions this implies $1.2\times10^{27}$ machine operations within the bounded-fidelity rejection-sampling model — more than a century of runtime on the Frontier supercomputer — to collect a $10^{6}$-sample ensemble, which takes only 19\,s on Nighthawk r2. To our knowledge, this is the first demonstration of
quantum advantage for a vanilla random-
circuit sampling on a commercially and broadly accessible
quantum processor that most non-expert
quantum computer users can easily replicate.
Read the full abstract ▾
All authors Tigran Sedrakyan, Yuxuan Zhang, Hovnatan Karapetyan, Joshua D. Baktay, Hrant Gharibyan, Hayk Tepanyan
Solar and Stellar Astrophysics
3D Radiation Hydrodynamics of Blue Supergiant Envelopes
Linhao Ma, Lars Bildsten and Yan-Fei Jiang
We present the first 3D radiation hydrodynamical simulation of the outer envelope of a blue supergiant. These near Eddington limited massive
stars in the Hertzsprung gap exhibit substantial photometric variability and have spectroscopic line widths indicating unexplained macroturbulent velocity fields. Our Athena++ simulations find that the convection associated with the opacity peak from iron directly extends to the photosphere, as found in previous 3D simulations of massive main sequence
stars. Unlike 1D
stellar models, this single convecting region imprints a velocity field that is consistent with spectroscopic models of macroturbulence. In addition, the convection in the outermost layer leads to such large density variations that the exiting radiation varies on large ($\lesssim 10 \% $) amplitudes and short ($\sim$ days) timescales. This leads to
stochastic low frequency (SLF) photometric variability, with amplitudes and power spectra broadly consistent with observations.
Read the full abstract ▾
Statistical Mechanics
Bayesian Monitoring of a Diffusive Particle in One Dimension
Federico Gerbino, Guido Giachetti, Pierre Le Doussal et al. (+1 more)
We study the
Bayesian monitoring of a single particle diffusing along a line, while an observer tracks it continuously using noisy measurements at each spatial point. The average Shannon
entropy $\overline{S(t)}$ associated with the posterior distribution of the particle's position quantifies the uncertainty that remains after conditioning on the sequence of measurements. We map the problem to the calculation of the moments of the partition functions of directed
polymers in $1 + 1$ dimensions. For a constant monitoring rate, the
entropy remains bounded for any finite measurement intensity and can be quantified using exact results from the Kardar-Parisi-Zhang equation, providing both the saturation value and the asymptotic behavior for long times. For a monitoring intensity that decreases according to a power law $\sim t^{-\alpha}$, two distinct asymptotic regimes emerge: $\overline{S(t)} \simeq \tfrac12 \ln t$ for $\alpha > 1/2$ and $\overline{S(t)} \simeq \alpha \ln t$ for $0 < \alpha < 1/2$. Both results are obtained using the replica method, which maps the problem onto an attractive Lieb-Liniger
Hamiltonian with a time-dependent coupling: the two regimes can be understood from a perturbative expansion around free
diffusion and around the attractive Lieb-Liniger ground state, respectively. We discuss the limiting case $\alpha = 1/2$ and compare the predictions with simulations of discrete Gaussian and log-gamma
polymer models.
Read the full abstract ▾
All authors Federico Gerbino, Guido Giachetti, Pierre Le Doussal, Andrea De Luca
Artificial Intelligence
Training Object Permanence in World Models
Haotian Zhang, Fengyuan Yu, Dezhi Luo et al. (+28 more)
Object permanence and solidity are hallmarks of human cognitive priors. Recent studies show that
video generation models, a paradigmatic class of current world models, have begun to show emerged
reasoning abilities, making them ideal candidates for building human-like physical intelligence. Do
video models have emerged object permanence in them? If not, could we train them with a core-cognition inspired
dataset? We introduce WROP (World
Reasoning with Object Permanence), a data infrastructure of 150 hand-designed cognitive science inspired tasks, divided into six cognitive categories. We build Blender generators that randomize speed, lighting, camera angle, and other nuisance parameters while preserving each task's cognitive structure, yielding 10,000+ samples per task. We release a 1.5M-sample training corpus and a 300-question exam. On this exam we evaluate 14
video models: 3 reference-to-
video, 7 edit, and 4 continuation, among which PWM-WROP, our 16B world model. In a blind pairwise Elo study, PWM-WROP ranks first among continuation models and third overall, behind only a statistical tie between two reference-to-
video models. We release the data, exam, model answers, scores, weights, and PWM, our native-PyTorch training stack on AWS Trainium2.
