Showing 1–50 of 1,626 results for author: Di, S

RemoteAgent: Bridging Vague Human Intents and Earth Observation with RL-based Agentic MLLMs

Authors: Liang Yao, Shengxiang Xu, Fan Liu, Chuanyi Zhang, Bishun Yao, Rui Min, Yongjun Li, Chaoqian Ouyang, Shimin Di, Min-Ling Zhang

Abstract: Earth Observation (EO) systems are essentially designed to support domain experts who often express their requirements through vague natural language rather than precise, machine-friendly instructions. Depending on the specific application scenario, these vague queries can demand vastly different levels of visual precision. Consequently, a practical EO AI system must bridge the gap between ambiguo… ▽ More Earth Observation (EO) systems are essentially designed to support domain experts who often express their requirements through vague natural language rather than precise, machine-friendly instructions. Depending on the specific application scenario, these vague queries can demand vastly different levels of visual precision. Consequently, a practical EO AI system must bridge the gap between ambiguous human queries and the appropriate multi-granularity visual analysis tasks, ranging from holistic image interpretation to fine-grained pixel-wise predictions. While Multi-modal Large Language Models (MLLMs) demonstrate strong semantic understanding, their text-based output format is inherently ill-suited for dense, precision-critical spatial predictions. Existing agentic frameworks address this limitation by delegating tasks to external tools, but indiscriminate tool invocation is computationally inefficient and underutilizes the MLLM's native capabilities. To this end, we propose RemoteAgent, an agentic framework that strategically respects the intrinsic capability boundaries of MLLMs. To empower this framework to understand real user intents, we construct VagueEO, a human-centric instruction dataset pairing EO tasks with simulated vague natural-language queries. By leveraging VagueEO for reinforcement fine-tuning, we align an MLLM into a robust cognitive core that directly resolves image- and sparse region-level tasks. Consequently, RemoteAgent processes suitable tasks internally while intelligently orchestrating specialized tools via the Model Context Protocol exclusively for dense predictions. Extensive experiments demonstrate that RemoteAgent achieves robust intent recognition capabilities while delivering highly competitive performance across diverse EO tasks. △ Less

Submitted 8 April, 2026; originally announced April 2026.

Euclid Quick Data Release (Q1). AgileLens: A scalable CNN-based pipeline for strong gravitational lens identification

Authors: Euclid Collaboration, X. Xu, R. Chen, T. Li, A. R. Cooray, S. Schuldt, J. A. Acevedo Barroso, D. Stern, D. Scott, M. Meneghetti, G. Despali, J. Chopra, Y. Cao, M. Cheng, J. Buda, J. Zhang, J. Furumizo, R. Valencia, Z. Jiang, C. Tortora, N. E. P. Lines, T. E. Collett, S. Fotopoulou, A. Galan, A. Manjón-García , et al. (286 additional authors not shown)

Abstract: We present an end-to-end, iterative pipeline for efficient identification of strong galaxy--galaxy lensing systems, applied to the Euclid Q1 imaging data. Starting from VIS catalogues, we reject point sources, apply a magnitude cut (I$_E$ $\leq$ 24) on deflectors, and run a pixel-level artefact/noise filter to build 96 $\times$ 96 pix cutouts; VIS+NISP colour composites are constructed with a VIS-… ▽ More We present an end-to-end, iterative pipeline for efficient identification of strong galaxy--galaxy lensing systems, applied to the Euclid Q1 imaging data. Starting from VIS catalogues, we reject point sources, apply a magnitude cut (I$_E$ $\leq$ 24) on deflectors, and run a pixel-level artefact/noise filter to build 96 $\times$ 96 pix cutouts; VIS+NISP colour composites are constructed with a VIS-anchored luminance scheme that preserves VIS morphology and NISP colour contrast. A VIS-only seed classifier supplies clear positives and typical impostors, from which we curate a morphology-balanced negative set and augment scarce positives. Among the six CNNs studied initially, a modified VGG16 (GlobalAveragePooling + 256/128 dense layers with the last nine layers trainable) performs best; the training set grows from 27 seed lenses (augmented to 1809) plus 2000 negatives to a colour dataset of 30,686 images. After three rounds of iterative fine-tuning, human grading of the top 4000 candidates ranked by the final model yields 441 Grade A/B candidate lensing systems, including 311 overlapping with the existing Q1 strong-lens catalogue, and 130 additional A/B candidates (9 As and 121 Bs) not previously reported. Independently, the model recovers 740 out of 905 (81.8%) candidate Q1 lenses within its top 20,000 predictions, considering off-centred samples. Candidates span I$_E$ $\simeq$ 17--24 AB mag (median 21.3 AB mag) and are redder in Y$_E$--H$_E$ than the parent population, consistent with massive early-type deflectors. Each training iteration required a week for a small team, and the approach easily scales to future Euclid releases; future work will calibrate the selection function via lens injection, extend recall through uncertainty-aware active learning, explore multi-scale or attention-based neural networks with fast post-hoc vetters that incorporate lens models into the classification. △ Less

Submitted 7 April, 2026; originally announced April 2026.

Comments: 30 pages, 16 figures

The Wonderful World of Binary Stars

Authors: Andrea Barone, Henri M. J. Boffin, Beatrice Caccherano, Simona Di Stefano, Akhila Divakaran, Alexandra S. Murphy, María José Rain, Elyar Sedaghati, Paul V. Steimle

Abstract: During the 2026 ESO La Silla Observing school, about twenty students attended lectures and performed observations to learn various aspects of observational astronomy. The school, which took place during the first two weeks of February 2026, made use of EFOSC2/NTT and HARPS+NIRPS/3.6m. One of the groups was devoted to the study of binary stars. Several projects were considered and followed up by so… ▽ More During the 2026 ESO La Silla Observing school, about twenty students attended lectures and performed observations to learn various aspects of observational astronomy. The school, which took place during the first two weeks of February 2026, made use of EFOSC2/NTT and HARPS+NIRPS/3.6m. One of the groups was devoted to the study of binary stars. Several projects were considered and followed up by some of the six students in this group. The first subgroup used HARPS to study the Rossiter-McLaughlin effect in binary stars to infer the relative inclination of the rotation axis of the primary with respect to the orbital plane. A detailed study of the contact binary system HD 115264 led to the conclusion that the primary is well aligned, likely as a result of strong tidal forces within the binary. The second subgroup analysed blue straggler stars (BSS) in open clusters, using both HARPS and EFOSC2. With HARPS, they looked at some well-known long-period binary with the aim of determining their chemical abundances, thereby confirming their membership to the cluster, as well as looking for any chemical anomalies that might be explained by mass transfer. EFOSC2 was used to derive radial velocities of rapidly varying BSS. For one of them - the star Rediet - the students clearly detected and analysed the radial velocity variations due to the second overtone pulsation, thereby confirming its delta Scuti character. Finally, one student used EFOSC2 to study planetary nebulae (PN) - taking nice images of some of these intricate objects, as well as doing time-resolved photometry and spectra of some others. In one case, the binary nature of the central star of the PN was proven, confirming some previous estimates done with ZTF. Each subgroup was thus able to obtain useful research results, which we present hereafter. △ Less

Submitted 7 April, 2026; originally announced April 2026.

FactReview: Evidence-Grounded Reviews with Literature Positioning and Execution-Based Claim Verification

Authors: Hang Xu, Ling Yue, Chaoqian Ouyang, Yuchen Liu, Libin Zheng, Shaowu Pan, Shimin Di, Min-Ling Zhang

Abstract: Peer review in machine learning is under growing pressure from rising submission volume and limited reviewer time. Most LLM-based reviewing systems read only the manuscript and generate comments from the paper's own narrative. This makes their outputs sensitive to presentation quality and leaves them weak when the evidence needed for review lies in related work or released code. We present FactRev… ▽ More Peer review in machine learning is under growing pressure from rising submission volume and limited reviewer time. Most LLM-based reviewing systems read only the manuscript and generate comments from the paper's own narrative. This makes their outputs sensitive to presentation quality and leaves them weak when the evidence needed for review lies in related work or released code. We present FactReview, an evidence-grounded reviewing system that combines claim extraction, literature positioning, and execution-based claim verification. Given a submission, FactReview identifies major claims and reported results, retrieves nearby work to clarify the paper's technical position, and, when code is available, executes the released repository under bounded budgets to test central empirical claims. It then produces a concise review and an evidence report that assigns each major claim one of five labels: Supported, Supported by the paper, Partially supported, In conflict, or Inconclusive. In a case study on CompGCN, FactReview reproduces results that closely match those reported for link prediction and node classification, yet also shows that the paper's broader performance claim across tasks is not fully sustained: on MUTAG graph classification, the reproduced result is 88.4%, whereas the strongest baseline reported in the paper remains 92.6%. The claim is therefore only partially supported. More broadly, this case suggests that AI is most useful in peer review not as a final decision-maker, but as a tool for gathering evidence and helping reviewers produce more evidence-grounded assessments. The code is public at https://github.com/DEFENSE-SEU/Review-Assistant. △ Less

Submitted 7 April, 2026; v1 submitted 5 April, 2026; originally announced April 2026.

Safe Control of Feedback-Interconnected Systems via Singular Perturbations

Authors: Stefano Di Gregorio, Guido Carnevale, Giuseppe Notarstefano

Abstract: Control Barrier Functions (CBFs) have emerged as a powerful tool in the design of safety-critical controllers for nonlinear systems. In modern applications, complex systems often involve the feedback interconnection of subsystems evolving at different timescales, e.g., two parts from different physical domains (e.g., the electrical and mechanical parts of robotic systems) or a physical plant and a… ▽ More Control Barrier Functions (CBFs) have emerged as a powerful tool in the design of safety-critical controllers for nonlinear systems. In modern applications, complex systems often involve the feedback interconnection of subsystems evolving at different timescales, e.g., two parts from different physical domains (e.g., the electrical and mechanical parts of robotic systems) or a physical plant and an (optimization or control) algorithm. In these scenarios, safety constraints often involve only a portion of the overall system. Inspired by singular perturbations for stability analysis, we develop a formal procedure to lift a safety certificate designed on a reduced-order model to the overall feedback-interconnected system. Specifically, we show that under a sufficient timescale separation between slow and fast dynamics, a composite CBF can be designed to certify the forward invariance of the safe set for the interconnected system. As a result, the online safety filter only needs to be solved for the lower-dimensional, reduced-order model. We numerically test the proposed approach on: (i) a robotic arm with joint motor dynamics, and (ii) a physical plant driven by an optimization algorithm. △ Less

Submitted 2 April, 2026; originally announced April 2026.

Euclid preparation. Non-Gaussianity of 2-pt statistics likelihood: Parameter inference with a non-Gaussian likelihood in Fourier and configuration space

Authors: Euclid Collaboration, S. Gouyou Beauchamps, J. Bel, P. Baratta, C. Carbone, B. Altieri, S. Andreon, N. Auricchio, C. Baccigalupi, M. Baldi, S. Bardelli, P. Battaglia, F. Bernardeau, A. Biviano, E. Branchini, M. Brescia, S. Camera, G. Cañas-Herrera, V. Capobianco, V. F. Cardone, J. Carretero, S. Casas, M. Castellano, G. Castignani, S. Cavuoti , et al. (255 additional authors not shown)

Abstract: In this work we account for this skewness in parameter inference by modelling the likelihood through an Edgeworth expansion which involves the complete skewness tensor, composed of 1-point, 2-point, and 3-point correlators. To simplify the calculations of this expansion we perform a change of basis which reduces the precision matrix to the identity. In this basis, the off-diagonal elements of the… ▽ More In this work we account for this skewness in parameter inference by modelling the likelihood through an Edgeworth expansion which involves the complete skewness tensor, composed of 1-point, 2-point, and 3-point correlators. To simplify the calculations of this expansion we perform a change of basis which reduces the precision matrix to the identity. In this basis, the off-diagonal elements of the skewness tensor are consistent with zero, while the amplitude of its diagonal match the level expected for a Gaussian underlying field. We perform parameter inference with this likelihood model and find that including only the diagonal part of the skewness is sufficient, while incorporating the full skewness tensor injects noise without improving accuracy. Despite the estimated excess skewness in the original basis, the cosmological constraints remain effectively unchanged when adopting a Gaussian likelihood or considering the more complete Edgeworth expansion, with variations in the figure of merit of cosmological parameters between the two cases below $5\%$. This result remains unchanged against variations of the survey volume and geometry, scale-cut, and 2-point statistic (power spectrum or correlation function). Using $10\, 000$ cloned \Euclid large mocks based on realistic galaxy catalogues with characteristics close to future \Euclid data, we find no detectable excess skewness on intermediate scales, due to the level of shot noise expected for the \Euclid spectroscopic sample. We conclude that the Gaussian likelihood assumption is robust for \Euclid 2-point statistics analyses in both Fourier and configuration space. △ Less

Submitted 1 April, 2026; originally announced April 2026.

