Search

Recent progress in large language models (LLMs) has focused on producing responses that meet human expectations and align with shared values - a process coined alignment. However, aligning LLMs remains challenging due to the inherent disconnect between the complexity of human values and the narrow nature of the technological approaches designed to address them. Current alignment methods often lead to misspecified objectives, reflecting the broader issue of incomplete contracts, the impracticality of specifying a contract between a model developer, and the model that accounts for every scenario in LLM alignment. In this paper, we argue that improving LLM alignment requires incorporating insights from societal alignment frameworks, including social, economic, and contractual alignment, and discuss potential solutions drawn from these domains. Given the role of uncertainty within societal alignment frameworks, we then investigate how it manifests in LLM alignment. We end our discussion by offering an alternative view on LLM alignment, framing the underspecified nature of its objectives as an opportunity rather than perfect their specification. Beyond technical improvements in LLM alignment, we discuss the need for participatory alignment interface designs.

We present a flip-chip architecture for an array of coupled superconducting qubits, in which circuit components reside inside individual microwave enclosures. In contrast to other flip-chip approaches, the qubit chips in our architecture are electrically floating, which guarantees a simple, fully modular assembly of capacitively coupled circuit components such as qubit, control, and coupling structures, as well as reduced crosstalk between the components. We validate the concept with a chain of three nearest neighbor coupled generalized flux qubits in which the center qubit acts as a frequency-tunable coupler. Using this coupler, we demonstrate a transverse coupling on/off ratio $\approx$ 50, zz-crosstalk $\approx$ 0.7 kHz between resonant qubits and isolation between the qubit enclosures > 60 dB.

Domain wall (DW) dynamics in antiferromagnetic (AFM) systems offer the advantages over their ferromagnetic counterparts of having faster and more energy efficient manipulation due to the absence of net magnetization, leading to reduced magnetic crosstalk and improved performance in spintronic devices. A comprehensive analysis of DW dynamics across regimes such as creep, depinning, and flow is well established in ferromagnetic systems but remains lacking in AFM-coupled systems. In this study, we explore the nature of DW dynamics in synthetic ferrimagnetic multilayers composed of Pt|Co|Tb|Al for different Tb thickness, focusing on the underlying pinning parameters, and on the different regimes of DW dynamics driven by spin-orbit torques (SOTs). We find that due to the AFM coupling between Co and Tb, the magnetic moment of Tb increases with Tb thickness resulting in a reduced saturation magnetization and an enhanced depinning field. The DW disorder interaction is found to vary weakly with the AFM coupling between Co and Tb, while the complete withdrawal of the Tb layer strongly increases the anisotropy and the DW pinning. Furthermore, we propose a novel approach to measure effective SOTs by comparing depinning transitions in current and field-induced DW motion. This research reveals new insights into DW dynamics in coupled AFM systems, highlighting enhancements in mobility through optimized SOTs and pinning landscapes., Comment: 12 pages, 9 figures

The Athena mission entered a redefinition phase in July 2022, driven by the imperative to reduce the mission cost at completion for the European Space Agency below an acceptable target, while maintaining the flagship nature of its science return. This notably called for a complete redesign of the X-ray Integral Field Unit (X-IFU) cryogenic architecture towards a simpler active cooling chain. Passive cooling via successive radiative panels at spacecraft level is now used to provide a 50 K thermal environment to an X-IFU owned cryostat. 4.5 K cooling is achieved via a single remote active cryocooler unit, while a multi-stage Adiabatic Demagnetization Refrigerator ensures heat lift down to the 50 mK required by the detectors. Amidst these changes, the core concept of the readout chain remains robust, employing Transition Edge Sensor microcalorimeters and a SQUID-based Time-Division Multiplexing scheme. Noteworthy is the introduction of a slower pixel. This enables an increase in the multiplexing factor (from 34 to 48) without compromising the instrument energy resolution, hence keeping significant system margins to the new 4 eV resolution requirement. This allows reducing the number of channels by more than a factor two, and thus the resource demands on the system, while keeping a 4' field of view (compared to 5' before). In this article, we will give an overview of this new architecture, before detailing its anticipated performances. Finally, we will present the new X-IFU schedule, with its short term focus on demonstration activities towards a mission adoption in early 2027., Comment: 44 pages, 14 figures, accepted for publication in Experimental Astronomy