Read the full abstract ▾
All authors Haotian Zhang, Fengyuan Yu, Dezhi Luo, Haoran Sun, Zehong Zhao, Qingying Gao, Yihan Li, Siyuan An, Huayi Qin, Yilan Zhang, Zhengze Jiang, Pinyuan Feng, Renrui Zhang, Ziyu Guo, Letian Wang, Mengyue Yang, Kangfu Mei, Maijunxian Wang, Ran Ji, Vikash Kumar, Freda Shi, Chandra Sripada, Vincent C. Muller, Philip Torr, Alan Yuille, Nikolaus Kriegeskorte, Felix Juefei-Xu, Lvmin Zhang, Jieneng Chen, Yilun Du, Hokin Deng
Information Retrieval
The Fellowship of the Query: Learning Retrieval Actions
Mohammed Al-Maamari, Saber Zerhoudi, Michael Granitzer et al. (+1 more)
Retrieval-augmented question answering requires control decisions about when to decompose a question, search, reformulate, extract evidence, synthesize facts, verify progress, and stop. We study whether trajectory fine-tuning can improve small
language models (SLMs) as next-action controllers. We additionally evaluate a low-resource setting in which a single SLM serves as both the controller and the final-answer generator. From accepted teacher search traces, we build a seven-way action-prediction task, where the model predicts the next structured teacher action from the current trajectory state, and evaluate LoRA-supervised fine-tuning across SLMs and xSLMs as controllers. On 1,646 held-out action examples, Granite 4.1 3B trained on 13,194 actions reaches macro-F1 0.6536, compared with 0.1736 for zero-shot prompting of the same model and 0.5399 for a TF-IDF logistic-regression baseline. In an end-to-end controller/generator swap evaluation over 149 held-out trajectories, using the fine-tuned model for both roles improves Exact Match from 0.7530 to 0.7946 and
token F1 from 0.7783 to 0.8295 compared with using the base model as both controller and generator. The cross-role conditions show that the fine-tuned controller increases evidence-fact recording when the generator is fixed, while controller-only final-answer gains are not statistically clear. Overall, trajectory supervision improves action prediction and evidence-recording behaviour in this evaluated pipeline. Code is available at https://github.com/padas-lab-de/
agent-action-controller
Read the full abstract ▾
All authors Mohammed Al-Maamari, Saber Zerhoudi, Michael Granitzer, Jelena Mitrović
Astrophysics of Galaxies
Mapping the Milky Way in Six Dimensions: A contiguous, homogenised, phase-space catalogue of Gaia DR3 tracers up to 250 kpc
Sutirtha Mukherjee, Amogh Srivastav and Subhabrata Majumdar
We present a comprehensive, homogenised catalogue of $32{,}552{,}876$
stellar sources with full 6D phase-space information, spanning Galactocentric distances from the inner
galaxy ($\gtrsim 5$\,kpc) to the outer
halo ($\sim\!250$\,kpc). By cross-matching
Gaia DR3 astrometry with spectrophotometric distances and line-of-sight velocities from 14 surveys (including desi, sdss-boss, apogee, lamost, galah, and ges), we achieve median fractional distance uncertainties of $\approx2.7\%$ for $d_{\rm helio}\le15$\,kpc, and $\approx29\%$ for $15<d_{\rm helio}\le250$\,kpc. The catalogue-wide median fractional radial velocity uncertainty is $\approx8.31\%$.
Gaia-anchored zero-point corrections and statistical normalisation suppress both random and systematic errors. Furthermore, we derive spectrophotometric parallaxes from joint photometric and spectroscopic data. Overlapping survey data is aggregated to replace unreliable values and supplement absent velocities, improving completeness and internal consistency. This multi-survey synthesis is externally validated against globular clusters,
dwarf galaxies, and
stellar streams, yielding precise kinematics. Notably, the sample features $517{,}123$
halo stars extending well beyond the
Gaia parallax limit ($\gtrsim\!10$\,kpc). These predominantly faint ($G > 17$) and metal-poor ($[\mathrm{Fe/H}] < -1$) tracers, including RR~Lyrae, blue horizontal-branch, and K~
giant stars, constitute the largest well-calibrated sample of outer-
halo tracers to date. This unique
dataset is ideal for multi-component mass modelling,
galactic archaeology, and dark matter searches across the Milky Way.