Comments: 25 pages, 12 figures, submitted to A&A

Euclid preparation. Impact of redshift distribution uncertainties on the joint analysis of photometric galaxy clustering and weak gravitational lensing

Authors: Euclid Collaboration, K. A. Bertmann, A. Porredon, V. Duret, J. Fonseca, H. Hildebrandt, I. Tutusaus, S. Camera, S. Escoffier, N. Aghanim, B. Altieri, A. Amara, S. Andreon, N. Auricchio, C. Baccigalupi, M. Baldi, S. Bardelli, P. Battaglia, A. Biviano, E. Branchini, M. Brescia, G. Cañas-Herrera, V. Capobianco, C. Carbone, V. F. Cardone , et al. (268 additional authors not shown)

Abstract: One of the $\textit{Euclid}$ mission's key projects is the so-called 3$\times$2pt analysis, that is, the combination of cosmic shear, photometric galaxy clustering, and galaxy-galaxy lensing. Although $\textit{Euclid}$ has established quality requirements for the photo-$z$ accuracy needed for the weak lensing galaxy sample, no such requirements have been set for the photometric clustering sample.… ▽ More One of the $\textit{Euclid}$ mission's key projects is the so-called 3$\times$2pt analysis, that is, the combination of cosmic shear, photometric galaxy clustering, and galaxy-galaxy lensing. Although $\textit{Euclid}$ has established quality requirements for the photo-$z$ accuracy needed for the weak lensing galaxy sample, no such requirements have been set for the photometric clustering sample. In this paper, we investigate the impact of redshift uncertainties on $\textit{Euclid}$'s photometric galaxy clustering analysis and its combination with weak gravitational lensing, focusing on data release 1 (DR1). In particular, we study whether having precise knowledge of the mean of the redshift distributions per bin is sufficient to avoid biases in the resulting cosmological constraints or whether accuracy in the higher-order moments of the distribution is required. We evaluate the results based on their constraining power on $w_{\mathrm{0}}$ and $w_{a}$ and define thresholds for the precision and accuracy of $\textit{Euclid}$'s redshift distribution of the photometric clustering sample. We find that the redshift distributions of the photometric clustering sample must be known at an accuracy of 0.004(1+$z$) in the mean in order to recover 80$\%$ of the constraining power in $\textit{Euclid}$'s DR1 $w_{\mathrm{0}}w_{a}$CDM 3$\times$2pt analysis. The impact of the uncertainty on the width is negligible, provided the mean redshift is constrained with sufficient accuracy. For most sources of redshift distribution error, attaining the requirement on the mean will also reduce uncertainty in the width well below the required level. △ Less

Submitted 1 April, 2026; originally announced April 2026.

Comments: 13+6 pages, 5+9 figures, 5+5 tables

ARCOL: Aspect Ratio Constrained Orthogonal Layout

Authors: Zainab Alsuwaykit, Yousef Rajeh, Alexandre Kouyoumdjian, Steve Kieffer, Dominik Engel, Sara Di Bartolomeo, Martin Nöllenburg, Ivan Viola

Abstract: Orthogonal graph layout algorithms aim to produce clear, compact, and readable network diagrams by arranging nodes and edges along horizontal and vertical lines, while minimizing bends and crossings. Most existing orthogonal layout methods focus primarily on quality criteria such as area usage, total edge length, and bend minimization. Explicitly controlling the global aspect ratio (AR) of the res… ▽ More Orthogonal graph layout algorithms aim to produce clear, compact, and readable network diagrams by arranging nodes and edges along horizontal and vertical lines, while minimizing bends and crossings. Most existing orthogonal layout methods focus primarily on quality criteria such as area usage, total edge length, and bend minimization. Explicitly controlling the global aspect ratio (AR) of the resulting layout is as of now unexplored. Existing orthogonal layout methods offer no control over the resulting AR and their rigid geometric constraints make adaptation of finished layouts difficult. With the increasing variety of aspect ratios encountered in daily life, from wide monitors to tall mobile devices or fixed-size interface panels, there is a clear need for aspect ratio control in orthogonal layout methods. To tackle this issue, we introduce Aspect Ratio-Constrained Orthogonal Layout (ARCOL). Building upon the Human-like Orthogonal Layout Algorithm (HOLA)~\cite{Kieffer2016}, we integrate aspect ratio at two different stages: (1) into the stress minimization phase, as a soft constraint, allowing the layout algorithm to gently guide node positions toward a specified target AR, while preserving visual clarity and topological faithfulness; and (2) into the tree reattachment phase, where we modify the cost function to favor placements that improve the AR. We evaluate our approach through quantitative evaluation and a user study, as well as expert interviews. Our evaluations show that ARCOL produces balanced and space efficient orthogonal layouts across diverse aspect ratios. △ Less

Submitted 31 March, 2026; originally announced March 2026.

Euclid Quick Data Release (Q1). The Strong Lensing Discovery Engine F -- Bright and low-redshift strong lenses

Authors: Euclid Collaboration, L. R. Ecker, M. Fabricius, S. Seitz, R. Saglia, N. E. P. Lines, P. Holloway, T. Li, A. Verma, F. Balzer, Q. Jin, A. Manjón-García, S. H. Vincken, J. Wilde, J. A. Acevedo Barroso, J. W. Nightingale, K. Rojas, S. Schuldt, M. Walmsley, T. E. Collett, G. Despali, A. Sonnenfeld, C. Tortora, R. B. Metcalf, R. Bender , et al. (324 additional authors not shown)

Abstract: We present 72 additional galaxy-galaxy strong lenses that complement the sample discovered in the Euclid Quick Release 1 data (63.1 deg^2) of the Strong Lens Discovery Engine (SLDE) papers A-E. It is shown that previous pre-selection of potential lenses, which excluded objects from the Gaia catalogue, led to missing several bright and low-redshift strong lenses, adding more than 10% new strong len… ▽ More We present 72 additional galaxy-galaxy strong lenses that complement the sample discovered in the Euclid Quick Release 1 data (63.1 deg^2) of the Strong Lens Discovery Engine (SLDE) papers A-E. It is shown that previous pre-selection of potential lenses, which excluded objects from the Gaia catalogue, led to missing several bright and low-redshift strong lenses, adding more than 10% new strong lens candidates compared to the previous search. In total, the catalogue includes 38 "grade A" (confident) and 34 "grade B" (probable) candidates. These lenses are identified through a combination of two independent searches for bright nearby objects: one based on machine-learning models followed by expert visual inspection, and the other based solely on expert visual inspection, targeting objects not included in the initial machine-learning selection (a limitation identified only after extensive visual inspection). With these additional strong lens candidates, we augment the expected number of high-confidence candidates in the Euclid Wide Survey from previous forecasts to 120000. Detailed semi-automated lens modelling confirms at least 41 systems out of 72, a fraction consistent with that found in SLDE A (315 out of 488). These include: multiple edge-on disc lenses; sources with arcs near the lens centre; "red sources"; and an edge-on disk galaxy lensing a galaxy merger, producing two sets of lensed features, an Einstein ring and a doubly imaged component. The median redshift of these systems is $Δ$ z ~ 0.3 lower than that of the SLDE A sample. △ Less

Submitted 30 March, 2026; originally announced March 2026.

Euclid preparation. Galaxy power spectrum and bispectrum modelling

Authors: Euclid Collaboration, K. Pardede, A. Eggemeier, D. Alkhanishvili, E. Sefusatti, A. Moradinezhad Dizgah, L. Christoph, A. Chudaykin, M. Kärcher, D. Linde, M. Marinucci, C. Porciani, A. Veropalumbo, M. Crocce, M. S. Cagliari, B. Camacho Quevedo, L. Castiblanco, E. Castorina, G. D'Amico, V. Desjacques, A. Farina, G. Gambardella, M. Guidi, J. Lesgourgues, C. Moretti , et al. (283 additional authors not shown)

Abstract: Higher-order correlation functions of the large-scale galaxy distribution offer access to information beyond that contained in standard 2-point statistics such as the power spectrum. In this work we assess this potential for the $\textit{Euclid}$ mission using synthetic catalogues of H$α$ galaxies based on the 54 $\, h^{-3} \, {\rm Gpc}^3$ Flagship I simulation, designed to reproduce the… ▽ More Higher-order correlation functions of the large-scale galaxy distribution offer access to information beyond that contained in standard 2-point statistics such as the power spectrum. In this work we assess this potential for the $\textit{Euclid}$ mission using synthetic catalogues of H$α$ galaxies based on the 54 $\, h^{-3} \, {\rm Gpc}^3$ Flagship I simulation, designed to reproduce the $\textit{Euclid}$ spectroscopic sample. We comprehensively validate the one-loop galaxy power spectrum and tree-level bispectrum predictions from perturbation theory in both real and redshift space. Assuming scale cuts consistent with our previous power spectrum study on the same catalogues, this modelling yields unbiased cosmological constraints for the bispectrum up to $k_{\rm max} = 0.15\,\, h \, {\rm Mpc}^{-1}$ in real space and $0.08 \, (0.1)\,\, h \, {\rm Mpc}^{-1}$ at the lowest (highest) redshift, corresponding to $z=0.9$ ($z=1.8$), for the monopole and quadrupole in redshift space using statistical uncertainties corresponding to the full simulation volume. With these scale cuts, adding bispectrum information to the power spectrum improves constraints on the amplitude of scalar perturbations and the matter density by up to 30 %, increasing the overall figure of merit for key cosmological parameters by a factor of about 2.5. Similar conclusions hold when statistical uncertainties are rescaled to a $\textit{Euclid}$-like volume, highlighting the importance of the bispectrum for fully exploiting the forthcoming $\textit{Euclid}$ data. Our analysis also provides the first detailed characterisation of the nonlinear bias model of H$α$ emitters, showing that bias relations calibrated on low-resolution \textit{N}-body simulations do not adequately describe the clustering of H$α$ galaxies at low redshift, whereas excursion-set and co-evolution relations for tidal biases remain accurate. △ Less

Submitted 29 March, 2026; originally announced March 2026.

Comments: 16+2 pages, 11 figures, 4+1 tables, abstract abridged for arXiv submission

Narrowband searches for continuous gravitational waves from known pulsars in the first two parts of the fourth LIGO--Virgo--KAGRA observing run

Authors: The LIGO Scientific Collaboration, the Virgo Collaboration, the KAGRA Collaboration, A. G. Abac, I. Abouelfettouh, F. Acernese, K. Ackley, A. Adam, C. Adamcewicz, S. Adhicary, D. Adhikari, N. Adhikari, R. X. Adhikari, V. K. Adkins, S. Afroz, A. Agapito, D. Agarwal, M. Agathos, N. Aggarwal, S. Aggarwal, O. D. Aguiar, I. -L. Ahrend, L. Aiello, A. Ain, P. Ajith , et al. (1831 additional authors not shown)

Abstract: Rotating non-axisymmetric neutron stars (NSs) are promising sources for continuous gravitational waves (CWs). Such CWs can, if detected, inform us about the internal structure and equation of state of NSs. Here, we present a narrowband search for CWs from known pulsars, for which an efficient and sensitive matched-filter search can be applied. Narrowband searches are designed to be robust to misma… ▽ More Rotating non-axisymmetric neutron stars (NSs) are promising sources for continuous gravitational waves (CWs). Such CWs can, if detected, inform us about the internal structure and equation of state of NSs. Here, we present a narrowband search for CWs from known pulsars, for which an efficient and sensitive matched-filter search can be applied. Narrowband searches are designed to be robust to mismatches between the electromagnetic (EM) and gravitational emissions, in contrast to fully targeted searches where the CW emission is assumed to be phase-locked to the EM one. In this work, we search for the CW counterparts emitted by 34 pulsars using data from the first and second parts of the fourth LIGO--Virgo--KAGRA observing run. This is the largest number of pulsars so far targeted for narrowband searches in the advanced detector era. We use the 5n-vector narrowband pipeline, which applies frequency-domain matched filtering. In previous searches, it covered a narrow range in the frequency -- frequency time derivative ($f$ -- $\dot{f}$) space. Here, we also explore a range in the second time derivative of the frequency $\ddot{f}$ around the value indicated by EM observations. Additionally, for the first time, we target sources in a binary system with this kind of search. We find no evidence for CWs and therefore set upper limits on the strain amplitude emitted by each pulsar, using simulated signals added in real data. For 20 analyses, we report an upper limit below the theoretical spin-down limit. The tightest constraint is for pulsar PSR J0534+2200 (the Crab pulsar), for which our strain upper limit on the CW amplitude is $\lesssim 2\%$ of its spin-down limit, corresponding to less than $0.04\%$ of the spin-down power being radiated in the CW channel. △ Less

Submitted 26 March, 2026; originally announced March 2026.

