Peer Revenue Equity in science

Papers

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

All of arXiv

Friday 28 August

Materials Science

Nanolamellar Hybrid High-Entropy Alloys with Superior Micromechanical Properties

Metallic materials with nanolamellar structures, such as pearlitic steels, exhibit high strength with appropriate ductility. Considering the potential ability of such nanolamellar structures to break the traditional strength-ductility trade-off in metallic alloys, this study aims at developing a unique nanolamellar structure with superior micromechanical properties by combining two different high-entropy alloys (HEAs). Al0.1CoCrFeNi with the face-centered cubic (FCC) structure is combined with TiZrHfNbTa with the body-centered cubic (BCC) structure using high-pressure torsion (HPT) of half discs of each alloy. That way, a layered hybrid structure was formed, with layer thickness down to about 61 nm. The BCC/FCC nanolamellar hybrid structure exhibits an exceptional combination of properties with an ultimate tensile strength of 2.4 GPa, a maximum bending strength of 4.0 GPa, and a hardness of 740 Hv, while retaining some ductility/plasticity under both tensile and bending loads. Detailed analyses by synchrotron diffraction, electron microscopy and atom probe tomography suggests that these high strength and hardness, which are superior to those of nanostructured HEAs, result from: (i) extreme grain boundary strengthening from nanograins with a mean size of 22 nm, (ii) presence of defects such as dislocations in FCC and BCC, stacking faults in FCC and twins in FCC, and (iii) interphase hardening from BCC/FCC nanolamellar boundaries with about 30% contribution to the total hardness. This work demonstrates that combining two HEAs using HPT into a defect-rich hybrid nanolamellar composite forms a promising synergy of ultrahigh strength and reasonable ductility/plasticity.
Read the full abstract

All authors Shivam Dangwal, Yoji Mine, Shohei Ueki, Xavier Sauvage, Fabien Cuvilly, Liliana Romero Resendiz, Muhammad Naeem, Kaveh Edalati

Machine Learning

Residual-Guided Randomized Neural Networks

Randomized neural networks enable fast and analytically tractable training by fixing the input to hidden layer parameters at random and learning the output weights in closed form; however, their performance critically depends on a single uninformed draw of hidden units. This one shot and task uninformed feature construction often leads to redundant representations and suboptimal utilization of model capacity. To address this limitation, we propose a simple and broadly applicable residual guided procedure that greedily constructs the hidden layer using a closed form residual decrease criterion. At each stage, we (i) generate a pool of random candidate units, (ii) score each candidate by the exact reduction it induces in the ridge regularized objective, (iii) select the top k units, and (iv) refit the readout in closed form using the standard design with direct input links. This procedure yields a progressive training process with a guaranteed monotonic decrease of the training objective. The method is model agnostic: only the candidate generation is architecture specific, while the scoring selection refitting loop is shared across models. Extensive experiments on 71 benchmark datasets from the UCI repository, covering both binary and multiclass classification tasks, demonstrate that the proposed residual-guided models consistently outperform their baseline counterparts in terms of accuracy, stability, and overall ranking performance.
Read the full abstract

Robotics

CoCoBench: A Cooperative Coordination Benchmark for Embodied Multi-Agent Task Planning

Agent systems powered by multimodal large language models (MLLMs) have advanced rapidly in recent years, yet existing embodied-agent benchmarks still lack fine-grained diagnostics for multi-agent coordination. Most benchmarks either focus on single-agent task completion or summarize multi-agent behavior with overall task success rates, which can obscure coordination failures such as duplicated work, violations of ordering constraints, resource contention, and desynchronized handoffs. In this paper, we introduce CoCoBench, a construct-level benchmark for evaluating multi-agent embodied coordination in executable household tasks. CoCoBench contains 897 oracle-validated instances organized around four recurring coordination constructs: task allocation, sequential ordering, mutual exclusion, and handoff coordination. In addition to task success rate, CoCoBench provides construct-level scores that measure whether agents coordinate effectively. We evaluate 11 leading MLLMs across different coordination modes, observation inputs, and numbers of agents. The results show that coordination ability is highly construct-specific: strong overall performance does not imply balanced competence across different coordination types. These findings point to new directions for designing targeted model architectures and improving multi-agent coordination ability.
Read the full abstract

All authors Yang Chen, Ye-Xin Xie, Lirong Che, Danyang Peng, Yuzhe Yang, Peiwen Lin, Xu Cao, Chuang Wang, Lei Yuan, Jian Su, Lan-Zhe Guo