Read the full abstract ▾
High Energy Astrophysical Phenomena
Dynamically Stable Magnetic Fields in Relativistic Neutron Stars: A Diverse Landscape of GRMHD Equilibria
Mai Li, Vasileios Paschalidis, Antonios Tsokaros et al. (+2 more)
Neutron stars endowed with purely poloidal or purely toroidal
magnetic fields are known to be unstable, and settle into mixed poloidal-toroidal configurations. The end state of these instabilities is still poorly understood. It is unclear whether this state is unique or whether the dynamically stable configuration has
memory of the initial conditions. In this work, we perform long-term, fully general relativistic magnetohydrodynamic simulations of
equilibrium configurations of slowly rotating
neutron stars endowed with self-consistent mixed poloidal and toroidal
magnetic fields. Our initial configurations differ in their initial magnetization and
magnetic field geometry. We evolve the initial
equilibria for tens of Alfvén timescales until well after any instabilities have saturated. Employing an array of diagnostic tools from visualizations to the structure of the
plasma four-current and a vector spherical harmonic decomposition of the settled
magnetic fields, we characterize the structure of the dynamically stable, settled
magnetic equilibria. We find that the settled configurations are not unique. Instead, they exhibit diverse multipolar structures, with higher-order modes contributing substantially. Our findings demonstrate that there is
memory of the
magnetic-field initial conditions, and that there is no unique dynamically stable
magnetic-field geometry for
neutron stars. Nevertheless, we find that the angle-averaged radial profile of the
magnetic field toroidal to poloidal amplitude in the bulk of the settled configurations exhibits some degree of universality with typical values of order \(20\)–\(40\%\).
Read the full abstract ▾
All authors Mai Li, Vasileios Paschalidis, Antonios Tsokaros, Maria Mutz, Koji Uryu
Combinatorics
A parking function analog of the Schröder numbers
Lucy Martinez and Doron Zeilberger
Pattern avoidance is a central topic in the study of permutations. Parking functions provide a natural setting in which to ask analogous questions. In this paper, we study parking functions whose parking permutations avoid the patterns $1234$ and $2134$, obtaining an analog of the Schröder numbers for parking functions. To enumerate these objects, we develop an enumeration scheme that runs in
polynomial time and adapt the prefix-based framework introduced by Zeilberger in 1998, and later improved by Vatter, by instead using “suffixes.”
Read the full abstract ▾
Number Theory
Straight-line programs for the solutions of Pell's equation
Bogdan Dumitru and Mihai Prunescu
For nonsquare $d\ge 2$, we construct fixed straight-line programs for the least non-trivial solution $(X_1,Y_1)$ of $x^2-dy^2=1$, using addition, truncated subtraction, multiplication, integer division, exponentiation, and remainder. A
geometric-sum identity recovers $(X_1,Y_1)$ from the coordinate sums of an initial segment of Pell solutions, without knowing how many solutions were summed. Weighted binary encodings make these sums accessible to arithmetic computation. Counting operations with reuse of computed values, one program uses $94$ operations, with intermediate bit lengths bounded by $2^{2^{2^{O(d)}}}$. Four additional operations give a $98$-operation program with the bound $2^{2^{O(d)}}$. Hua's bound gives corresponding programs with $103$ and $107$ operations and replaces $O(d)$ by $O(\sqrt d\log d)$ in these size bounds. Once the fundamental solution is known, generating-function formulas compute the $n$-th positive solution in $21$ further operations by extracting its coordinates as base-$b$ digits. We prove that $2X_1(X_1+1)-1$ is the least integer base for which both formulas are defined and correct for every $n\ge1$.