Comments: 30 pages, 6 figures, submitted to ApJ

Report number: LIGO-P2500612

Searches for Continuous Gravitational Waves from Supernova Remnants in the first part of the LIGO-Virgo-KAGRA Fourth Observing run

Authors: The LIGO Scientific Collaboration, the Virgo Collaboration, the KAGRA Collaboration, A. G. Abac, I. Abouelfettouh, F. Acernese, K. Ackley, C. Adamcewicz, S. Adhicary, D. Adhikari, N. Adhikari, R. X. Adhikari, V. K. Adkins, S. Afroz, A. Agapito, D. Agarwal, M. Agathos, N. Aggarwal, S. Aggarwal, O. D. Aguiar, I. -L. Ahrend, L. Aiello, A. Ain, P. Ajith, T. Akutsu , et al. (1742 additional authors not shown)

Abstract: We present results from directed searches for continuous gravitational waves from a sample of 15 nearby supernova remnants, likely hosting young neutron star candidates, using data from the first eight months of the fourth observing run (O4) of the LIGO-Virgo-KAGRA Collaboration. The analysis employs five pipelines: four semi-coherent methods -- the Band-Sampled-Data directed pipeline, Weave and t… ▽ More We present results from directed searches for continuous gravitational waves from a sample of 15 nearby supernova remnants, likely hosting young neutron star candidates, using data from the first eight months of the fourth observing run (O4) of the LIGO-Virgo-KAGRA Collaboration. The analysis employs five pipelines: four semi-coherent methods -- the Band-Sampled-Data directed pipeline, Weave and two Viterbi pipelines (single- and dual-harmonic) -- and PyStoch, a cross-correlation-based pipeline. These searches cover wide frequency bands and do not assume prior knowledge of the targets' ephemerides. No evidence of a signal is found from any of the 15 sources. We set 95\% confidence-level upper limits on the intrinsic strain amplitude, with the most stringent constraints reaching $\sim 4 \times 10^{-26}$ near 300 Hz for the nearby source G266.2$-$1.2 (Vela Jr.). We also derive limits on neutron star ellipticity and $r$-mode amplitudes for the same source, with the best constraints reaching $\lesssim 10^{-7}$ and $\lesssim 10^{-5}$, respectively, at frequencies above 400 Hz. These results represent the most sensitive wide-band directed searches for continuous gravitational waves from supernova remnants to date. △ Less

Submitted 2 April, 2026; v1 submitted 26 March, 2026; originally announced March 2026.

Experimental Analysis of FreeRTOS Dependability through Targeted Fault Injection Campaigns

Authors: Luca Mannella, Stefano Di Carlo, Alessandro Savino

Abstract: Real-Time Operating Systems (RTOSes) play a crucial role in safety-critical domains, where deterministic and predictable task execution is essential. Yet they are increasingly exposed to ionizing radiation, which can compromise system dependability. To assess FreeRTOS under such conditions, we introduce KRONOS, a software-based, non-intrusive post-propagation Fault Injection (FI) framework that… ▽ More Real-Time Operating Systems (RTOSes) play a crucial role in safety-critical domains, where deterministic and predictable task execution is essential. Yet they are increasingly exposed to ionizing radiation, which can compromise system dependability. To assess FreeRTOS under such conditions, we introduce KRONOS, a software-based, non-intrusive post-propagation Fault Injection (FI) framework that injects transient and permanent faults into Operating System-visible kernel data structures without specialized hardware or debug interfaces. Using KRONOS, we conduct an extensive FI campaign on core FreeRTOS kernel components, including scheduler-related variables and Task Control Blocks (TCBs), characterizing the impact of kernel-level corruptions on functional correctness, timing behavior, and availability. The results show that corruption of pointer and key scheduler-related variables frequently leads to crashes, whereas many TCB fields have only a limited impact on system availability. △ Less

Submitted 26 March, 2026; originally announced March 2026.

Comments: 6 pages; 5 figures; sent to the International Symposium on Design and Diagnostics of Electronic Circuits and Systems (DDECS) 2026

Enabling Homomorphic Analytical Operations on Compressed Scientific Data with Multi-stage Decompression

Authors: Xuan Wu, Sheng Di, Tripti Agarwal, Kai Zhao, Xin Liang, Franck Cappello

Abstract: Error-controlled lossy compressors have been widely used in scientific applications to reduce the unprecedented size of scientific data while keeping data distortion within a user-specified threshold. While they significantly mitigate the pressure for data storage and transmission, they prolong the time to access the data because decompression is required to transform the binary compressed data in… ▽ More Error-controlled lossy compressors have been widely used in scientific applications to reduce the unprecedented size of scientific data while keeping data distortion within a user-specified threshold. While they significantly mitigate the pressure for data storage and transmission, they prolong the time to access the data because decompression is required to transform the binary compressed data into meaningful floating-point numbers. This incurs noticeable overhead for common analytical operations on scientific data that extract or derive useful information, because the time cost of the operations could be much lower than that of decompression. In this work, we design an error-controlled lossy compression and analytical framework that features multi-stage decompression and homomorphic analytical operation algorithms on intermediate decompressed data for reduced data access latency. Our contributions are threefold. (1) We abstract a generic compression pipeline with partial decompression to multiple intermediate data representations and implement four instances based on state-of-the-art high-throughput scientific data compressors. (2) We carefully design homomorphic algorithms to enable direct operations on intermediate decompressed data for three types of analytical operations on scientific data. (3) We evaluate our approach using five real-world scientific datasets. Experimental evaluations demonstrate that our method achieves significant speedups when performing analytical operations on compressed scientific data across all three targeted analytical operation types. △ Less

Submitted 26 March, 2026; originally announced March 2026.

Comments: ICDE 2026

Euclid preparation. Cosmology Likelihood for Observables in Euclid (CLOE). 2. Code implementation

Authors: Euclid Collaboration, S. Joudaki, V. Pettorino, L. Blot, M. Bonici, S. Camera, G. Cañas-Herrera, V. F. Cardone, P. Carrilho, S. Casas, S. Davini, S. Di Domizio, S. Farrens, L. W. K. Goh, S. Gouyou Beauchamps, S. Ilić, F. Keil, A. M. C. Le Brun, M. Martinelli, C. Moretti, A. Pezzotta, Z. Sakr, A. G. Sánchez, D. Sciotti, K. Tanidis , et al. (323 additional authors not shown)

Abstract: We provide a description of the code implementation and structure of Cosmology Likelihood for Observables in Euclid (CLOE), developed by members of the Euclid Consortium. CLOE is a modular Python code for computing the theoretical predictions of cosmological observables and evaluating them against state-of-the-art data from galaxy surveys such as Euclid in a unified likelihood. This primarily incl… ▽ More We provide a description of the code implementation and structure of Cosmology Likelihood for Observables in Euclid (CLOE), developed by members of the Euclid Consortium. CLOE is a modular Python code for computing the theoretical predictions of cosmological observables and evaluating them against state-of-the-art data from galaxy surveys such as Euclid in a unified likelihood. This primarily includes the core observables of weak gravitational lensing, photometric galaxy clustering, galaxy-galaxy lensing, and spectroscopic galaxy clustering, but also extended probes such as the clusters of galaxies and cross-correlations of galaxy positions and shapes with the cosmic microwave background. While CLOE has been developed to serve as the unified framework for the parameter inferences in Euclid, it has general capabilities that can serve the broader cosmological community. It is different from other comparable cosmological tools in that it is written entirely in Python, performs the full likelihood calculation, and includes both photometric and spectroscopic observables. We will focus on the primary probes of Euclid and will describe the overall code structure, rigorous code development practices, extensive documentation, unique features, speed optimization, and future development plans. CLOE is publicly available at https://github.com/cloe-org/cloe. △ Less

Submitted 23 March, 2026; originally announced March 2026.

Comments: Second in a series of six papers presenting CLOE, the Euclid likelihood code; 43 pages, 11 figures, accepted for publication in A&A (with a table of contents added)

Electric-field-induced X-ray Nonreciprocal Dichroism in Hematite

Authors: Takeshi Hayashida, Koei Matsumoto, Keito Arakawa, Yves Joly, Sergio Di Matteo, Kenji Tamasaku, Yoshikazu Tanaka, Tsuyoshi Kimura

Abstract: Hematite (alpha-Fe2O3) is a prototypical room temperature antiferromagnet whose time-reversal-odd magnetic structure has recently attracted renewed attention. While such magnetic symmetry can be characterized in terms of higher-order multipoles beyond the magnetic dipole, their manifestation in measurable physical phenomena has remained largely elusive. In this work, we investigate x-ray absorptio… ▽ More Hematite (alpha-Fe2O3) is a prototypical room temperature antiferromagnet whose time-reversal-odd magnetic structure has recently attracted renewed attention. While such magnetic symmetry can be characterized in terms of higher-order multipoles beyond the magnetic dipole, their manifestation in measurable physical phenomena has remained largely elusive. In this work, we investigate x-ray absorption near the Fe K-edge of hematite under an applied electric field, which explicitly breaks space-inversion symmetry. We observe an electric-field-induced x-ray nonreciprocal linear dichroism (E-induced XNLD) that reflects the time-reversal-odd nature of the magnetic order. Numerical simulations based on ab-initio density functional theory reproduce the observed spectra, including their dependence on the antiferromagnetic domain and x-ray polarization. Furthermore, a symmetry-resolved multipole analysis reveals that this response originates from the magnetic quadrupole and the magnetic toroidal octupole induced by the applied electric field. These results demonstrate that electric-field-modulated x-ray absorption provides direct access to the antiferroic order of higher-order multipoles in time-reversal-odd antiferromagnets, thereby establishing a general framework to uncover hidden symmetry properties in magnetic materials. △ Less

Submitted 21 March, 2026; originally announced March 2026.

Advanced Virgo Plus for O5 -- Design Report Overview

Authors: F. Acernese, A. Agapito, D. Agarwal, I. -L. Ahrend, L. Aiello, A. Ain, S. Albanesi, W. Ali, C. Alléné, A. Allocca, W. Amar, A. Amato, F. Amicucci, C. Amra, M. Andia, T. Andrić, S. Ansoldi, S. Antier, E. Z. Appavuravther, M. Arca Sedda, F. Arciprete, F. Armato, N. Arnaud, L. Asprea, M. Assiduo , et al. (556 additional authors not shown)

Abstract: This document presents an overview of the design, implementation, and expected performance of the Advanced Virgo Plus (AdV+) upgrades in view of the O5 observing run. Following the experience gained during the O4 commissioning and operations, the Virgo Collaboration has revised the upgrade strategy to address limitations associated with marginally stable recycling cavities. The O5 upgrade program… ▽ More This document presents an overview of the design, implementation, and expected performance of the Advanced Virgo Plus (AdV+) upgrades in view of the O5 observing run. Following the experience gained during the O4 commissioning and operations, the Virgo Collaboration has revised the upgrade strategy to address limitations associated with marginally stable recycling cavities. The O5 upgrade program combines elements from the original AdV+ Phase II project with new design solutions, including the implementation of stable recycling cavities, a major modification to the central interferometer layout, and a comprehensive renewal of critical subsystems. The planned upgrades are organized in two steps, targeting progressive improvements in operational stability, noise reduction, and detector sensitivity. Key developments include new vacuum infrastructures, suspensions, mirrors, optical configurations, quantum noise reduction systems, and high-power laser technologies. The resulting configuration is expected to significantly enhance the interferometer performance, enabling a substantial increase in astrophysical reach and scientific return during O5. △ Less

Submitted 31 March, 2026; v1 submitted 20 March, 2026; originally announced March 2026.

GWTC-4.0: Tests of General Relativity. III. Tests of the Remnants

Authors: The LIGO Scientific Collaboration, the Virgo Collaboration, the KAGRA Collaboration, A. G. Abac, I. Abouelfettouh, F. Acernese, K. Ackley, C. Adamcewicz, S. Adhicary, D. Adhikari, N. Adhikari, R. X. Adhikari, V. K. Adkins, S. Afroz, A. Agapito, D. Agarwal, M. Agathos, N. Aggarwal, S. Aggarwal, O. D. Aguiar, I. -L. Ahrend, L. Aiello, A. Ain, P. Ajith, T. Akutsu , et al. (1757 additional authors not shown)

Abstract: This is the third paper of the set recording the results of the suite of tests of general relativity (GR) performed on the signals from the fourth Gravitational-Wave Transient Catalog (GWTC-4.0), where we focus on the remnants of the binary mergers. We examine for the first time 42 events from the first part of the fourth observing run of the LIGO-Virgo-KAGRA detectors, alongside events from the p… ▽ More This is the third paper of the set recording the results of the suite of tests of general relativity (GR) performed on the signals from the fourth Gravitational-Wave Transient Catalog (GWTC-4.0), where we focus on the remnants of the binary mergers. We examine for the first time 42 events from the first part of the fourth observing run of the LIGO-Virgo-KAGRA detectors, alongside events from the previous observation runs, restricting our analysis to the confident signals, which were measured in at least two detectors and that have false alarm rates $\le 10^{-3} \mathrm{yr}^{-1}$. This paper focuses on seven tests of the coalescence remnants. Three of these are tests of the ringdown and its consistency with the expected quasinormal mode spectrum of a Kerr black hole. Specifically, two tests analyze just the ringdown in the time domain, and the third test analyzes the entire signal in the frequency domain. Four tests allow for the existence of possible echoes arriving after the end of the ringdown, which are not expected in GR. We find overall consistency of the remnants with GR. When combining events by multiplying likelihoods (hierarchically), one analysis finds that the GR prediction lies at the boundary of the $98.6^{+1.4}_{-9.4}\%$ ($99.3^{+0.7}_{-4.5}\%$) credible region, an increase from $93.8^{+6.1}_{-20.0}\%$ ($94.9^{+4.4}_{-18.2}\%$) for GWTC-3.0. Here the ranges of values comes from bootstrapping to account for the finite number of events analyzed and suggest that some of the apparently significant deviation could be attributed to variance due to the finite catalog. Since the significance also decreases to 92.2% (96.2%) when including the more recent very loud event GW250114, there is no strong evidence for a GR deviation. We find no evidence for post-merger echoes in the events that were analyzed. (Abridged) △ Less

Submitted 19 March, 2026; originally announced March 2026.