General Relativity and Quantum Cosmology

Investigation of global structure of Kerr-Newman-Bertotti-Robinson spacetime

We conduct an investigation of the global structure of Kerr-Newman-BR spacetime and its various subcases such as Schwarzschild-BR spacetime. In this work, we extend the result of our previous studies, where we found that $r\to \infty$ is not the proper conformal infinity. By using the reciprocal coordinate $y=1/r$, we show that the whole family of Kerr-Newman-BR spacetimes can be continuously extended through the surface $r\to \infty$. However, the conformal structure after passing $r\to \infty$ strongly depends on the type of spacetime one is considering (either rotation is present, or not) and on the polar direction one chooses to be fixed (namely, whether we consider generic polar angle $\theta$, or we focus on the behaviour near the equatorial plane $\theta=\pi/2$, or near the poles $\theta=0,\pi$). Physically this extension can be interpreted as kind of “confinement box,” created solely by the magnetic field. Finally, we discuss the universality of such a continuous extension.
Read the full abstract

Artificial Intelligence

Finding Where the Buck Stops: An Automated Failure Attribution-Based Reflection Framework for Multi-Agent Collaboration

Multi-agent systems (MAS) powered by large language models have shown promise for complex tasks but suffer from high failure rates. Current self-reflection methods for MAS require all agents to reflect upon failure, overlooking a critical reality: failures typically stem from a specific agent leading the task astray, namely the decisive error agent, while others merely fulfill their regular duties. Forcing regular-behaving agents to reflect contaminates their memory with wrong insights. Hence, we propose DoCtOR (Diagnose-then-Correct PPO-enhanced Reflection), a novel reflection framework that enhances multi-agent collaboration. DoCtOR first identifies the decisive error step and decisive error agent through automated failure attribution, then employs counterfactual reasoning to generate a corrected decisive error step, and finally engages only the decisive error agent to produce targeted reflections. Experimental results show DoCtOR achieves 22%, 26%, and 27% improvements over initial success rates on HotPotQA, ChartQAPro, and Mind2Web datasets, outperforming Reflexion, Retroformer, and COPPER. We further establish the generalizability of our diagnose-then-correct paradigm and demonstrate that in low-resource settings, focusing reflection on reasoning steps after the decisive error step achieves comparable quality to reflecting on the complete failure trajectory.
Read the full abstract

All authors Xiaoqing Wang, Keman Huang, Bin Liang, Hongyu Li, Xiaoyong Du, Wuqiong Pan

Mesoscale and Nanoscale Physics

Rapid Charge Stability Diagram Generation from Device-level Modeling of Semiconductor Quantum Dots

Self-consistent Schrödinger-Poisson calculations are a powerful tool for predicting the behavior of layered semiconductor quantum dot devices. However, characterization of charge stability diagrams through fully simulated gate-voltage sweeps is computationally expensive. Combining a Multi-Domain Multi-Model (MDMM) approach with an automated tuning routine, we identify gate voltages associated with selected charge configurations. This small set of self-consistent simulations can be augmented with Full Configuration Interaction (FCI) energy calculations to extract charging energies, lever arms, and interdot Coulomb interactions to directly parameterize a Hubbard model for rapid charge stability diagram generation. For an Intel Tunnel Falls Si/SiGe device, we demonstrate the Hubbard model's ability to reproduce charge stability diagrams at a fraction of the computational cost in comparison to voltage bias sweeps. We further compare the simulated diagrams to experimental data and demonstrate qualitative agreement. Our result represents a step towards predictive digital twin models for semiconductor quantum dot devices. Finally, we apply this workflow towards lever arm engineering in a second device, demonstrating that the method extends to multiple architectures.
Read the full abstract

All authors Ron Nodel, David W. Kanaar, Connor Nasseraddin, Tim J. Wilson, Hong-Wen Jiang, Jason R. Petta, Chris Anderson, Mark F. Gyure

Machine Learning

SinkSLOT: Sinkhorn via Sparse Lifted Optimal Transport

Entropic optimal transport (EOT) has been shown to offer a computationally tractable approximation to exact optimal transport. However, the standard Sinkhorn-Knopp algorithm has two main limitations. First, given discrete measures with $N$ points, each iteration requires $O(N^2)$ operations, which restricts its use on large-scale datasets (e.g. $N\geq10^4$). Second, it uses the independent coupling as a reference measure for regularisation. This assigns mass to high-cost transport edges at moderate regularisation strengths. We propose SinkSLOT, which addresses both limitations by putting forth the expected sliced lifted transport plan as a natural way to sparsify the Gibbs kernel with a non-independent prior coupling. We prove that: 1) SinkSLOT converges; 2) with $L$ slices, each resulting sparse Sinkhorn iteration costs $O(LN)$; and 3) the resulting objective is a divergence requiring no debiasing. Experiments on synthetic benchmarks show that SinkSLOT delivers substantial speedups over state-of-the-art dense and sparse EOT methods. We also demonstrate the applicability of the proposed divergence in a gradient flow experiment. The code is publicly available at https://github.com/cai4cai/SinkSLOT.
Read the full abstract