This paper proposes a novel k-medoids approximation algorithm to handle large-scale datasets with reasonable computational time and memory complexity. We develop a local-search algorithm that iteratively improves the medoid selection based on the estimation of the k-medoids objective. A single batch of size m << n provides the estimation, which reduces the required memory size and the number of pairwise dissimilarities computations to O(mn), instead of O(n^2) compared to most k-medoids baselines. We obtain theoretical results highlighting that a batch of size m = O(log(n)) is sufficient to guarantee, with strong probability, the same performance as the original local-search algorithm. Multiple experiments conducted on real datasets of various sizes and dimensions show that our algorithm provides similar performances as state-of-the-art methods such as FasterPAM and BanditPAM++ with a drastically reduced running time., Comment: Paper accepted by AAAI 2025

The International Semantic Web Research School (ISWS) is a week-long intensive program designed to immerse participants in the field. This document reports a collaborative effort performed by ten teams of students, each guided by a senior researcher as their mentor, attending ISWS 2023. Each team provided a different perspective to the topic of creative AI, substantiated by a set of research questions as the main subject of their investigation. The 2023 edition of ISWS focuses on the intersection of Semantic Web technologies and Creative AI. ISWS 2023 explored various intersections between Semantic Web technologies and creative AI. A key area of focus was the potential of LLMs as support tools for knowledge engineering. Participants also delved into the multifaceted applications of LLMs, including legal aspects of creative content production, humans in the loop, decentralised approaches to multimodal generative AI models, nanopublications and AI for personal scientific knowledge graphs, commonsense knowledge in automatic story and narrative completion, generative AI for art critique, prompt engineering, automatic music composition, commonsense prototyping and conceptual blending, and elicitation of tacit knowledge. As Large Language Models and semantic technologies continue to evolve, new exciting prospects are emerging: a future where the boundaries between creative expression and factual knowledge become increasingly permeable and porous, leading to a world of knowledge that is both informative and inspiring., Comment: Technical Report

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 2,700 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai., Comment: 27 pages, 6 figures

The advent of both multi-messenger and time-domain astrophysics over the last decade has seen a large interest in the development of small-scale, cheap, and robust gamma-ray detectors. This has been further encouraged by the availability of CubeSat platforms. Of particular interest are detectors capable of producing spectral and localization measurements of X and gamma-ray transients to allow for accurate follow-up measurements at different wavelengths. A vast number of the instruments developed for such purposes in the last years use a combination of scintillators and Silicon Photomultipliers (SiPMs) for photon detection. Here, we present the design, performance, and space qualification of a readout system capable of reading out 64 SiPM channels. This low-power and low-cost system was originally designed for the POLAR-2 mission, a large scale gamma-ray polarimeter. However, its flexible design makes it equally suitable for use on various CubeSat missions. The system was found to perform well when reading out both plastic and high Z scintillators using a total of 1.8~W. The space qualified design furthermore relies on commercial off-the-shelf components, thereby also removing most international export issues. In this paper, we will present the overall design, the performance of the electronics, its performance when reading out various scintillators and the successful space-qualification of this design., Comment: Submitted for publication in JINST

Exactly computing the full output distribution of linear optical circuits remains a challenge, as existing methods are either time-efficient but memory-intensive or memory-efficient but slow. Moreover, any realistic simulation must account for noise, and any viable quantum computing scheme based on linear optics requires feedforward. In this paper, we propose an algorithm that models the output amplitudes as partial derivatives of a multivariate polynomial. The algorithm explores the lattice of all intermediate partial derivatives, where each derivative is used to compute more efficiently ones with higher degree. In terms of memory, storing one path from the root to the leaves is sufficient to iterate over all amplitudes and requires only $2^n$ elements, as opposed to $\binom{n+m-1}{n}$ for the fastest state of the art method. This approach effectively balances the time-memory trade-off while extending to both noisy and feedforward scenarios with negligible cost. To the best of our knowledge, this is the first approach in the literature to meet all these requirements. We demonstrate how this method enables the simulation of systems that were previously out of reach, while providing a concrete implementation and complexity analysis.