Comments: As part of the Astrophysical Journal Letters Focus Issue on the Gravitational Wave Transient Catalog

Report number: LIGO-P2500067

GWTC-4.0: Tests of General Relativity. II. Parameterized Tests

Authors: The LIGO Scientific Collaboration, the Virgo Collaboration, the KAGRA Collaboration, A. G. Abac, I. Abouelfettouh, F. Acernese, K. Ackley, C. Adamcewicz, S. Adhicary, D. Adhikari, N. Adhikari, R. X. Adhikari, V. K. Adkins, S. Afroz, A. Agapito, D. Agarwal, M. Agathos, N. Aggarwal, S. Aggarwal, O. D. Aguiar, I. -L. Ahrend, L. Aiello, A. Ain, P. Ajith, T. Akutsu , et al. (1761 additional authors not shown)

Abstract: In this second of three papers on tests of general relativity (GR) applied to the compact binary coalescence signals in the fourth Gravitational-Wave Transient Catalog (GWTC-4.0), we present the results of the parameterized tests of GR and constraints on line-of-sight acceleration. We include events up to and including the first part of the fourth observing run (O4a) of the LIGO Virgo KAGRA detect… ▽ More In this second of three papers on tests of general relativity (GR) applied to the compact binary coalescence signals in the fourth Gravitational-Wave Transient Catalog (GWTC-4.0), we present the results of the parameterized tests of GR and constraints on line-of-sight acceleration. We include events up to and including the first part of the fourth observing run (O4a) of the LIGO Virgo KAGRA detectors. As in the other two papers in this series, we restrict our analysis to the 42 confident signals, measured by at least two detectors, that have false alarm rates $\le 10^{-3} \mathrm{yr}^{-1}$ from O4a, in addition to the 49 such events from previous observing runs. This paper focuses on the eight tests that constrain parameterized deviations from the expected GR (or unaccelerated) values. These include modifications of post-Newtonian (PN) parameters, spin-induced quadrupole moments different from those of a binary black hole, and possible dispersive or birefringent propagation effects. Overall, we find no evidence for physics beyond GR, for spin-induced quadrupole moments different from those of a Kerr black hole in GR, or for line of sight acceleration, with more than 90% of the events including the null result (no deviation) within their 90% credible intervals. We discuss possible systematics affecting the other events and tests, even though they are statistically not surprising, given noise. We improve the bounds on deviations from the GR PN coefficients by factors of 1.2-5.5 and provide illustrative translations to constraints on some modified theories. Also, we update the bound on the mass of the graviton, at 90% credibility, to $m_g \leq 1.92\times 10^{-23} \mathrm{eV}/c^2$. Thus, we see that GR holds, and many of the bounds on possible deviations derived from our events are the best to date. △ Less

Submitted 19 March, 2026; originally announced March 2026.

Comments: As part of the Astrophysical Journal Letters Focus Issue on the Gravitational Wave Transient Catalog

Report number: LIGO-P2500066

GWTC-4.0: Tests of General Relativity. I. Overview and General Tests

Authors: The LIGO Scientific Collaboration, the Virgo Collaboration, the KAGRA Collaboration, A. G. Abac, I. Abouelfettouh, F. Acernese, K. Ackley, C. Adamcewicz, S. Adhicary, D. Adhikari, N. Adhikari, R. X. Adhikari, V. K. Adkins, S. Afroz, A. Agapito, D. Agarwal, M. Agathos, N. Aggarwal, S. Aggarwal, O. D. Aguiar, I. -L. Ahrend, L. Aiello, A. Ain, P. Ajith, T. Akutsu , et al. (1759 additional authors not shown)

Abstract: The worldwide LIGO-Virgo-KAGRA network of gravitational-wave (GW) detectors continues to increase in sensitivity, thus increasing the quantity and quality of the detected GW signals from compact binary coalescences. These signals allow us to perform ever-more sensitive tests of general relativity (GR) in the dynamical and strong-field regime of gravity. This paper is the first of three, where we p… ▽ More The worldwide LIGO-Virgo-KAGRA network of gravitational-wave (GW) detectors continues to increase in sensitivity, thus increasing the quantity and quality of the detected GW signals from compact binary coalescences. These signals allow us to perform ever-more sensitive tests of general relativity (GR) in the dynamical and strong-field regime of gravity. This paper is the first of three, where we present the results of a suite of tests of GR using the binary signals included in the fourth GW Transient Catalog (GWTC-4.0), i.e., up to and including the first part of the fourth observing run of the detectors (O4a). We restrict our analysis to the 91 confident signals, henceforth called events, that were measured by at least two detectors, and have false alarm rates $\le 10^{-3} \mathrm{yr}^{-1}$. These include 42 events from O4a. This first paper presents an overview of the methods, selection of events and GR tests, and serves as a guidemap for all three papers. Here we focus on the four general tests of consistency, where we find no evidence for deviations from our models. Specifically, for all the events considered, we find consistency of the residuals with noise. The final mass and final spin as inferred from the low- and high-frequency parts of the waveform are consistent with each other. We also find no evidence for deviations from the GR predictions for the amplitudes of subdominant GW multipole moments, or for non-GR modes of polarization. We thus find that GR, without new physics beyond it, is still consistent with these GW events. The results of the two additional papers in this trio also find overall consistency with vacuum GR, with more than 90% of the events being consistent with GR at the 90% credible level. While one of the ringdown analyses finds the GR value in the tails for its combined results, this may be due in part to catalog variance. △ Less

Submitted 19 March, 2026; originally announced March 2026.

Comments: As part of the Astrophysical Journal Letters Focus Issue on the Gravitational Wave Transient Catalog

Report number: LIGO-P2500065

On the K-theory of matroids with Tutte coverings

Authors: Luigi Caputi, Sabino Di Trani

Abstract: The aim of this work is to explicitly compute the K-theory of the category of matroids with respect to the covering family of Tutte coverings. In particular, we show that this is equivalent to the K-theory spectrum of the category of graphic matroids on looped forests, with the covering family generated by isomorphisms. Further, we show that this yields an equivalence of $C_2$-spectra. The aim of this work is to explicitly compute the K-theory of the category of matroids with respect to the covering family of Tutte coverings. In particular, we show that this is equivalent to the K-theory spectrum of the category of graphic matroids on looped forests, with the covering family generated by isomorphisms. Further, we show that this yields an equivalence of $C_2$-spectra. △ Less

Submitted 18 March, 2026; originally announced March 2026.

Comments: 12 pages

MSC Class: 19M05; 05B35

First $^{94}$Nb($n,γ$) Measurement: Constraining the Nucleosynthetic Origin of $^{94}$Mo in Presolar Grains

Authors: J. Balibrea-Correa, J. Lerendegui-Marco, C. Domingo-Pardo, V. Babiano-Suarez, I. Ladarescu, M. Krtivcka, G. Cescutti, S. Cristallo, D. Vescovi, N. Liu, E. A. Maugeri, U. Köster, I. M\önch, A. Casanovas, V. Alcayne, D. Cano-Ott, E. Mendoza, O. Aberle, J. Andrzejewski, S. Altieri, S. Amaducci, M. Bacak, C. Beltrami, S. Bennett, A. P. Bernardes , et al. (114 additional authors not shown)

Abstract: Isotopic measurements of presolar silicon carbide grains from dying stars have revealed a puzzling overabundance of $^{94}$Mo that stellar nucleosynthesis models have failed to reproduce for two decades. This discrepancy challenged our understanding of the slow neutron-capture process ($s$-process) that forges approximately half of the elements heavier than iron. The key uncertainty lies at… ▽ More Isotopic measurements of presolar silicon carbide grains from dying stars have revealed a puzzling overabundance of $^{94}$Mo that stellar nucleosynthesis models have failed to reproduce for two decades. This discrepancy challenged our understanding of the slow neutron-capture process ($s$-process) that forges approximately half of the elements heavier than iron. The key uncertainty lies at $^{94}$Nb, a radiactive branching point where competition between neutron capture and beta decay governs the $^{94}$Mo production, yet the neutron-capture cross section had never been measured. Here we report the first experimental determination of the $^{94}$Nb(n,$γ$)$^{95}$Nb cross section important for Mo isotopic abundances. The measurement was enabled by a coordinated effort involving high-purity target preparation at Institute of Solid State and Materials Research (IFW) Dresden, radioactive sample production at the Institut Laue-Langevin (ILL) Grenoble, radiochemical characterization at Paul Scherrer Institute (PSI) Villigen, and the Time-of-Flight CERN n$\_$TOF facility using for the first time segmented total-energy detectors. Incorporation of the resulting Maxwellian-averaged cross section into fully coupled nucleosynthesis models of low-mass asymptotic giant branch (AGB) stars brings them into agreement with the presolar grain data. These results remove a major nuclear-physics input uncertainty at the $^{94}$Nb branching point and provide a firmer foundation for understanding the origin of $^{94}$Mo in the solar system. △ Less

Submitted 18 March, 2026; originally announced March 2026.

Search for Sterile Neutrinos with CUPID-0

Authors: O. Azzolini, J. W. Beeman, F. Bellini, M. Beretta, M. Biassoni, C. Brofferio, C. Bucci, S. Capelli, V. Caracciolo, L. Cardani, P. Carniti, N. Casali, E. Celi, D. Chiesa, M. Clemenza, I. Colantoni, O. Cremonesi, A. Cruciani, A. D'Addabbo, I. Dafinei, S. Di Domizio, F. Ferroni, L. Gironi, A. Giuliani, P. Gorla , et al. (28 additional authors not shown)

Abstract: Sterile neutrinos are well-motivated extensions of the Standard Model, introduced to address fundamental questions such as the origin of neutrino masses and the nature of dark matter. Exploiting the precise data reconstruction achieved by the CUPID-0 experiment, we searched for spectral distortions in the double $β$-decay of $^{82}$Se compatible with the emission of a sterile neutrino. The analysi… ▽ More Sterile neutrinos are well-motivated extensions of the Standard Model, introduced to address fundamental questions such as the origin of neutrino masses and the nature of dark matter. Exploiting the precise data reconstruction achieved by the CUPID-0 experiment, we searched for spectral distortions in the double $β$-decay of $^{82}$Se compatible with the emission of a sterile neutrino. The analysis relies on the construction of a detailed background model down to 200 keV, enabling an accurate characterization of the main sources of contamination. Using a Zn$^{82}$Se exposure of 9.95 kg$\cdot$yr, we explored sterile neutrino mass hypotheses between 0.5 MeV and 1.5 MeV. No evidence for a signal was observed in any scenario; therefore, we derived 90% C.I. upper limits on the active-sterile mixing probability $\sin^2θ$, obtaining the most stringent bound, $\sin^2θ<8\times 10^{-3}$, for a sterile neutrino mass of 0.7 MeV. △ Less

Submitted 18 March, 2026; originally announced March 2026.

Calculation for Electric Dipole Moments of Lepton and Neutron in the N-B-LSSM via the Mass Insertion Approximation

Authors: Shuang Di, Wei-Hang Zhang, Rong-Zhi Sun, Xing-Xing Dong, Guo-Zhu Ning, Shu-Min Zhao

Abstract: In the N-B-LSSM, we calculate the electric dipole moments (EDMs) of lepton and neutron at the one loop level via the Mass Insertion Approximation (MIA). In the Standard Model (SM), charge parity (CP) violation originates only from the single phase of the Cabibbo-Kobayashi-Maskawa (CKM) matrix, and the predicted EDMs of lepton and neutron are far below the current experimental upper limits. Thus, E… ▽ More In the N-B-LSSM, we calculate the electric dipole moments (EDMs) of lepton and neutron at the one loop level via the Mass Insertion Approximation (MIA). In the Standard Model (SM), charge parity (CP) violation originates only from the single phase of the Cabibbo-Kobayashi-Maskawa (CKM) matrix, and the predicted EDMs of lepton and neutron are far below the current experimental upper limits. Thus, EDMs serve as sensitive probes for exploring CP-violating phases in new physics. The N-B-LSSM extends the Minimal Supersymmetric Standard Model (MSSM) by introducing right-handed neutrino superfields and additional singlet Higgs superfields, which enriches the particle spectrum and the sources of CP violation. We derive the one loop analytical expressions for lepton and quark EDMs, and reveal their dependence on model parameters such as $g_{YB}$, $θ_{μ_H}$, $θ_{1'}$, $θ_{BB'}$ and $\tanβ$. Numerical analyses demonstrate that within a reasonable parameter space, the EDMs of leptons (electron, muon, tau) and the neutron can satisfy the current experimental limitations. This study provides a systematic theoretical tool and numerical reference for exploring CP violation and new physics under the N-B-LSSM. △ Less

Submitted 15 March, 2026; originally announced March 2026.