All authors Ian Hsieh, Soumya Snigdha Kundu, Tom Vercauteren, Reuben Dorent

Optics

Quantum many-body effects in the optical response of ideal thin films

We study quantum many-body effects in the long-wavelength optical response of confined electrons at finite temperatures. We simulate homogeneous electron gas confined in one dimension into a slab of nanoscale thickness. We demonstrate how the slab boundaries break down the ideal Drude response of free charge carriers, giving rise to scattering effects due to both the surfaces and quantum many-body interactions. We use a recent path-integral Monte Carlo (PIMC) approach developed in [Tiihonen et al. Phys. Rev. A 113, 053711] to quantify these effects in high accuracy. We perform phenomenological fits to Drude and Drude-Lorentz models parameters, manifesting various trends of the optical response with physical parameters like density and temperature, and numerical effects like finite size and the quantum statistics.
Read the full abstract

All authors David Trejo-Garcia, Tapio T. Rantala, Marco Ornigotti, Juha Tiihonen

Cosmology and Nongalactic Astrophysics

Guide for the Hubble Tension: Mid-Time Solutions and Generalized PAge Parameterizations

In cosmology, the Hubble tension has become one of the most serious challenges in the last decade. Assuming that all the observational data are right, one of the ways out is to modify the standard $\Lambda$CDM cosmology. In the literature, the theoretical modifications are mainly made in the early and the late/local universes. But on both sides, some no-go arguments have been proposed recently. In particular, the general no-go arguments for the late-time modifications are mainly based on the PAge parameterization. In the present work, we note that there are some flaws in the (original) PAge parameterization, and then propose various generalized PAge (GPAge) parameterizations, which can be extended to $z>z_{\rm CMB}$ and hence the cosmic microwave background (CMB) can be taken into account, while they are also more accurate at low redshifts. With these GPAge parameterizations, we revisit the Hubble tension by using not only the late-time observations but also the observational data of CMB. We find that the Hubble tension could be alleviated (or even resolved) in GPAges significantly different from the standard $\Lambda$CDM cosmology, while they are overwhelmingly preferred over $\Lambda$CDM by the observational data in terms of all the information criteria. Surprisingly, we find that the new physics in the mid-time might be the key to resolve the Hubble tension.
Read the full abstract

Nuclear Experiment

Lifetime measurements in neutron-rich odd-A yttrium isotopes ($^{93-99}$Y): Investigation of shape coexistence and the intertwined quantum phase transition

Lifetimes of 16 excited states in the neutron-rich odd-$A$ nuclei $^{93-99}$Y were measured using fast-timing $\gamma$-$\gamma$ coincidence spectroscopy with fast scintillation detectors at the LOHENGRIN recoil separator. Particular attention is given to the region around $N \approx 59$, where rapid changes in nuclear deformation and shape coexistence occur. The lifetimes, determined using the generalized centroid difference method, are compared with interacting boson-fermion model calculations with configuration mixing, in which the odd-$A$ yttrium isotopes are described as a proton coupled to a bosonic core containing normal and intruder configurations. The results provide new constraints on theoretical descriptions of shape coexistence and structural evolution in neutron-rich nuclei near $A \approx 100$, particularly for odd-$A$ systems where experimental information remains limited.
Read the full abstract

All authors A. Pfeil, N. Gavrielov, U. Köster, Y. H. Kim, N. Cieplicka-Oryńczak, J. Dudouet, A. Esmaylzadeh, Ł. W. Iskra, M. Ley, J.-M. Régis, D. Reygadas, J. Jolie

Solar and Stellar Astrophysics

Search for magnetic fields in seven slowly rotating A stars

A small fraction of A-type stars host strong fossil magnetic fields. Identifying them relies on indirect indicators such as chemical peculiarities, slow rotation, and rotational modulation, which are also shared by non-magnetic stars. Assessing their reliability is essential for future surveys. We magnetically characterise seven slowly rotating, chemically peculiar A-type stars selected through two complementary strategies. Three were selected from abundance analysis combined with potential intrinsically slow rotation, and four from a spectral depression at 5200 Angstrom combined with rotational modulation in TESS photometry. We obtained high-resolution spectropolarimetry, applied least squares deconvolution, and measured longitudinal field strengths for detections. For non-detections, we derived upper limits on polar field strength and evaluated the critical fields of Zahn and Spruit. We detect magnetic fields in all four stars selected via the spectral depression and photometric criterion, with longitudinal field strengths of approximately 145-2900 G, while none of the three stars selected via abundance and slow rotation shows a Zeeman signature. Combined upper limits for two non-detections lie below or near the magnetic desert boundary, but only one is also below the critical fields required for rigid rotation. The non-detections are most naturally interpreted as Am stars. Among the detections, HD 63843 stands out as a magnetic $\delta$ Scuti pulsator, promising for magneto-asteroseismology. In conclusion, slow rotation combined with chemical peculiarity alone does not reliably trace large scale magnetic fields, whereas adding the 5200 Angstrom depression and stable rotational modulation in space photometry appears to select magnetic A-type stars with high efficiency, providing a practical guideline for future spectropolarimetric surveys.
Read the full abstract