A planet's albedo is a fundamental property that sets its energy budget by dictating the fraction of incident radiation absorbed versus reflected back to space. Generally, optical eclipse observations have revealed the majority of hot, giant planets to have low albedos, indicating dayside atmospheres dominated by absorption instead of reflection. However, there are several exceptions to this rule, including the ultra-hot-Neptune LTT 9779b, which have been found to have high geometric albedos. We observed four eclipses of LTT 9779b with the G280 grism of the Hubble Space Telescope's WFC3 UVIS mode; targeting the scattering signatures of the cloud condensate species causing the planet's elevated reflectivity. However, we do not definitively detect the planet's eclipse in our observations, with injection-recovery tests yielding a 3-$\sigma$ upper limit of 113 ppm on the eclipse depth of LTT 9779b in the 0.2-0.8$\mu$m waveband. We create reflectance spectrum grids for LTT 9779b's dayside using VIRGA/PICASO and compare to our UVIS limit, as well as previously published CHEOPS and TESS eclipse photometry. We find that silicate condensates are best able to explain LTT 9779b's highly-reflective dayside. Our forward model grids only enable weak constraints on vertical mixing efficiency, and suggest that, regardless of their particular composition, the clouds are likely composed of smaller and more reflective particles. Our work facilitates a deeper understanding of the reflectance properties of LTT 9779b as well as the UVIS spectroscopic mode itself, which will remain the community's primary access to UV wavelengths until next-generation telescopes like the Habitable Worlds Observatory., Comment: Accepted to MNRAS

Systematic expansion schemes in functional approaches require the inclusion of higher order vertices. These vertices are expanded in independent tensor bases with a rapidly increasing number of basis elements. Amongst the related tasks are the construction of bases and projection operators, the importance ordering of their elements, and the optimisation of such tensor bases, as well as an analysis of their regularity in momentum space. We present progress in all these directions and introduce the Mathematica package TensorBases designed for the aforementioned tasks., Comment: 24 pages, 2 figures

The Segment Anything Model (SAM) has revolutionized open-set interactive image segmentation, inspiring numerous adapters for the medical domain. However, SAM primarily relies on sparse prompts such as point or bounding box, which may be suboptimal for fine-grained instance segmentation, particularly in endoscopic imagery, where precise localization is critical and existing prompts struggle to capture object boundaries effectively. To address this, we introduce S4M (Segment Anything with 4 Extreme Points), which augments SAM by leveraging extreme points -- the top-, bottom-, left-, and right-most points of an instance -- prompts. These points are intuitive to identify and provide a faster, structured alternative to box prompts. However, a na\"ive use of extreme points degrades performance, due to SAM's inability to interpret their semantic roles. To resolve this, we introduce dedicated learnable embeddings, enabling the model to distinguish extreme points from generic free-form points and better reason about their spatial relationships. We further propose an auxiliary training task through the Canvas module, which operates solely on prompts -- without vision input -- to predict a coarse instance mask. This encourages the model to internalize the relationship between extreme points and mask distributions, leading to more robust segmentation. S4M outperforms other SAM-based approaches on three endoscopic surgical datasets, demonstrating its effectiveness in complex scenarios. Finally, we validate our approach through a human annotation study on surgical endoscopic videos, confirming that extreme points are faster to acquire than bounding boxes.

General-purpose multilingual vector representations, used in retrieval, regression and classification, are traditionally obtained from bidirectional encoder models. Despite their wide applicability, encoders have been recently overshadowed by advances in generative decoder-only models. However, many innovations driving this progress are not inherently tied to decoders. In this paper, we revisit the development of multilingual encoders through the lens of these advances, and introduce EuroBERT, a family of multilingual encoders covering European and widely spoken global languages. Our models outperform existing alternatives across a diverse range of tasks, spanning multilingual capabilities, mathematics, and coding, and natively supporting sequences of up to 8,192 tokens. We also examine the design decisions behind EuroBERT, offering insights into our dataset composition and training pipeline. We publicly release the EuroBERT models, including intermediate training checkpoints, together with our training framework., Comment: 26 pages, 6 figures, 11 tables