All-sky Searches for Continuous Gravitational Waves from Isolated Neutron Stars in the Data from the First Part of the Fourth LIGO-Virgo-KAGRA Observing Run

Authors: The LIGO Scientific Collaboration, the Virgo Collaboration, the KAGRA Collaboration, A. G. Abac, I. Abouelfettouh, F. Acernese, K. Ackley, A. Adam, C. Adamcewicz, S. Adhicary, D. Adhikari, N. Adhikari, R. X. Adhikari, V. K. Adkins, S. Afroz, A. Agapito, D. Agarwal, M. Agathos, N. Aggarwal, S. Aggarwal, O. D. Aguiar, I. -L. Ahrend, L. Aiello, A. Ain, P. Ajith , et al. (1804 additional authors not shown)

Abstract: We present results from an all-sky search for continuous gravitational waves, using three different methods applied to the first eight months of LIGO data from the fourth LIGO-Virgo-KAGRA Collaboration s observing run. We aim at signals potentially emitted by rotating, non-axisymmetric isolated neutron star in the Milky Way. The analysis spans a frequency range from 20 Hz to 2000 Hz and accommodat… ▽ More We present results from an all-sky search for continuous gravitational waves, using three different methods applied to the first eight months of LIGO data from the fourth LIGO-Virgo-KAGRA Collaboration s observing run. We aim at signals potentially emitted by rotating, non-axisymmetric isolated neutron star in the Milky Way. The analysis spans a frequency range from 20 Hz to 2000 Hz and accommodates frequency derivative magnitudes up to $10^{-8}$ Hz/s. No statistically significant periodic gravitational wave signals were detected. We establish 95% confidence-level (CL) frequentist upper limits on the dimensionless strain amplitudes. The most stringent population-averaged strain upper limits reach 9.7 $\times$ $10^{-26}$ near 290 Hz, matching the best previous constraints from 250 to $\sim$1700 Hz while extending coverage to a much broader spin-down range. At higher frequencies, the new limits improve upon previous results by factors of approximately $\sim$1.6. These constraints are applied to three astrophysical scenarios: 1) the distribution of galactic neutron stars as a function of spin frequency and ellipticity; 2) the contribution of millisecond pulsars to the GeV excess near the galactic center; and 3) the possible dark matter fraction composed of nearby inspiraling primordial binary black holes with asteroid-scale masses. △ Less

Submitted 14 March, 2026; originally announced March 2026.

Comments: 45 pages, 17 figures

Report number: LIGO-P2500416

Superhydrides on the way to ambient pressure: weak localization and persistent X-ray photoconductivity in BaSiH$_{8}$

Authors: Dmitrii V. Semenok, Di Zhou, Sven Luther, Toni Helm, Hirokazu Kadobayashi, Yuki Nakamoto, Katsuya Shimizu, Kirill S. Pervakov, Andrei V. Sadakov, Oleg A. Sobolevskiy, Vladimir M. Pudalov, Simone Di Cataldo, Roman Lucrezi, Lilia Boeri, Michele Galasso, Frederico G. Alabarse, Ivan A. Troyan, Viktor V. Struzhkin

Abstract: Reducing the stabilization pressure of superhydrides represents one of the most important challenges in hydrogen-saturated compound chemistry. Moving in this direction, we studied the Ba-Si-H system at 0-142 GPa using transport measurements, 1H nuclear magnetic resonance, single-crystal and powder X-ray diffraction in the temperature range of 4-317 K. We synthesized the previously predicted cubic… ▽ More Reducing the stabilization pressure of superhydrides represents one of the most important challenges in hydrogen-saturated compound chemistry. Moving in this direction, we studied the Ba-Si-H system at 0-142 GPa using transport measurements, 1H nuclear magnetic resonance, single-crystal and powder X-ray diffraction in the temperature range of 4-317 K. We synthesized the previously predicted cubic BaSiH$_{8}$ at pressures of 18-31 GPa. Remarkably, we demonstrate that BaSiH$_8$ remains stable upon decompression to ambient conditions and can be recovered from the diamond anvil cell. Obtained Ba-Si polyhydrides exhibit metallic and superconducting properties ($\textit{T$_c$}$ = 9 K, $\textit{B$_{c2}$}$(0)=13-16 T) at 142 GPa. However, at pressures below 50 GPa, these hydrides behave as degenerate semiconductors (bandgap < 0.4 meV) or poor metals with weak electron localization, negative magnetoresistance, photovoltaic effect, and persistent photoconductivity in the X-ray and visible range. Our work demonstrates the high-pressure synthesis of Ba-Si polyhydrides that remain stable upon decompression to ambient conditions, overcoming a critical bottleneck in superhydride chemistry and establishing a foundation for practical applications in hydrogen storage. △ Less

Submitted 14 March, 2026; originally announced March 2026.

NCCL EP: Towards a Unified Expert Parallel Communication API for NCCL

Authors: Amos Goldman, Nimrod Boker, Maayan Sheraizin, Nimrod Admoni, Artem Polyakov, Subhadeep Bhattacharya, Fan Yu, Kai Sun, Georgios Theodorakis, Hsin-Chun Yin, Peter-Jan Gootzen, Aamir Shafi, Assaf Ravid, Salvatore Di Girolamo, James Dinan, Xiaofan Li, Manjunath Gorentla Venkata, Gil Bloch

Abstract: Mixture-of-Experts (MoE) architectures have become essential for scaling large language models, driving the development of specialized device-initiated communication libraries such as DeepEP, Hybrid-EP, and others. These libraries demonstrate the performance benefits of GPU-initiated RDMA for MoE dispatch and combine operations. This paper presents NCCL EP (Expert Parallelism), a ground-up MoE c… ▽ More Mixture-of-Experts (MoE) architectures have become essential for scaling large language models, driving the development of specialized device-initiated communication libraries such as DeepEP, Hybrid-EP, and others. These libraries demonstrate the performance benefits of GPU-initiated RDMA for MoE dispatch and combine operations. This paper presents NCCL EP (Expert Parallelism), a ground-up MoE communication library built entirely on NCCL's Device API. NCCL EP provides unified ncclEpDispatch and ncclEpCombine primitives with both C and Python interfaces, supporting Low-Latency (LL) mode for inference decoding and High-Throughput (HT) mode for training and inference prefill. LL targets small batch sizes (1-128 tokens) using direct all-to-all RDMA+NVLink mesh connectivity with double-buffered communication for overlapping dispatch and combine phases. HT targets large batches (4096+ tokens) using hierarchical communication that aggregates tokens within NVLink domains before inter-node RDMA transmission. Both modes leverage Device API for both intra- and inter-node communications, taking advantage of its topology awareness and optimized GPU-initiated implementation. We evaluate NCCL EP on an H100-based cluster across multi-node configurations, demonstrating competitive LL kernel performance and presenting end-to-end results with vLLM integration. By building MoE communication natively within NCCL, NCCL EP provides a supported path for expert parallelism on current and emerging NVIDIA platforms. △ Less

Submitted 2 April, 2026; v1 submitted 13 March, 2026; originally announced March 2026.

Comments: 13 pages, 8 figures, 7 tables

MSC Class: 68W10; 68M10; 65Y05 ACM Class: C.2.1; C.2.4; C.1.2; C.1.4

Euclid preparation. Far-infrared predictions for Euclid galaxy catalogues: cluster, protocluster, and field

Authors: Euclid Collaboration, A. Parmar, D. L. Clements, M. Bolzonella, O. Cucciati, L. Pozzetti, H. Dannerbauer, G. Castignani, S. Serjeant, L. Wang, R. Hill, D. Scott, J. G. Sorce, M. Magliocchetti, F. Pace, T. T. Thai, N. Aghanim, B. Altieri, S. Andreon, N. Auricchio, C. Baccigalupi, M. Baldi, S. Bardelli, A. Biviano, W. Bon , et al. (275 additional authors not shown)

Abstract: The MAMBO mock galaxy catalogue, based on the Millennium Simulation with empirically assigned galaxy properties, provides predictions of FIR fluxes and physical parameters of Euclid-detectable galaxies. Predicted FIR flux distributions confirm that only the brightest Euclid sources will be detectable in existing FIR surveys. We employ stacking to measure the mean dust properties as a function of s… ▽ More The MAMBO mock galaxy catalogue, based on the Millennium Simulation with empirically assigned galaxy properties, provides predictions of FIR fluxes and physical parameters of Euclid-detectable galaxies. Predicted FIR flux distributions confirm that only the brightest Euclid sources will be detectable in existing FIR surveys. We employ stacking to measure the mean dust properties as a function of stellar mass and redshift. We find dust temperatures and infrared luminosities increase with redshift across all mass bins, while dust masses remain roughly constant. FIR number counts from MAMBO show overall good agreement with observations, and the total infrared luminosity function reproduces published estimates across most redshift ranges, extending to z~10. Comparing the Euclid Wide and Deep Surveys, we find that the EDS recovers the total IRLF to fainter luminosities and higher redshifts (up to z~6 in $I_E$), although its detectability falls below 80% at z>4, whereas the EWS becomes strongly incomplete beyond z~2. We also examine the dependence of the IRLF on environment. Schechter fits indicate that the faint-end slope $α$ flattens with redshift for cluster and protocluster galaxies, while remaining approximately constant for field populations. Imposing additional detection limits from Herschel-PACS and SPIRE shows that only the most luminous ($L_{IR}$ > $10^{12.5}$ $L_{\odot}$) galaxies remain detectable at z~4, but the limited MAMBO area (3.14$deg^2$) is inadequate for statistically robust (>3$σ$) constraints. Survey areas at least 30 times larger are required. Overall, the MAMBO FIR extension reproduces key number count and IRLF trends, provides realistic predictions for FIR-detected Euclid galaxies, and highlights the importance of synergies with current and future FIR/sub-mm facilities to probe environmental dependence with sufficient depth and area. △ Less

Submitted 13 March, 2026; originally announced March 2026.

Comments: 25 pages, 17 figures, submitted to A&A

Euclid preparation. Simulated galaxy catalogues for non-standard cosmological models

Authors: Euclid Collaboration, M. -A. Breton, P. Fosalba, S. Avila, M. Baldi, C. Carbone, M. Kärcher, G. Rácz, M. Bolzonella, F. J. Castander, C. Giocoli, K. Koyama, A. M. C. Le Brun, L. Pozzetti, A. G. Adame, V. Gonzalez-Perez, G. Yepes, B. Altieri, S. Andreon, C. Baccigalupi, S. Bardelli, P. Battaglia, A. Biviano, E. Branchini, M. Brescia , et al. (267 additional authors not shown)

Abstract: Stage-IV galaxy surveys will provide the opportunity to test cosmological models and the underlying theory of gravity with unparalleled precision. In this context, it is crucial for the Euclid mission to leverage its spectroscopic and photometric probes to systematically investigate and incorporate non-standard cosmological models, including modified gravity, alternative dark energy scenarios, mas… ▽ More Stage-IV galaxy surveys will provide the opportunity to test cosmological models and the underlying theory of gravity with unparalleled precision. In this context, it is crucial for the Euclid mission to leverage its spectroscopic and photometric probes to systematically investigate and incorporate non-standard cosmological models, including modified gravity, alternative dark energy scenarios, massive neutrinos, and primordial non-Gaussianity. We produce and release publicly simulated galaxy catalogues from a broad suite of non-standard cosmological simulations, which we processed through a model-independent analytical pipeline, making use of Rockstar for halo identification, and a modified version of the SciPic library for the galaxy-halo connection using the halo occupation distribution framework. We investigate their galaxy-clustering characteristics via the multipoles of the 2PCF in redshift space and VDG, a highly performant model for galaxy clustering. Across a wide range of models, the linear growth rate multiplied by the matter density within spheres of radius 12,Mpc, fs12, exhibits a notable robustness to the choice of cosmological template. Compared to previous works, our study extends this result to numerous scenarios with markedly distinct gravitational or dark energy dynamics. We find that the most of the scatter in cosmological parameter inference already appears when using the cosmological model of the simulations as templates. Using a `wrong' template can also introduce an additional scatter, although with smaller amplitude. Often, we find deviations much larger than error bars, meaning that the Gaussian approximation for the covariance might need to be further studied. Future cosmological investigations must broaden their scope to include a diverse array of non-standard theoretical frameworks, extending beyond LCDM and rudimentary dynamic dark energy models. △ Less

Submitted 13 March, 2026; originally announced March 2026.