All authors Pablo de Frutos Rull, Coralie Neiner, Jonathan Labadie-Bartz, Frédéric Royer, Richard Monier, Keegan Thomson-Paressant

Rings and Algebras

On Locally Finite Derivations in Ore Extensions

Let $\Bbbk$ be an algebraically closed field of characteristic zero. We classify the locally finite derivations of arbitrary Ore extensions of $\Bbbk[x]$, thus extending van den Essen's \cite{V92} classification for the polynomial algebra $\Bbbk[x,y]$ to this noncommutative setting. More precisely, we consider the three families arising in the classification of Ore extensions of $\Bbbk[x]$: the quantum plane, the first quantum Weyl algebra, and the differential Ore extensions \[ A_h=\Bbbk[x][t;h(x)\partial_x]. \] For both the quantum plane and the first quantum Weyl algebra, we determine the locally finite derivations and explain how the resulting classifications are related to the work of Suárez-Alvarez and Vivas \cite{SuarezVivas} on generalized Weyl algebras. For the algebras $A_h$, with $h$ nonconstant, we obtain a complete classification in both the square-free and non-square-free cases. As a consequence, we show that $\LFD(A_h)$ is a solvable and weakly locally finite Lie subalgebra of $\Der(A_h)$, although it is not locally finite as a set of derivations.
Read the full abstract

All authors R. Baltazar, A. Bianchi, M. Veloso, J. Schwarz

Artificial Intelligence

Physics-Guided Flow Matching for CT Image Reconstruction

Deep generative models have recently emerged as powerful priors for solving ill-posed inverse problems in CT, with diffusion-based approaches achieving state-of-the-art reconstruction performance. However, diffusion models typically rely on stochastic sampling procedures, long inference trajectories, and carefully tuned noise schedules, which can limit computational efficiency and numerical stability, especially at high spatial resolutions. In this work, we investigate Flow Matching as an alternative generative prior for CT reconstruction. We train a high-resolution Rectified Flow Matching model on 256x256 chest images from the Mayo Clinic Low-Dose CT dataset. To mitigate overfitting and limited anatomical variability, we employ a two-stage training strategy consisting of an initial phase with strong, anatomically informed data augmentation, followed by a fine-tuning phase with reduced or no augmentation to refine structural fidelity. The resulting model is capable of generating high-quality and anatomically coherent CT-like images, serving as a strong learned prior. We then evaluate multiple reconstruction methods specifically designed for Flow Matching models, including Plug-and-Play Flow, FlowDPS, Flower, and Flow-Priors (ICTM), and compare them against state-of-the-art diffusion-based reconstruction algorithms such as DDRM, DPS, and DiffPIR. Experimental results across several CT inverse problem settings show that Flow Matching-based approaches consistently outperform diffusion-based methods in terms of PSNR, SSIM, and perceptual quality, while requiring fewer sampling steps. Finally, we publicly release the trained Flow Matching model and accompanying code to facilitate reproducibility and future research. Overall, this work demonstrates that Flow Matching provides a stable, efficient, and effective alternative to diffusion models for high-resolution CT image reconstruction.
Read the full abstract

Number Theory

Rational and integral values of rational functions at rational points

The basic issue concerns sets of values of rational functions at rational points of an algebraic variety, namely image f(X(k)), where X is an algebbaric variety and f is a rational function on X, defined over the number field k. For instance, we shall prove that for X an abelian variety, the map between rational points is never surjective. This is reminiscent of the Hilbert Property, but here the fibers may have arbitrary dimension. One of our examples concerns the classical Hilbert Property: we produce a simply connected affine surface whose set of integral points is Zariski-dense and thin, disproving a plausible expectation. We shall also discuss hieghts and integrality issues; in this context, a role will be played by 'gcd estimates'. In the first Appendix, written by D. Masser, an effective estimate of some relevant gcd is provided.
Read the full abstract