In the framework of the MACIV project, a consortium of French laboratories has deployed a temporary seismic network of 100 broadband stations in the French Massif Central (FMC) for 3-4 years (2023-2027). The project aims at imaging the crust and upper mantle of the FMC to better assess the sources of volcanism, and the impacts of the Variscan inheritance or the Cenozoic rift system on volcanic systems. A large-scale array of 35 broadband stations covers the entire FMC and complements the permanent networks to reach a homogeneous coverage with ~35 km spacing. This network, with XP code, is the French contribution to AdriaArray. The XP array is complemented with 3 quasi-linear north-south, east-west and northwest-southeast profiles with inter-station spacing of 5-20 km, making up the XF network of 65 stations. The profiles cross volcanic areas and the main Variscan structures. We describe the experimental setup designed to optimize the performance/cost ratio and minimize the number of field visits, the deployment, the state-of-health monitoring, the data management and the data quality control strategies, outcomes of our 15-years' experience with major temporary seismic experiments in France and neighboring countries, including AlpArray. We also show some preliminary results including hypocenter locations and receiver function analysis. The 2 broadband arrays will be supplemented in 2025 by a month-long deployment of 3 large-N dense arrays of 625 3-C short-period nodes. These dense arrays will complete our multi-scale seismic experiment and illuminate active faults and possible plumbing systems of the youngest volcanoes.

Gravitational wave observations are a powerful tool to constrain fundamental physics. This work considers dark matter that carries charge under a dark abelian massive vector field. If such dark matter is bound inside coalescing neutron stars, the presence of the new force will modify the total energy of the binary, and the emission of dark radiation modes will impact the rate of inspiral. At leading order, the dark matter corrections introduce a Yukawa term to the potential energy and a dipole radiation mode. We calculate modifications to the binary inspiral waveform to first post-Newtonian order, incorporate these corrections into a state-of-the-art waveform model for neutron star binaries, and use this model in a Bayesian parameter estimation analysis on real data collected by current, second-generation ground-based detectors. Through this study, we place the first robust constraints on the Yukawa interaction strength and dipole emission parameter. We find that the strongest constraints are obtained from analysis of the GW170817 event, for which the Yukawa interaction strength is constrained to be less than $O(10^{-2})$ and the dipole emission parameter is less than $O(10^{-4})$ at 90% credibility., Comment: 33 pages, 10 figures

Orbital magnetization, a key property arising from the orbital motion of electrons, plays a crucial role in determining the magnetic behavior of molecules and solids. Despite its straightforward calculation in finite systems, the computation in periodic systems poses challenges due to the ill-defined position operator and surface current contributions. The modern theory of orbital magnetization, implemented in the Density Functional Theory (DFT) framework, offers an accurate solution via the "converse approach." Here, we introduce QE-CONVERSE, a refactored, modular implementation of this method, replacing outdated routines from Quantum ESPRESSO (version 3.2). QE-CONVERSE integrates modern computational libraries like scaLAPACK and ELPA, enhancing scalability, especially for large supercell calculations. This work focuses on providing the community with a reliable, accurate orbital magnetization package for properties such as Electron Paramagnetic Resonance (EPR) g-tensors and Nuclear Magnetic Resonance (NMR) chemical shifts, particularly where perturbative methods fail. We demonstrate QE-CONVERSE's effectiveness with benchmark cases, including the NMR shifts of ${}^{27}$Al in alumina and ${}^{17}$O and ${}^{29}$Si in $\alpha$-quartz, as well as the EPR g-tensor for $_{ }^{n}\Sigma(n\geq 2)$ radicals and nitrogen defects in silicon. Results show excellent agreement with theoretical and experimental data, with improved accuracy in EPR calculations over linear response methods. QE-CONVERSE is fully compatible with recent Quantum ESPRESSO versions, enabling new possibilities for studying complex materials

Ongoing demands for smaller and more energy efficient electronic devices necessitate alternative interconnect materials with lower electrical resistivity at reduced dimensions. Despite the emergence of many promising candidates, synthesizing high quality nanostructures remains a major bottleneck in evaluating their performance. Here, we report the successful synthesis of Weyl semimetal NbAs nanowires via thermomechanical nanomolding, achieving single crystallinity and controlled diameters as small as 40 nm. Our NbAs nanowires exhibit a remarkably low room-temperature resistivity of 9.7 +/- 1.6 microOhm-cm, which is three to four times lower than their bulk counterpart. Theoretical calculations corroborate the experimental observations, attributing this exceptional resistivity reduction to surface dominant conduction with long carrier lifetime at finite temperatures. Further characterization of NbAs nanowires and bulk single crystals reveals high breakdown current density, robust stability, and superior thermal conductivity. Collectively, these properties highlight the strong potential of NbAs nanowires as next-generation interconnects, which can surpass the limitations of current copper-based interconnects. Technologically, our findings present a practical application of topological materials, while scientifically showcasing the fundamental properties uniquely accessible in nanoscale platforms., Comment: 5 figures