Comments: 14+9 pages, 3+4 figures, submitted

OmniStream: Mastering Perception, Reconstruction and Action in Continuous Streams

Authors: Yibin Yan, Jilan Xu, Shangzhe Di, Haoning Wu, Weidi Xie

Abstract: Modern visual agents require representations that are general, causal, and physically structured to operate in real-time streaming environments. However, current vision foundation models remain fragmented, specializing narrowly in image semantic perception, offline temporal modeling, or spatial geometry. This paper introduces OmniStream, a unified streaming visual backbone that effectively perceiv… ▽ More Modern visual agents require representations that are general, causal, and physically structured to operate in real-time streaming environments. However, current vision foundation models remain fragmented, specializing narrowly in image semantic perception, offline temporal modeling, or spatial geometry. This paper introduces OmniStream, a unified streaming visual backbone that effectively perceives, reconstructs, and acts from diverse visual inputs. By incorporating causal spatiotemporal attention and 3D rotary positional embeddings (3D-RoPE), our model supports efficient, frame-by-frame online processing of video streams via a persistent KV-cache. We pre-train OmniStream using a synergistic multi-task framework coupling static and temporal representation learning, streaming geometric reconstruction, and vision-language alignment on 29 datasets. Extensive evaluations show that, even with a strictly frozen backbone, OmniStream achieves consistently competitive performance with specialized experts across image and video probing, streaming geometric reconstruction, complex video and spatial reasoning, as well as robotic manipulation (unseen at training). Rather than pursuing benchmark-specific dominance, our work demonstrates the viability of training a single, versatile vision backbone that generalizes across semantic, spatial, and temporal reasoning, i.e., a more meaningful step toward general-purpose visual understanding for interactive and embodied agents. △ Less

Submitted 12 March, 2026; originally announced March 2026.

Comments: Technical Report. Project Page: https://go2heart.github.io/omnistream/

Enhanced Seismicity Monitoring in the Rapid Scientific Response to the 2025 Santorini Crisis

Authors: Margarita Segou, Foteini Dervisi, Xing Tan, Rajat Choudhary, Patricia Martínez-Garzón, Francesco Scotto di Uccio, Gregory Beroza, Genny Giacomuzzi, Claudio Chiarabba, Wayne Shelley, Stephanie Prejean, Jeremy Pesicek, John J. Wellik, Marco Bohnhoff, David Pyle, Costas Synolakis, Tom Parsons, Athanassios Ganas, William Ellsworth, Brian Baptie, Gaetano Festa, Piero Poli, Warner Marzocchi

Abstract: We used a deep learning workflow to enhance earthquake detection during the 2025 seismic unrest between Santorini and Amorgos islands to track the evolution of the crisis in near real-time. We analysed the continuous seismic waveforms daily (1/2 - 3/3/25) as the crisis unfolded. Our analysis enhanced the earthquake catalogue from around 4,000 to 80,000 earthquakes. The enhanced catalogue allowed t… ▽ More We used a deep learning workflow to enhance earthquake detection during the 2025 seismic unrest between Santorini and Amorgos islands to track the evolution of the crisis in near real-time. We analysed the continuous seismic waveforms daily (1/2 - 3/3/25) as the crisis unfolded. Our analysis enhanced the earthquake catalogue from around 4,000 to 80,000 earthquakes. The enhanced catalogue allowed this international expert group to identify the volcanic-tectonic character, clearly revealing burst-like, spasmodic seismicity swarms, which is a pattern associated with fluid-driven processes from early stages of the crisis. Detailed moment tensor inversions in early events characterised by a significant non-double couple component indicated the involvement of magmatic or high-pressure hydrothermal fluids driving the unrest. Concurrent DL-enhanced tomography efforts identified a third, deep magmatic reservoir beneath Anydros Islet, consistent with pressure-driven processes. To date, volcanic-tectonic swarms in which >200 earthquakes of ML > 4 occurred within only a few weeks, largely within episodic bursts of seismicity, have not been observed elsewhere. △ Less

Submitted 11 March, 2026; originally announced March 2026.

Thousand-GPU Large-Scale Training and Optimization Recipe for AI-Native Cloud Embodied Intelligence Infrastructure

Authors: Yongjian Guo, Yunxuan Ma, Haoran Sun, Zhong Guan, Shuai Di, Jing Long, Wanting Xu, Xiaodong Bai, Wen Huang, Yucheng Guo, Chen Zhou, Qiming Yang, Mingxi Luo, Tianyun Zhao, Hedan Yang, Song Wang, Xiaomeng Tian, Xiaolong Xiang, Zhen Sun, Yu Wei, Luqiao Wang, Yuzhen Li, Chenfeng Gu, Junwu Xiong, Yicheng Gong

Abstract: Embodied intelligence is a key step towards Artificial General Intelligence (AGI), yet its development faces multiple challenges including data, frameworks, infrastructure, and evaluation systems. To address these issues, we have, for the first time in the industry, launched a cloud-based, thousand-GPU distributed training platform for embodied intelligence, built upon the widely adopted LeRobot f… ▽ More Embodied intelligence is a key step towards Artificial General Intelligence (AGI), yet its development faces multiple challenges including data, frameworks, infrastructure, and evaluation systems. To address these issues, we have, for the first time in the industry, launched a cloud-based, thousand-GPU distributed training platform for embodied intelligence, built upon the widely adopted LeRobot framework, and have systematically overcome bottlenecks across the entire pipeline. At the data layer, we have restructured the data pipeline to optimize the flow of embodied training data. In terms of training, for the GR00T-N1.5 model, utilizing thousand-GPU clusters and data at the scale of hundreds of millions, the single-round training time has been reduced from 15 hours to just 22 minutes, achieving a 40-fold speedup. At the model layer, by combining variable-length FlashAttention and Data Packing, we have moved from sample redundancy to sequence integration, resulting in a 188% speed increase; π-0.5 attention optimization has accelerated training by 165%; and FP8 quantization has delivered a 140% speedup. On the infrastructure side, relying on high-performance storage, a 3.2T RDMA network, and a Ray-driven elastic AI data lake, we have achieved deep synergy among data, storage, communication, and computation. We have also built an end-to-end evaluation system, creating a closed loop from training to simulation to assessment. This framework has already been fully validated on thousand-GPU clusters, laying a crucial technical foundation for the development and application of next-generation autonomous intelligent robots, and is expected to accelerate the arrival of the era of human-machine integration. △ Less

Submitted 18 March, 2026; v1 submitted 11 March, 2026; originally announced March 2026.

Rate-Distortion Bounds for Heterogeneous Random Fields on Finite Lattices

Authors: Sujata Sinha, Vishwas Rao, Robert Underwood, David Lenz, Sheng Di, Franck Cappello, Lingjia Liu

Abstract: Since Shannon's foundational work, rate-distortion theory has defined the fundamental limits of lossy compression. Classical results, derived for memoryless and stationary ergodic sources in the asymptotic regime, have shaped both transform and predictive coding architectures, as well as practical standards such as JPEG. Finite-blocklength refinements, initiated by the non-asymptotic achievability… ▽ More Since Shannon's foundational work, rate-distortion theory has defined the fundamental limits of lossy compression. Classical results, derived for memoryless and stationary ergodic sources in the asymptotic regime, have shaped both transform and predictive coding architectures, as well as practical standards such as JPEG. Finite-blocklength refinements, initiated by the non-asymptotic achievability and converse bounds of Kostina and Verdu, provide precise characterizations under excess-distortion probability constraints, but primarily for memoryless or statistically homogeneous models. In contrast, error-bounded practical lossy compressors for scientific computing, such as SZ, ZFP, MGARD, and SPERR, are designed for finite, high-dimensional, spatially correlated, and statistically heterogeneous random fields. These compressors partition data into fixed-size tiles that are processed independently, making tile size a central architectural constraint. Structural heterogeneity, finite lattice effects, and tiling constraints are not addressed by existing finite-blocklength analyses. This paper introduces a finite-blocklength rate-distortion framework for heterogeneous random fields on finite lattices, explicitly accounting for the tile-based architectures used in high-performance scientific compressors. The field is modeled as piecewise homogeneous with regionwise stationary second-order statistics, and tiling constraints are incorporated directly into the source model. Under an excess-distortion probability criterion, we establish non-asymptotic achievability, converse bounds and derive a second-order expansion that quantifies the impact of spatial correlation, region geometry, heterogeneity, and tile size on the rate and dispersion. △ Less

Submitted 10 March, 2026; originally announced March 2026.

ToolRosetta: Bridging Open-Source Repositories and Large Language Model Agents through Automated Tool Standardization

Authors: Shimin Di, Xujie Yuan, Hanghui Guo, Chaoqian Ouyang, Zhangze Chen, Ling Yue, Libin Zheng, Jia Zhu, Shaowu Pan, Jian Yin, Min-Ling Zhang, Yong Rui

Abstract: Reusing and invoking existing code remains costly and unreliable, as most practical tools are embedded in heterogeneous code repositories and lack standardized, executable interfaces. Although large language models (LLMs) and Model Context Protocol (MCP)-based tool invocation frameworks enable natural language task execution, current approaches rely heavily on manual tool curation and standardizat… ▽ More Reusing and invoking existing code remains costly and unreliable, as most practical tools are embedded in heterogeneous code repositories and lack standardized, executable interfaces. Although large language models (LLMs) and Model Context Protocol (MCP)-based tool invocation frameworks enable natural language task execution, current approaches rely heavily on manual tool curation and standardization, which fundamentally limits scalability. In this paper, we propose ToolRosetta, a unified framework that automatically translates open-source code repositories and APIs into MCP-compatible tools that can be reliably invoked by LLMs. Given a user task, ToolRosetta autonomously plans toolchains, identifies relevant codebases, and converts them into executable MCP services, enabling end-to-end task completion with minimal human intervention. In addition, ToolRosetta incorporates a security inspection layer to mitigate risks inherent in executing arbitrary code. Extensive experiments across diverse scientific domains demonstrate that ToolRosetta can automatically standardize a large number of open-source tools and reduce the human effort required for code reproduction and deployment. Notably, by seamlessly leveraging specialized open-source tools, ToolRosetta-powered agents consistently improve task completion performance compared to commercial LLMs and existing agent systems. △ Less

Submitted 10 March, 2026; originally announced March 2026.

Comments: 20 pages

DC-Merge: Improving Model Merging with Directional Consistency

Authors: Han-Chen Zhang, Zi-Hao Zhou, Mao-Lin Luo, Shimin Di, Min-Ling Zhang, Tong Wei

Abstract: Model merging aims to integrate multiple task-adapted models into a unified model that preserves the knowledge of each task. In this paper, we identify that the key to this knowledge retention lies in maintaining the directional consistency of singular spaces between merged multi-task vector and individual task vectors. However, this consistency is frequently compromised by two issues: i) an imbal… ▽ More Model merging aims to integrate multiple task-adapted models into a unified model that preserves the knowledge of each task. In this paper, we identify that the key to this knowledge retention lies in maintaining the directional consistency of singular spaces between merged multi-task vector and individual task vectors. However, this consistency is frequently compromised by two issues: i) an imbalanced energy distribution within task vectors, where a small fraction of singular values dominate the total energy, leading to the neglect of semantically important but weaker components upon merging, and ii) the geometric inconsistency of task vectors in parameter space, which causes direct merging to distort their underlying directional geometry. To address these challenges, we propose DC-Merge, a method for directional-consistent model merging. It first balances the energy distribution of each task vector by smoothing its singular values, ensuring all knowledge components are adequately represented. These energy-balanced vectors are then projected onto a shared orthogonal subspace to align their directional geometries with minimal reconstruction error. Finally, the aligned vectors are aggregated in the shared orthogonal subspace and projected back to the original parameter space. Extensive experiments on vision and vision-language benchmarks show that DC-Merge consistently achieves state-of-the-art performance in both full fine-tuning and LoRA settings. The implementation code is available at https://github.com/Tobeginwith/DC-Merge. △ Less

Submitted 15 March, 2026; v1 submitted 6 March, 2026; originally announced March 2026.

Comments: Accepted by CVPR 2026 Main Track

Atmospheric neutrino constraints on Lorentz invariance violation with the first six detection units of KM3NeT/ORCA

Authors: KM3NeT Collaboration, O. Adriani, A. Albert, A. R. Alhebsi, S. Alshalloudi, S. Alves Garre, F. Ameli, M. Andre, L. Aphecetche, M. Ardid, S. Ardid, J. Aublin, F. Badaracco, L. Bailly-Salins, B. Baret, A. Bariego-Quintana, L. Barigione, M. Barnard, Y. Becherini, M. Bendahman, F. Benfenati Gualandi, M. Benhassi, D. M. Benoit, Z. Beňušová, E. Berbee , et al. (272 additional authors not shown)

Abstract: Lorentz invariance is a fundamental symmetry underlying both the Standard Model of particle physics and General Relativity. Testing its validity provides a direct means of searching for new physics emerging near the Planck scale. A search for isotropic Lorentz invariance violation with 1.4 years of atmospheric neutrino data collected by a partial configuration of the KM3NeT/ORCA detector comprisin… ▽ More Lorentz invariance is a fundamental symmetry underlying both the Standard Model of particle physics and General Relativity. Testing its validity provides a direct means of searching for new physics emerging near the Planck scale. A search for isotropic Lorentz invariance violation with 1.4 years of atmospheric neutrino data collected by a partial configuration of the KM3NeT/ORCA detector comprising six detection units is presented. No evidence for such violation is found; thus, competitive limits are set on a subset of isotropic Lorentz invariance violating coefficients, which complement and extend existing experimental constraints. △ Less

Submitted 4 March, 2026; originally announced March 2026.