Combinatorics

Simplicial arrangements in real projective three-space revisited

In this paper we study irreducible simplicial arrangements of projective planes in $\mathbb{P}^{3}(\mathbb{R})$ from combinatorial and projective geometry viewpoints. We first formulate a simpliciality criterion in terms of incidences between rank-two and rank-three flats, together with equivalent formulations using face numbers, reduced restrictions, and characteristic polynomials. We also relate the classical planar restriction data of Grünbaum-Shephard to Ziegler multirestrictions. Our principal result concerns the rank-four special-vertex property: among the irreducible crystallographic Coxeter arrangements of rank four, the arrangements of types $A_4$ and $B_4$ admit a special vertex, whereas those of types $D_4$ and $F_4$ do not. A simplicial deletion chain inside $B_4$ supplies further irreducible examples with a special vertex. Finally, we compare rank-flat, Purdy-type, and Grünbaum-Shephard defects for these arrangements.
Read the full abstract

High Energy Physics (Theory)

Revisiting the $U(1)$-$CP$ tension in QCD from a 1PI-action perspective

We discuss the tension one encounters in trying to solve, simultaneously, the $U(1)$ and strong-$CP$ problems in QCD with light quarks and a small $\bar{\theta}$ angle within QCD itself. We do so by considering the low-energy expansion of a $1$PI effective action $\Gamma$, a functional of a properly chosen set of gauge-invariant (composite) operators. After enforcing Ward-identity constraints on $\Gamma$, we prove, under minor assumptions, that $CP$ violation at small non-vanishing $\bar{\theta}$ can be avoided only if the anomaly-induced contribution to the pseudoscalar mass matrix is completely negligible when compared to its non-anomalous counterpart - that is, only if the $U(1)$ problem is not solved. The topological susceptibility in QCD and a precise generalization of the one in pure Yang-Mills theory emerge as the most important players in establishing the above tension and in determining the corrections to various known results at leading order in $\bar{\theta}$.
Read the full abstract

All authors Paolo Di Vecchia, Francesco Sannino, Graham M. Shore, Gabriele Veneziano, Shimon Yankielowicz

Artificial Intelligence

Regime-Aware Portfolio Management via Retrieval-Augmented LLM-Guided Expert Switching

Financial markets are inherently non-stationary, making the effectiveness of individual portfolio-management strategies highly dependent on changing market conditions. This work proposes a retrieval-augmented expert-switching framework that dynamically selects portfolio management experts based on their historical performance under similar market situations. A dual-stream variational autoencoder represents asset-level and market-wide information, while a retrieval-based knowledge base stores historical situations and expert performance. During inference, an instruction-tuned LLM reasons over the retrieved evidence to identify the most appropriate expert rather than directly generating portfolio actions. We further establish a monotonicity property showing that adding a locally superior expert cannot degrade the switching mechanism's performance. Experiments across cryptocurrency, stock, and foreign-exchange markets show that the proposed selector achieves the highest cumulative return and Sharpe ratio among the evaluated selection strategies in all three markets. In the stock market, for example, cumulative return increases from 26% for the best fixed expert to 34%, while the Sharpe ratio improves from 0.74 to 0.96. Ablation results confirm the importance of both retrieval and LLM reasoning, while experiments with different expert-pool sizes demonstrate the value of complementary expertise. Overall, the findings support retrieval-grounded expert switching as an effective approach to adaptive portfolio management in non-stationary financial environments.
Read the full abstract

Quantum Physics

Role of the Drive in Mediating Correlations Between Two Qubits Through a Shared Dissipative Cavity

Using a numerically exact master equation, we demonstrate that two qubits, coupled solely through a shared damped, driven cavity, can become correlated. The drive influences both the amount and the type of correlation. For parametric, coherent, and resonantly modulated drives, the qubits develop quantum discord that increases with cavity temperature, while the logarithmic negativity remains numerically zero. This indicates the presence of discord without entanglement. In contrast, a time-modulated parametric drive is the only one that generates genuine two-qubit entanglement, achieving \(E_\mathcal{N} \simeq 0.15\) and concurrence \(\simeq 0.16\) at \((\eps, \gamma) = (0.3, 0.2)\), which rises to \(E_\mathcal{N} \simeq 0.32\) in the weak-coupling, moderate-damping region. Heating eventually destroys this entanglement around \(n_{\mathrm{th}} \simeq 0.2\), while discord continues to grow, resulting in a temperature-driven transition from entanglement to discord within a single drive. Moreover, the parametric drive offers the best protection for single-qubit coherence, unlike the coherent and modulated drives. An adiabatic-elimination model indicates that the cavity generates an effective coupling and a collective dephasing channel, both of which increase with temperature, explaining the observed discord without entanglement.
Read the full abstract