Read the full abstract ▾
Solar and Stellar Astrophysics
External photoevaporation in the center of Tr14
Katia Gkimisi, Da Eun Kang, Dominika Itrich et al. (+11 more)
Most
stars form in massive
stellar clusters, where intense far-ultraviolet (FUV) radiation from OB
stars can accelerate the dispersal of protoplanetary disks through external photoevaporation. Quantifying the impact of this process on disk
evolution remains challenging, particularly in distant and crowded
star-forming regions. We investigate the low-mass
stellar population in the highly irradiated center of Trumpler 14 and examine how the local FUV radiation field influences circumstellar disk
evolution. We extracted
stellar spectra from VLT/MUSE observations using an enhanced background-optimized extraction method, carefully corrected for instrumental response as a function of wavelength and finally derived the
stellar parameters with the SAPSAL deep-learning framework. We combined these results with a probabilistic estimate of the local FUV field and near-infrared photometry to identify disk-bearing
stars. We identify 310 bona fide cluster members with an age of $0.66^{+0.73}_{-0.35}$ Myr and local radiation fields spanning $\log(FUV)\simeq4.2-6.0\ G_{\mathrm{0}}$. The fraction of
stars exhibiting near-infrared excess decreases from 30% at the lowest FUV fluxes to under 5% at $\log(FUV)\gtrsim5.6 \ G_{\mathrm{0}}$, while no significant correlation is found between our simplified estimate of the optical veiling and the local FUV field. The pronounced decline in the disk fraction with ambient FUV flux provides strong evidence for external photoevaporation rapidly dispersing circumstellar disks in the core of Trumpler 14. Our results confirm previous disk dispersal estimates and extend to higher values of the FUV field strength.
Read the full abstract ▾
All authors Katia Gkimisi, Da Eun Kang, Dominika Itrich, Rossella Anania, Eleonora Fiorellino, Sacha Gavino, Ralf S. Klessen, Victor F. Ksoll, Giuseppe Milazzo, Anna McLeod, Thomas Preibisch, Veronica Roccatagliata, Leonardo Testi, Claudia Toci
Strongly Correlated Electrons
Compact representation of strongly correlated Green's functions: the MOR+EC way to explore phase space
Norman Hogan, A. F. Kemper and Carlos Mejuto-Zaera
Repeated evaluation of the single-particle Green's function (GF) across parameter space is a recurring bottleneck in many-body methods, limiting the resolution at which phase boundaries may be probed and the frequency resolution of computed spectra. We introduce MOR+EC, a framework that constructs reusable reduced-order models for computing single-particle Green's functions by combining eigenvector continuation (EC) for parameter space exploration and model order reduction (MOR) for extrapolation in frequency space. Contrary to conventional parameterized MOR, we construct these reduced-order models with a parameter-independent resolvent, further reducing the number of required full-space evaluations. Benchmarking against exact diagonalization as the
impurity solver for our example case of DMFT calculations for single- and two-band models, MOR+EC reproduces the DMFT-converged
impurity GF to an average relative error below $10^{-3}$, with a median wall-time speedup of $9.5$-$92\times$. This accuracy and efficiency together resolve a fine-grained scan of the
impurity occupation as doping is tuned and produce a high-resolution phase diagram of an orbital-selective Mott transition. The reduced-order model also reproduces real- and imaginary-frequency spectra at no additional cost in full-space evaluations. This framework applies broadly to GF-based methods requiring repeated parametric evaluation and high resolution of the frequency axis.
Read the full abstract ▾
Lattice Field Theory
Sign of Wilson-Fermion Determinant using Contour-Integral Spectral Projection
Bhabani Sankar Tripathy and M. Padmanath
Key properties of a large sparse operator can often be determined by only a small, spectrally localized subset of its eigenmodes. In
lattice QCD with Wilson
fermions, the sign of the
fermion determinant is determined by the parity of the number of
eigenvalues lying on the negative real axis of the non-Hermitian Wilson-Dirac operator. Determination of this signature can be challenging, particularly in parameter regimes where near-zero modes can lead to exceptional configurations. We present a contour-integral-based spectral projection method to count the negative-real
eigenvalues of the non-Hermitian Wilson-Dirac operator $D$ that determine the sign of its determinant. The spectral projection is supplemented by an adapted deflation-accelerated solver for shifted linear systems and singular-value exclusions to reliably set spectral bounds and contour boundaries. We demonstrate the robustness of the method through systematic convergence tests, including variations of the contour size, and show its ability to resolve
eigenvalues close to the negative real axis. Beyond
fermion determinant sign evaluation, the approach provides a systematic means of isolating physically relevant
eigenvalues embedded in dense complex spectra, with potential applications to non-Hermitian
quantum systems, stability analyses, and large-scale
eigenvalue problems.