Safe Whole-Body Loco-Manipulation via Combined Model and Learning-based Control

Authors: Alexander Schperberg, Yeping Wang, Stefano Di Cairano

Abstract: Simultaneous locomotion and manipulation enables robots to interact with their environment beyond the constraints of a fixed base. However, coordinating legged locomotion with arm manipulation, while considering safety and compliance during contact interaction remains challenging. To this end, we propose a whole-body controller that combines a model-based admittance control for the manipulator arm… ▽ More Simultaneous locomotion and manipulation enables robots to interact with their environment beyond the constraints of a fixed base. However, coordinating legged locomotion with arm manipulation, while considering safety and compliance during contact interaction remains challenging. To this end, we propose a whole-body controller that combines a model-based admittance control for the manipulator arm with a Reinforcement Learning (RL) policy for legged locomotion. The admittance controller maps external wrenches--such as those applied by a human during physical interaction--into desired end-effector velocities, allowing for compliant behavior. The velocities are tracked jointly by the arm and leg controllers, enabling a unified 6-DoF force response. The model-based design permits accurate force control and safety guarantees via a Reference Governor (RG), while robustness is further improved by a Kalman filter enhanced with neural networks for reliable base velocity estimation. We validate our approach in both simulation and hardware using the Unitree Go2 quadruped robot with a 6-DoF arm and wrist-mounted 6-DoF Force/Torque sensor. Results demonstrate accurate tracking of interaction-driven velocities, compliant behavior, and safe, reliable performance in dynamic settings. △ Less

Submitted 2 March, 2026; originally announced March 2026.

Comments: Accepted to IEEE International Conference on Robotics and Automation (ICRA), June 2026, in Vienna, Austria

The ESPRESSO Redshift Drift Experiment III -- The Third Epoch of QSO J052915.80-435152.0

Authors: Andrea Trost, Catarina M. J. Marques, S. Cristiani, Guido Cupani, Simona Di Stefano, Valentina D'Odorico, Francesco Guarneri, Carlos J. A. P. Martins, Dinko Milaković, Luca Pasquini, Ricardo Génova Santos, Paolo Molaro, Michael T. Murphy, Nelson J. Nunes, Tobias M. Schmidt, Yann Alibert, Konstantina Boutsia, Giorgio Calderone, J. I. González Hernández, Andrea Grazian, Gaspare Lo Curto, Enric Palle, Francesco Pepe, Matteo Porru, Nuno C. Santos , et al. (3 additional authors not shown)

Abstract: The Sandage-Loeb test probes cosmic expansion directly by measuring the redshift drift in quasar absorption features in a model-independent way. In this series of papers, we have launched an observational campaign to assess whether current instrumentation is capable of measuring this effect and what systematic effects might interfere with a detection. We report the observations and analysis of the… ▽ More The Sandage-Loeb test probes cosmic expansion directly by measuring the redshift drift in quasar absorption features in a model-independent way. In this series of papers, we have launched an observational campaign to assess whether current instrumentation is capable of measuring this effect and what systematic effects might interfere with a detection. We report the observations and analysis of the third epoch of ESPRESSO observations of the bright quasar J052915.80-435152.0 (SB2, z=3.962), extending the temporal baseline to $\sim2$ years, and providing the tightest constraints on the redshift drift in the series so far. We acquired 9.5 hours of ESPRESSO observations, complementing the 12 hours presented in the first paper of the series, with one year of separation from the second epoch. The complete dataset was analysed and compared to spline-based Lyman-$α$ forest models calibrated on simulations, to measure the presence of any velocity drift among the spectra. The measurement was carried out with two independent methods. Both approaches give a consistent null result, $\dot{v} = -3.5 \pm 3.6 ~{\rm m s^{-1} yr^{-1}}$ (or $\dot{z} = (-5.3\pm5.6)\times 10^{-8}~{\rm yr^{-1}}$ in redshift space), in agreement with $Λ$CDM expectations, systematic effects remain subdominant at the present level of noise. By extrapolating the results from the observed sightline to the complete QUBRICS Golden Sample, we show that ESPRESSO alone could detect the signal on century timescales, while a joint ESPRESSO+ANDES programme would reach first detection before 2080. A future analysis of the other quasars of the QUBRICS Golden Sample is required to improve this estimate. We show that the program would greatly benefit from a complementary effort with radio facilities targeting low-z HI 21 cm absorption lines. Such synergy could reduce the experiments' timeline by up to $\sim10$ years. △ Less

Submitted 2 March, 2026; originally announced March 2026.

Comments: 13 pages, 9 figures, accepted for publication in Astronomy & Astrophysics, abstract abridged

RxnNano:Training Compact LLMs for Chemical Reaction and Retrosynthesis Prediction via Hierarchical Curriculum Learning

Authors: Ran Li, Shimin Di, Haowei LI, Luanshi Bu, Jiachuan Wang, Wangze Ni, Lei Chen

Abstract: Chemical reaction prediction is pivotal for accelerating drug discovery and synthesis planning. Despite advances in data-driven models, current approaches are hindered by an overemphasis on parameter and dataset scaling. Some methods coupled with evaluation techniques that bypass fundamental challenges in reaction representation and fail to capture deep chemical intuition like reaction common sens… ▽ More Chemical reaction prediction is pivotal for accelerating drug discovery and synthesis planning. Despite advances in data-driven models, current approaches are hindered by an overemphasis on parameter and dataset scaling. Some methods coupled with evaluation techniques that bypass fundamental challenges in reaction representation and fail to capture deep chemical intuition like reaction common sense and {topological atom mapping logic}. We argue that the core challenge lies in instilling these knowledge into the models. To this end, we propose a unified framework that prioritizes chemical understanding over scale through three key innovations: (1) a {Latent Chemical Consistency} objective that models reactions as movements on a continuous chemical manifold, ensuring reversible and physically plausible transformations; (2) a {Hierarchical Cognitive Curriculum} that trains the model through progressive stages, from syntax mastery to semantic reasoning, building robust chemical intuition; (3) {Atom-Map Permutation Invariance (AMPI)}, which force the model to learn invariant relational topology and balance multi-task learning. (4)and structured plan-based reasoning to improve the performance of the LLMs. Our compact {0.5B-parameter model}, \textbf{RxnNano} significantly outperforms fine-tuned LLMs ten times larger (>7B) and all the domain baselines, achieving a 23.5\% Top-1 accuracy improvement on rigorous benchmarks without test-time augmentation. https://github.com/rlisml/RxnNano. △ Less

Submitted 10 February, 2026; originally announced March 2026.

ImpCresst -- A versatile simulation tool focusing on solid-state detectors at keV energies

Authors: G. Angloher, S. Banik, A. Bento, A. Bertolini, R. Breier, C. Bucci, J. Burkhart, L. Burmeister, L. Canonica, F. Casadei, E. Cipelli, S. Di Lorenzo, J. Dohm, F. Dominsky, A. Erb, E. Fascione, F. von Feilitzsch, S. Fichtinger, D. Fuchs, A. Fuss, V. M. Ghete, P. Gorla, P. V. Guillaumon, D. Hauff, M. Ješkovský , et al. (35 additional authors not shown)

Abstract: We present ImpCresst, a Geant4-based Monte Carlo tool to simulate backgrounds from natural and cosmogenic radionuclides, and calibration signals in solid-state detectors and their response to it. It is tuned for a fast-evolving and heterogeneous detector environment with a focus on physics at the keV range. This tool was originally developed and validated by the CRESST collaboration; however, its… ▽ More We present ImpCresst, a Geant4-based Monte Carlo tool to simulate backgrounds from natural and cosmogenic radionuclides, and calibration signals in solid-state detectors and their response to it. It is tuned for a fast-evolving and heterogeneous detector environment with a focus on physics at the keV range. This tool was originally developed and validated by the CRESST collaboration; however, its flexibility and configurability make it suitable for other experiments with similar requirements. Key features of ImpCresst include the dynamic geometry implementation directly from CAD files, ROOT-based data persistency of the whole event topology and automatic metadata annotation for data provenance, and interfaces to various particle generators, particularly for radiogenic and cosmogenic radionuclides. It includes also a newly developed particle generator for radioactive bulk and surface contaminations which is completely independent of any user defined confinement volumes. The auxiliary tool CresstDS applies detector-specific energy and time resolution based on a user-provided data set of empirical parameterization. We discuss also how to manage an ImpCresst based workflow in an HPC environment based on Apptainer and nextflow. △ Less

Submitted 2 March, 2026; originally announced March 2026.

SPARe: Stacked Parallelism with Adaptive Reordering for Fault-Tolerant LLM Pretraining Systems with 100k+ GPUs

Authors: Jin Lee, Zhonghao Chen, Xuhang He, Robert Underwood, Bogdan Nicolae, Franck Cappello, Xiaoyi Lu, Sheng Di, Zheng Zhang

Abstract: In large-scale LLM pre-training systems with 100k+ GPUs, failures become the norm rather than the exception, and restart costs can dominate wall-clock training time. However, existing fault-tolerance mechanisms are largely unprepared for this restart-dominant regime. To address this challenge, we propose SPARe - Stacked Parallelism with Adaptive Reordering - a fault-tolerance framework that masks… ▽ More In large-scale LLM pre-training systems with 100k+ GPUs, failures become the norm rather than the exception, and restart costs can dominate wall-clock training time. However, existing fault-tolerance mechanisms are largely unprepared for this restart-dominant regime. To address this challenge, we propose SPARe - Stacked Parallelism with Adaptive Reordering - a fault-tolerance framework that masks node failures during gradient synchronization by stacking redundant data shards across parallelism groups and adaptively reordering execution. SPARe achieves availability comparable to traditional replication while maintaining near-constant computation overhead of only 2~3x, even under high redundancy where traditional replication would require linearly inflating overhead. We derive closed-form expressions for endurable failure count and computation overhead, validate them via SimGrid-based discrete-event simulation, and jointly optimize redundancy and checkpointing to minimize time-to-train. At extreme scale with up to 600k GPUs, SPARe reduces time-to-train by 40~50% compared to traditional replication. △ Less

Submitted 27 February, 2026; originally announced March 2026.

LUMOS: Democratizing SciML Workflows with L0-Regularized Learning for Unified Feature and Parameter Adaptation

Authors: Shouwei Gao, Xu Zheng, Dongsheng Luo, Sheng Di, Wenqian Dong

Abstract: The rapid growth of scientific machine learning (SciML) has accelerated discovery across diverse domains, yet designing effective SciML models remains a challenging task. In practice, building such models often requires substantial prior knowledge and manual expertise, particularly in determining which input features to use and how large the model should be. We introduce LUMOS, an end-to-end frame… ▽ More The rapid growth of scientific machine learning (SciML) has accelerated discovery across diverse domains, yet designing effective SciML models remains a challenging task. In practice, building such models often requires substantial prior knowledge and manual expertise, particularly in determining which input features to use and how large the model should be. We introduce LUMOS, an end-to-end framework based on L0-regularized learning that unifies feature selection and model pruning to democratize SciML model design. By employing semi-stochastic gating and reparameterization techniques, LUMOS dynamically selects informative features and prunes redundant parameters during training, reducing the reliance on manual tuning while maintaining predictive accuracy. We evaluate LUMOS across 13 diverse SciML workloads, including cosmology and molecular sciences, and demonstrate its effectiveness and generalizability. Experiments on 13 SciML models show that LUMOS achieves 71.45% parameter reduction and a 6.4x inference speedup on average. Furthermore, Distributed Data Parallel (DDP) training on up to eight GPUs confirms the scalability of △ Less

Submitted 25 February, 2026; originally announced February 2026.

Mitigating Artifacts in Pre-quantization Based Scientific Data Compressors with Quantization-aware Interpolation

Authors: Pu Jiao, Sheng Di, Jiannan Tian, Mingze Xia, Xuan Wu, Yang Zhang, Xin Liang, Franck Cappello

Abstract: Error-bounded lossy compression has been regarded as a promising way to address the ever-increasing amount of scientific data in today's high-performance computing systems. Pre-quantization, a critical technique to remove sequential dependency and enable high parallelism, is widely used to design and develop high-throughput error-controlled data compressors. Despite the extremely high throughput o… ▽ More Error-bounded lossy compression has been regarded as a promising way to address the ever-increasing amount of scientific data in today's high-performance computing systems. Pre-quantization, a critical technique to remove sequential dependency and enable high parallelism, is widely used to design and develop high-throughput error-controlled data compressors. Despite the extremely high throughput of pre-quantization based compressors, they generally suffer from low data quality with medium or large user-specified error bounds. In this paper, we investigate the artifacts generated by pre-quantization based compressors and propose a novel algorithm to mitigate them. Our contributions are fourfold: (1) We carefully characterize the artifacts in pre-quantization based compressors to understand the correlation between the quantization index and compression error; (2) We propose a novel quantization-aware interpolation algorithm to improve the decompressed data; (3) We parallelize our algorithm in both shared-memory and distributed-memory environments to obtain high performance; (4) We evaluate our algorithm and validate it with two leading pre-quantization based compressors using five real-world datasets. Experiments demonstrate that our artifact mitigation algorithm can effectively improve the quality of decompressed data produced by pre-quantization based compressors while maintaining their high compression throughput. △ Less

Submitted 23 February, 2026; originally announced February 2026.