Atomic Physics

Coherently Enhanced Cherenkov Radiation by Highly Relativistic and Ultra-Compact Electron Beams

Cherenkov radiation is emitted when charged particles penetrate dielectric media with velocity higher than the speed of light in the medium. Recent developments in plasma-based accelerator research enable creation of highly relativistic electron beams with femtosecond length and tens of micrometer scale radii, at energies of up to several GeV. Here, we consider Cherenkov radiation emitted by such extreme beams penetrating a gas target with refractive index very close to one. We demonstrate coherent enhancement in the emitted Cherenkov radiation up to the optical or even near-ultraviolet regime, which is due to the high beam energy and density. We also explore the dependence of the coherently enhanced Cherenkov radiation spectrum on the beam shape.
Read the full abstract

Computer Vision

Synth-JDoc: Synthesizing a Japanese Document Image Dataset for OCR with Diverse Layouts and Embedded Images

The ability of Large Vision Language Models (LVLMs) to read text within document images is crucial, as it enables various applications such as Document Visual Question Answering. To enhance the text-reading capabilities of LVLMs, high-quality OCR datasets are essential. This need is particularly critical for Japanese documents, which often feature vertically written text alongside horizontally written text. Current LVLMs demonstrate considerably lower performance on vertically written Japanese text than on horizontally written text, necessitating specialized OCR datasets to bridge this gap. However, manually constructing OCR datasets is expensive and difficult to scale. Alternatively, constructing datasets by extracting text from existing document images using OCR models introduces challenges, such as text recognition errors and the prerequisite of sourcing document images. To address these issues, we construct an OCR dataset by synthesizing document images directly from text. Leveraging HTML and CSS, we generate multi-column documents that incorporate both vertical and horizontal writing styles. Furthermore, to ensure the visual realism of the documents, we embed images generated by text-to-image models within the layout. Additionally, to foster model robustness, we apply noise and degradation filters to the synthesized document images. In our experiments, we compared the performance of models fine-tuned on our synthetic dataset against baselines fine-tuned on synthetic datasets from prior work and those generated by a high-performance text-to-image model. Evaluation results demonstrate that our synthetic dataset is the most effective approach for improving LVLM performance on reading vertically written Japanese text. Our dataset and code are publicly available (https://github.com/llm-jp/synth-jdoc).
Read the full abstract

Computer Vision

A comprehensive and trustworthy benchmark of AI methods for change detection in Earth observation

Change detection in Earth observation (EO) is critical for monitoring land surface transformations, yet recent research in the field is constrained by inconsistent evaluation protocols and a narrow focus on predictive accuracy without regard for computational efficiency. To address this, we present a standardized, open-source benchmark for evaluating state-of-the-art (SOTA) deep learning methods for Earth observation change detection. We conduct a comprehensive analysis of ten representative model architectures, ranging from convolutional networks (CNNs) to vision transformers (ViTs), across ten heterogeneous change detection datasets. We rigorously evaluate these models with identical experimental protocols, comparing models trained from scratch against those utilizing pre-trained weights. Furthermore, we evaluate predictive performance alongside computational efficiency, including parameter counts and inference latency. Our findings reveal that well-optimized classical architectures, such as Siamese U-Nets, frequently outperform more complex contemporary models when computational efficiency is factored in, and that pre-training consistently provides a significant performance boost with no additional inference cost. To ensure complete transparency and reproducibility, all experimental resources, including standardized data splits, training scripts, training logs, and model checkpoints are publicly available and adhere to FAIR principles (Findable, Accessible, Interoperable, and Reusable).
Read the full abstract

All authors Tadej Tomanič, Alice Baudhuin, Jan Sotošek, Jure Brence, Panče Panov, Nikola Simidjievski, Dragi Kocev

Robotics

Training-free Suction Grasp Detection for Deformed Aseptic Cartons Using Vision-Language Models and Geometric Surface Scoring

Robotic sorting of recyclable waste is challenging due to the deformable and geometrically inconsistent nature of target objects. We present a training-free suction grasping system for sorting deformed aseptic beverage cartons, decoupling target identification from grasp-point selection. An open-vocabulary vision-language model detects cartons from a text prompt, SAM2 refines each detection into an instance mask, and a geometric scoring method selects the suction point by combining surface flatness with normal alignment. Three geometric methods are compared: k-nearest-neighbour PCA, Sobel cross-product, and RANSAC plane fitting. Evaluated on a real robot across three deformation levels and 35 cluttered scenes, single-object grasp success reaches 88.2% and end-to-end retrieval in clutter is 72.6%.
Read the full abstract

Machine Learning (Statistics)