Read the full abstract ▾
Computer Vision
PePESeg3D: Perception Prior Enhances Multi-Scale Segmentation for 3D Gaussian Splatting
Sungjae Choi, Seunghee Koh and Junmo Kim
Recent advancements in 3D Gaussian Splatting (3DGS) have extended its capabilities to multi-scale
segmentation. Existing methods reconstruct a scene with Gaussian primitives and learn multi-scale
segmentation features separately, which leaves the geometry unaware of semantic structure and the feature learning dependent on incomplete mask supervision. To address these limitations, we present PePESeg3D, a novel framework that injects perception priors into a multi-scale 3D Gaussian
segmentation pipeline. To fully exploit perception priors, we integrate them not only into contrastive feature learning but also into the upstream geometry reconstruction. Specifically, PePE Reconstruction incorporates
monocular depth and mask constraints to ensure semantically coherent object structures. Building on this aligned geometry, PePE Contrastive Learning leverages dense depth-color cues and view-consistent centroid supervision to compensate for the incompleteness of multi-scale masks obtained from a 2D foundation model. Extensive experiments on the
SPIn-NeRF, LERF-Mask, and NVOS
benchmarks demonstrate that PePESeg3D achieves state-of-the-art performance in both multi-scale
segmentation and scene reconstruction, highlighting the importance of integrating perception priors into both geometry optimization and feature learning for accurate multi-scale 3D
segmentation. Our code is available at https://github.com/BeCow5X5/PePESeg3D.
Read the full abstract ▾
High Energy Physics (Theory)
Ignacio Quiroz Vargas and Merna Youssef
We construct a
holographic open
quantum system by coupling two CFTs through a double trace deformation and tracing out the bath degrees of freedom within the Schwinger-Keldysh formalism. This procedure yields an effective description of the nonunitary reduced dynamics of the remaining sector, whose linear response is encoded in its retarded Green's function. Working perturbatively in the double trace coupling, we derive closed form expressions for the quasinormal modes of several coupled geometries involving pure AdS and rotating or nonrotating BTZ black holes, considering both possible system/bath assignments. For the nonrotating configurations previously investigated in arXiv:2509.14312, our analytic expressions agree with the numerical results, while our treatment extends the analysis to configurations involving rotating BTZ black holes. Finally, moving beyond perturbation theory, we derive a relation between dissipation and transmission in the general AdS-BTZ setup.
Read the full abstract ▾
Strongly Correlated Electrons
Defect Poisoning of Quantum Spin Ice
Alaric L. Sanders, Gautam K. Naik, Jonathan N. Hallén et al. (+1 more)
The hunt for a material realization of
quantum spin ice has motivated more than two decades of experimental effort. The candidate materials inevitably contain
crystal imperfections, such as
magnetic vacancies, whose effects are often disregarded. Here, we show that experimentally relevant levels of dilution can qualitatively reshape the low-energy behavior, as nearby vacancies generate
quantum fluctuations that are absent in the clean system. Already at dilution levels as low as two percent, well below those reported in cerium-based pyrochlores, these vacancy-induced processes connect percolating clusters of
spins and dominate over the conventional
quantum-
spin-ice dynamics. We therefore argue that
magnetic vacancies in current experiments can strongly contaminate, and potentially completely obscure, the sought-after signatures of
quantum spin ice. We support these conclusions using large-scale, unbiased
quantum Monte Carlo simulations and exact diagonalization.
Read the full abstract ▾
All authors Alaric L. Sanders, Gautam K. Naik, Jonathan N. Hallén, Robin Schäfer
Mesoscale and Nanoscale Physics
Interaction induced flattening of optical transition quantum geometry
Xu Yang and Justin C. W. Song
The Riemannian geometry of optical transition dipoles has become a useful picture for understanding linear and nonlinear optical response. Here we argue that the interacting
quantum geometry of optical transitions possesses a rich structure and can be naturally delineated into two types: localized and delocalized particle-hole excitations. The former possess uniform
quantum geometry with flat (vanishing) Hermitian
curvature; the latter possess non-uniform
quantum geometry with non-vanishing Hermitian
curvature. As a striking example, we find that uniform
quantum geometry can be produced by electron-hole interactions: even when composed from extended Bloch states in the particle and hole bands, we find
excitons have uniform and flat
quantum geometry. By developing a many-body length gauge formulation of nonlinear response, we find this uniform and flat excitonic
quantum geometry zeros its third-order circular photoconductivity in non-
magnetic materials in stark contrast to its non-interacting counterparts. Similarly, the zero Hermitian
curvature of localized optical transitions locks their
Hall response to that of the ground state. This demonstrates the rich landscape of many-body optical response controlled by an interacting
quantum geometry.
Read the full abstract ▾