TopoSZp: Lightweight Topology-Aware Error-controlled Compression for Scientific Data

Authors: Tripti Agarwal, Sheng Di, Xin Liang, Zhaoyuan Su, Yuxiao Li, Ganesh Gopalakrishnan, Hanqi Guo, Franck Cappello

Abstract: Error-bounded lossy compression is essential for managing the massive data volumes produced by large-scale HPC simulations. While state-of-the-art compressors such as SZ and ZFP provide strong numerical error guarantees, they often fail to preserve topological structures (example, minima, maxima, and saddle points) that are critical for scientific analysis. Existing topology-aware compressors addr… ▽ More Error-bounded lossy compression is essential for managing the massive data volumes produced by large-scale HPC simulations. While state-of-the-art compressors such as SZ and ZFP provide strong numerical error guarantees, they often fail to preserve topological structures (example, minima, maxima, and saddle points) that are critical for scientific analysis. Existing topology-aware compressors address this limitation but incur substantial computational overhead. We present TopoSZp, a lightweight, topology-aware, error-controlled lossy compressor that preserves critical points and their relationships while maintaining high compression and decompression performance. Built on the high-throughput SZp compressor, TopoSZp integrates efficient critical point detection, local ordering preservation, and targeted saddle point refinement, all within a relaxed but strictly enforced error bound. Experimental results on real-world scientific datasets show that TopoSZp achieves 3 to 100 times fewer non-preserved critical points, introduces no false positives or incorrect critical point types, and delivers 100 to 10000 times faster compression and 10 to 500 times faster decompression compared to existing topology-aware compressors, while maintaining competitive compression ratios. △ Less

Submitted 19 February, 2026; originally announced February 2026.

Comments: 11 pages, 9 figures, 2 tables

Euclid preparation. Impact of galaxy intrinsic alignment modelling choices on Euclid 3x2pt cosmology

Authors: Euclid Collaboration, D. Navarro-Gironés, I. Tutusaus, M. Crocce, S. Gouyou Beauchamps, R. Paviot, B. Joachimi, J. Ruiz-Zapatero, D. Sciotti, N. Tessore, G. Cañas-Herrera, P. Carrilho, J. M. Coloma-Nadal, H. Hoekstra, A. Porredon, B. Altieri, S. Andreon, C. Baccigalupi, M. Baldi, S. Bardelli, A. Biviano, E. Branchini, M. Brescia, S. Camera, V. Capobianco , et al. (251 additional authors not shown)

Abstract: The Euclid galaxy survey will provide unprecedented constraints on cosmology, but achieving unbiased results will require an optimal characterisation and mitigation of systematic effects. Among these, the intrinsic alignments (IA) of galaxies are one of the dominant contaminants of the weak lensing (WL) and galaxy-galaxy lensing (GGL) probes. In this work, we assess IA modelling choices for Euclid… ▽ More The Euclid galaxy survey will provide unprecedented constraints on cosmology, but achieving unbiased results will require an optimal characterisation and mitigation of systematic effects. Among these, the intrinsic alignments (IA) of galaxies are one of the dominant contaminants of the weak lensing (WL) and galaxy-galaxy lensing (GGL) probes. In this work, we assess IA modelling choices for Euclid DR1 3x2pt analyses by comparing the performance of the two most commonly used IA models, nonlinear alignment (NLA) and tidal alignment tidal torquing (TATT), along with several variations. Our analyses combine three perspectives: i) the constraining power on the IA and cosmological parameters for each IA model, ii) the bias that results when the IA analysis model differs from the model used to generate the synthetic data vector, and iii) the degeneracies between IA and photometric redshift (photo-z) nuisance parameters. Among the IA models analysed, the redshift-dependent TATT model (zTATT) provides the most flexible description of IA, with a similar constraining power compared to simpler IA models, making it a well-motivated choice for Euclid DR1 3x2pt analyses. △ Less

Submitted 7 March, 2026; v1 submitted 18 February, 2026; originally announced February 2026.

Comments: 23 pages, 16 figures, submitted to A&A. Updated acknowledgements and affiliations

PeroMAS: A Multi-agent System of Perovskite Material Discovery

Authors: Yishu Wang, Wei Liu, Yifan Li, Shengxiang Xu, Xujie Yuan, Ran Li, Yuyu Luo, Jia Zhu, Shimin Di, Min-Ling Zhang, Guixiang Li

Abstract: As a pioneer of the third-generation photovoltaic revolution, Perovskite Solar Cells (PSCs) are renowned for their superior optoelectronic performance and cost potential. The development process of PSCs is precise and complex, involving a series of closed-loop workflows such as literature retrieval, data integration, experimental design, and synthesis. However, existing AI perovskite approaches fo… ▽ More As a pioneer of the third-generation photovoltaic revolution, Perovskite Solar Cells (PSCs) are renowned for their superior optoelectronic performance and cost potential. The development process of PSCs is precise and complex, involving a series of closed-loop workflows such as literature retrieval, data integration, experimental design, and synthesis. However, existing AI perovskite approaches focus predominantly on discrete models, including material design, process optimization,and property prediction. These models fail to propagate physical constraints across the workflow, hindering end-to-end optimization. In this paper, we propose a multi-agent system for perovskite material discovery, named PeroMAS. We first encapsulated a series of perovskite-specific tools into Model Context Protocols (MCPs). By planning and invoking these tools, PeroMAS can design perovskite materials under multi-objective constraints, covering the entire process from literature retrieval and data extraction to property prediction and mechanism analysis. Furthermore, we construct an evaluation benchmark by perovskite human experts to assess this multi-agent system. Results demonstrate that, compared to single Large Language Model (LLM) or traditional search strategies, our system significantly enhances discovery efficiency. It successfully identified candidate materials satisfying multi-objective constraints. Notably, we verify PeroMAS's effectiveness in the physical world through real synthesis experiments. △ Less

Submitted 10 February, 2026; originally announced February 2026.

Learning to Compose for Cross-domain Agentic Workflow Generation

Authors: Jialiang Wang, Shengxiang Xu, Hanmo Liu, Jiachuan Wang, Yuyu Luo, Shimin Di, Min-Ling Zhang, Lei Chen

Abstract: Automatically generating agentic workflows -- executable operator graphs or codes that orchestrate reasoning, verification, and repair -- has become a practical way to solve complex tasks beyond what single-pass LLM generation can reliably handle. Yet what constitutes a good workflow depends heavily on the task distribution and the available operators. Under domain shift, current systems typically… ▽ More Automatically generating agentic workflows -- executable operator graphs or codes that orchestrate reasoning, verification, and repair -- has become a practical way to solve complex tasks beyond what single-pass LLM generation can reliably handle. Yet what constitutes a good workflow depends heavily on the task distribution and the available operators. Under domain shift, current systems typically rely on iterative workflow refinement to discover a feasible workflow from a large workflow space, incurring high iteration costs and yielding unstable, domain-specific behavior. In response, we internalize a decompose-recompose-decide mechanism into an open-source LLM for cross-domain workflow generation. To decompose, we learn a compact set of reusable workflow capabilities across diverse domains. To recompose, we map each input task to a sparse composition over these bases to generate a task-specific workflow in a single pass. To decide, we attribute the success or failure of workflow generation to counterfactual contributions from learned capabilities, thereby capturing which capabilities actually drive success by their marginal effects. Across stringent multi-domain, cross-domain, and unseen-domain evaluations, our 1-pass generator surpasses SOTA refinement baselines that consume 20 iterations, while substantially reducing generation latency and cost. △ Less

Submitted 11 February, 2026; originally announced February 2026.

HII-DPO: Eliminate Hallucination via Accurate Hallucination-Inducing Counterfactual Images

Authors: Yilin Yang, Zhenghui Guo, Yuke Wang, Omprakash Gnawali, Sheng Di, Chengming Zhang

Abstract: Large Vision-Language Models (VLMs) have achieved remarkable success across diverse multimodal tasks but remain vulnerable to hallucinations rooted in inherent language bias. Despite recent progress, existing hallucination mitigation methods often overlook the underlying hallucination patterns driven by language bias. In this work, we design a novel pipeline to accurately synthesize Hallucination-… ▽ More Large Vision-Language Models (VLMs) have achieved remarkable success across diverse multimodal tasks but remain vulnerable to hallucinations rooted in inherent language bias. Despite recent progress, existing hallucination mitigation methods often overlook the underlying hallucination patterns driven by language bias. In this work, we design a novel pipeline to accurately synthesize Hallucination-Inducing Images (HIIs). Using synthesized HIIs, we reveal a consistent scene-conditioned hallucination pattern: models tend to mention objects that are highly typical of the scene even when visual evidence is removed. To quantify the susceptibility of VLMs to this hallucination pattern, we establish the Masked-Object-Hallucination (MOH) benchmark to rigorously evaluate existing state-of-the-art alignment frameworks. Finally, we leverage HIIs to construct high-quality preference datasets for fine-grained alignment. Experimental results demonstrate that our approach effectively mitigates hallucinations while preserving general model capabilities. Specifically, our method achieves up to a 38% improvement over the current state-of-the-art on standard hallucination benchmarks. △ Less

Submitted 10 February, 2026; originally announced February 2026.

VersaViT: Enhancing MLLM Vision Backbones via Task-Guided Optimization

Authors: Yikun Liu, Yuan Liu, Shangzhe Di, Haicheng Wang, Zhongyin Zhao, Le Tian, Xiao Zhou, Jie Zhou, Jiangchao Yao, Yanfeng Wang, Weidi Xie

Abstract: Multimodal Large Language Models (MLLMs) have recently achieved remarkable success in visual-language understanding, demonstrating superior high-level semantic alignment within their vision encoders. An important question thus arises: Can these encoders serve as versatile vision backbones, capable of reliably performing classic vision-centric tasks as well? To address the question, we make the fol… ▽ More Multimodal Large Language Models (MLLMs) have recently achieved remarkable success in visual-language understanding, demonstrating superior high-level semantic alignment within their vision encoders. An important question thus arises: Can these encoders serve as versatile vision backbones, capable of reliably performing classic vision-centric tasks as well? To address the question, we make the following contributions: (i) we identify that the vision encoders within MLLMs exhibit deficiencies in their dense feature representations, as evidenced by their suboptimal performance on dense prediction tasks (e.g., semantic segmentation, depth estimation); (ii) we propose VersaViT, a well-rounded vision transformer that instantiates a novel multi-task framework for collaborative post-training. This framework facilitates the optimization of the vision backbone via lightweight task heads with multi-granularity supervision; (iii) extensive experiments across various downstream tasks demonstrate the effectiveness of our method, yielding a versatile vision backbone suited for both language-mediated reasoning and pixel-level understanding. △ Less

Submitted 10 February, 2026; originally announced February 2026.

Sci-VLA: Agentic VLA Inference Plugin for Long-Horizon Tasks in Scientific Experiments

Authors: Yiwen Pang, Bo Zhou, Changjin Li, Xuanhao Wang, Shengxiang Xu, Deng-Bao Wang, Min-Ling Zhang, Shimin Di

Abstract: Robotic laboratories play a critical role in autonomous scientific discovery by enabling scalable, continuous experimental execution. Recent vision-language-action (VLA) models offer a promising foundation for robotic laboratories. However, scientific experiments typically involve long-horizon tasks composed of multiple atomic tasks, posing a fundamental challenge to existing VLA models. While VLA… ▽ More Robotic laboratories play a critical role in autonomous scientific discovery by enabling scalable, continuous experimental execution. Recent vision-language-action (VLA) models offer a promising foundation for robotic laboratories. However, scientific experiments typically involve long-horizon tasks composed of multiple atomic tasks, posing a fundamental challenge to existing VLA models. While VLA models fine-tuned for scientific tasks can reliably execute atomic experimental actions seen during training, they often fail to perform composite tasks formed by reordering and composing these known atomic actions. This limitation arises from a distributional mismatch between training-time atomic tasks and inference-time composite tasks, which prevents VLA models from executing necessary transitional operations between atomic tasks. To address this challenge, we propose an Agentic VLA Inference Plugin for Long-Horizon Tasks in Scientific Experiments. It introduces an LLM-based agentic inference mechanism that intervenes when executing sequential manipulation tasks. By performing explicit transition inference and generating transitional robotic action code, the proposed plugin guides VLA models through missing transitional steps, enabling reliable execution of composite scientific workflows without any additional training. This inference-only intervention makes our method computationally efficient, data-efficient, and well-suited for open-ended and long-horizon robotic laboratory tasks. We build 3D assets of scientific instruments and common scientific operating scenes within an existing simulation environment. In these scenes, we have verified that our method increases the average success rate per atomic task by 42\% during inference. Furthermore, we show that our method can be easily transferred from the simulation to real scientific laboratories. △ Less

Submitted 10 February, 2026; originally announced February 2026.