I-FLOP: Fast Learning of Order and Parents from Interventional Data

We extend the FLOP (fast learning of order and parents) algorithm recently proposed by Wienöbst et al. (2026) from observational to interventional data. In particular, we use the interventional BIC score of Hauser and Bühlmann (2012), adapting it to be used with the iterative Cholesky-based score updates that are partly responsible for FLOP's speed. We show that, in the sample limit, I-FLOP recovers a DAG in the same interventional Markov equivalence class as the data-generating DAG. We compare I-FLOP to existing causal structure learning algorithms on real and simulated interventional data, where it performs favorably in terms of both performance and run time.
Read the full abstract

High Energy Physics (Phenomenology)

HEPLocalAgent 1.0: Running Collider Simulations from Plain-Language Requests on Your Own Computer

We present HEPLocalAgent, an open source local interface that builds a bounded class of collider simulation workflows from natural language requests. A locally served language model proposes a typed workflow representation, and deterministic software then restores recognized user stated quantities, builds the HEP tool inputs, validates the supported workflow, and presents the artifacts for approval before execution. In a same response comparison on 47 evaluable model request cases, the first structured proposal gave 7 unmodified artifacts satisfying the external benchmark scorer, against 19 after the full deterministic pipeline. Under the fixed representation normalization defined by the benchmark, the counts were 11 and 43. The direction of improvement is unchanged. The gap between the two views arises because the released builder and the benchmark scorers disagree on three bookkeeping conventions, namely launch form, two fixed control lines, and the output directory name, not on physics content. Four of seven approved workflows ran to completion on the managed local software stack, with cross sections consistent between repeats. In a separate challenge set, 57 of 96 problematic requests still reached the approval stage after part of the request was dropped, defaulted, or reinterpreted. No tested unsafe payload was retained in an executable artifact before the approval gate, but this does not establish operating system level containment. The deterministic backend supports MadGraph, Pythia8, Delphes, and a restricted MadAnalysis 5 plan. Reliable natural language routing to the MadAnalysis stage was not demonstrated in the tested examples. Version 1 should therefore be seen as an inspectable, validation gated workflow constructor requiring expert approval rather than an autonomous or scientifically self validating agent.
Read the full abstract

Commutative Algebra

On the existence of the maximal ideal in the set of associated primes of monomial ideals

In this paper, we investigate the behavior of associated primes of powers of monomial ideals, with particular emphasis on the presence of the maximal ideal and the phenomenon of fluctuation. We establish criteria for determining when the maximal ideal belongs to the set of associated primes of powers of monomial ideals in $K[x,y,z]$, and provide examples illustrating its appearance and disappearance among the associated primes of successive powers. Furthermore, we prove that for every $n\geq 3$, there exist infinitely many monomial ideals in $K[x_1,\ldots,x_n]$ whose powers exhibit fluctuations in their sets of associated primes. Finally, we construct infinitely many examples of nearly normally torsion-free (respectively, co-nearly normally torsion-free) monomial ideals that fail to satisfy the persistence (respectively, copersistence) property.
Read the full abstract

Machine Learning

Spectral Features Dominate BCG Respiratory-Event Detection: A Large-Scale Patient-Independent Comparison of Feature Groups in Sleep Apnea Patients

Unobtrusive ballistocardiographic (BCG) sensing is a promising modality for long-term sleep-apnea monitoring, yet it remains unclear which signal features are most discriminative for respiratory-event detection. We present a literature-guided, patient-independent comparison of ten BCG feature groups using a 512-sensor capacitive pressure mat recorded simultaneously with respiratory polygraphy in 155 patients (52 female, 103 male) undergoing in-hospital evaluation for obstructive sleep apnea. Features were extracted from six spatially distinct signal channels, yielding a 191-dimensional feature vector spanning general statistical, time-domain, frequency-domain, wavelet, frame-energy, and nonlinear complexity descriptors. Under strict leave-one-patient-out cross-validation for binary classification of respiratory-event windows versus event-free reference windows, Random Forest and Histogram Gradient Boosting achieved AUC-ROC of 0.967 and 0.969 and AUC-PR of 0.977 and 0.979, respectively. Feature-importance analysis revealed that frequency-domain features dominate discrimination: breathing-band power in the 0.1-0.4 Hz range accounted for 30.3% of total discriminative information across all spatial channels, and Fast Fourier Transform spectral-shape descriptors of the adaptively preprocessed channel contributed a further 15.1%. AUC and curve-length features provided the main complementary time-domain evidence (21.5%), whereas wavelet-derived and nonlinear features contributed smaller secondary effects (10.4% combined across 59 features). Frequency-domain and time-domain features together accounted for 67% of total discriminative information, demonstrating that a compact, interpretable subset of the full feature library achieves clinically relevant performance under patient-independent validation and providing an empirical basis for feature selection in future BCG systems.
Read the full abstract

All authors Israel Campero Jurado, Zoe Bousraou, Lara Benning, Sara Padilla Neira, Alexander Breuss, Robert Riener, Esther Irene Schwarz, Elisabeth Wilhelm

Artificial Intelligence

Beyond Task-Only Matching: Personalized Skill Routing with Counterfactual Evaluation

The rapid expansion of reusable skill repositories makes skill routing a critical capability for large language model (LLM) agents. Existing methods treat routing as task-only semantic matching. However, when users with incompatible constraints issue an identical request, this assumption conflates task relevance with skill suitability: a task-only router can select a semantically plausible skill that is unsuitable for the requesting user. To expose this failure mode, we formulate personalized skill routing as profile-conditioned retrieval, in which relevance depends jointly on the task and the user profile. We first introduce a profile-counterfactual benchmark, in which the task is held fixed while changes in the user profile induce changes in the reference skill. We further construct paired counterfactual supervision and propose SkillFeed, a progressive retrieve-and-rerank framework that first establishes task–skill alignment and then learns profile-conditioned discrimination. By retrieving body-level evidence and reranking semantically similar but profile-conflicting candidates, SkillFeed identifies skills that satisfy both task requirements and user constraints. On SkillFeed-Bench, SkillFeed attains 75.1% top-1 retrieval accuracy, a 23.1-point improvement over the corresponding pretrained routing baseline. Adding profile conditioning yields a 35.1-point gain on queries where user profile changes the reference skill. This contrast shows that user profiles are most consequential precisely when they change skill suitability. Our website is publicly available at http://www.aiskillfeed.com .
Read the full abstract

All authors Tianle Wang, Yanghe Zou, Xiang Liu, Ziyao Huang, Chenchen Fu, Weiwei Wu

Computer Vision

WilLaGS: Latent-Conditional 3D Appearance Fields for Robust Gaussian Splatting In-the-Wild

3D Gaussian Splatting (3DGS) delivers real-time and high-fidelity rendering but remains challenged by unconstrained in-the-wild scenes, where drastic appearance variations and transient objects violate multi-view consistency. Existing methods are fundamentally limited by independent and discrete embeddings that struggle to capture continuous environmental changes or model spatially-varying local illumination. To address these limitations, we propose WilLaGS, a unified framework for robust 3D scene reconstruction and generative appearance synthesis under unconstrained settings. Specifically, we introduce a generative appearance model where a $\beta$-VAE learns a structured and continuous manifold of global appearance. Conditioned on the latent code, we construct a 3D neural appearance field that generates dynamic Tri-Plane features to encode spatially-varying local illumination effects. Furthermore, to suppress transient artifacts, we present a self-supervised perceptual masking mechanism that leverages a Teacher-Student (EMA) architecture to derive a stable scene consensus, robustly identifying inconsistent regions via perceptual discrepancies. Extensive experiments on multiple datasets demonstrate that WilLaGS achieves state-of-the-art performance in reconstruction quality and novel view appearance synthesis, while maintaining real-time rendering efficiency.
Read the full abstract

All authors Yuhao Bai, Qianqiu Tan, Lilong Chen, Huanhuan Lv, Lijun Chen

Materials Science

QUBO-Compatible Active Learning for Inverse Design of High-Entropy Alloys

Machine-learned forward models can rapidly predict alloy properties, but their use for inverse design remains challenging when the search should also retain compatibility with quadratic unconstrained binary optimization (QUBO). Here, we develop a QUBO-compatible active-learning framework for inverse design of high-entropy alloys using a pretrained graph-neural-network predictor as a fixed property oracle. A property-guided binary variational autoencoder provides a binary latent representation, while an ensemble of quadratic factorization machines guides candidate selection. We systematically benchmark the framework through controlled latent-space ablations and comparison with direct composition-space optimization. The results show that candidate generation is a major determinant of search performance: local perturbations around previously high-performing latent codes provide the largest workflow-specific improvement, while surrogate-based selection further prioritizes candidates within the enriched search pool. The resulting QUBO-compatible workflow remains competitive with strong classical optimization strategies, although a composition-space genetic algorithm achieves the highest mean score. Finally, the learned quadratic surrogate can be exported directly as a QUBO. These results show that effective data acquisition can be separated from the final QUBO optimization endpoint, providing a benchmarked route for QUBO-compatible data-driven materials inverse design.
Read the full abstract

All authors Giorgio Silvi, Kirsten Bark, Rolando Reiner, Nicolas Vogt, Thomas Plehn, Daniel Barragan-Yani, Marc Landmann, David Melching