Assessing the quality of automatic image segmentation is crucial in clinical practice, but often very challenging due to the limited availability of ground truth annotations. In this paper, we introduce In-Context Reverse Classification Accuracy (In-Context RCA), a novel framework for automatically estimating segmentation quality in the absence of ground-truth annotations. By leveraging recent in-context learning segmentation models and incorporating retrieval-augmentation techniques to select the most relevant reference images, our approach enables efficient quality estimation with minimal reference data. Validated across diverse medical imaging modalities, our method demonstrates robust performance and computational efficiency, offering a promising solution for automated quality control in clinical workflows, where fast and reliable segmentation assessment is essential. The code is available at https://github.com/mcosarinsky/In-Context-RCA.

The electrification of road transport, as the predominant mode of transportation in Africa, represents a great opportunity to reduce greenhouse gas emissions and dependence on costly fuel imports. However, it introduces major challenges for local energy infrastructures, including the deployment of charging stations and the impact on often fragile electricity grids. Despite its importance, research on electric mobility planning in Africa remains limited, while existing planning tools rely on detailed local mobility data that is often unavailable, especially for privately owned passenger vehicles. In this study, we introduce a novel framework designed to support private vehicle electrification in data-scarce regions and apply it to Addis Ababa, simulating the mobility patterns and charging needs of 100,000 electric vehicles. Our analysis indicate that these vehicles generate a daily charging demand of approximately 350 MWh and emphasize the significant influence of the charging location on the spatial and temporal distribution of this demand. Notably, charging at public places can help smooth the charging demand throughout the day, mitigating peak charging loads on the electricity grid. We also estimate charging station requirements, finding that workplace charging requires approximately one charging point per three electric vehicles, while public charging requires only one per thirty. Finally, we demonstrate that photovoltaic energy can cover a substantial share of the charging needs, emphasizing the potential for renewable energy integration. This study lays the groundwork for electric mobility planning in Addis Ababa while offering a transferable framework for other African cities.

Primordial black holes (PBHs) in the asteroid-mass range remain a viable and until now unconstrained dark matter (DM) candidate. If such PBHs exist, they could be captured by stars in DM-dominated environments with low velocity dispersion such as ultra-faint dwarf galaxies (UFDs). The capture probability increases with the stellar mass, and captured PBHs would rapidly destroy their host stars. As a result, the presence of PBHs in UFDs would alter their stellar mass functions. Using photometric observations of three ultra-faint dwarf galaxies from the Hubble Space Telescope, we show that it is unlikely that their mass functions have been significantly modified by PBHs, and we place constraints on the PBH abundance. In the ultra-faint dwarf galaxy Triangulum II, PBHs around $10^{19}$g are excluded at the $2\sigma$ ($3\sigma$) level from constituting more than $\sim55\%$ ($\sim78\%$) of the dark matter, while the possibility that PBHs represent the entirety of the DM is excluded at the $3.7\sigma$ level., Comment: 9+5 pages. Comments are welcome!

We show that for any fixed $(2+1)$-dimensional TQFT over $\mathbb{C}$ of either Turaev-Viro-Barrett-Westbury or Reshetikhin-Turaev type, the problem of (exactly) computing its invariants on closed 3-manifolds is either solvable in polynomial time, or else it is $\#\mathsf{P}$-hard to (exactly) contract certain tensors that are built from the TQFT's fusion category. Our proof is an application of a dichotomy result of Cai and Chen [J. ACM, 2017] concerning weighted constraint satisfaction problems over $\mathbb{C}$. We leave for future work the issue of reinterpreting the conditions of Cai and Chen that distinguish between the two cases (i.e. $\#\mathsf{P}$-hard tensor contractions vs. polynomial time invariants) in terms of fusion categories. We expect that with more effort, our reduction can be improved so that one gets a dichotomy directly for TQFTs' invariants of 3-manifolds rather than more general tensors built from the TQFT's fusion category., Comment: 16 pages, 11 figures. Comments welcome!

Context. Radiation plays a significant role in solar and astrophysical environments as it may constitute a sizeable fraction of the energy density, momentum flux, and the total pressure. Modelling the dynamic interaction between radiation and magnetized plasmas in such environments is an intricate and computationally costly task. Aims. The goal of this work is to demonstrate the capabilities of the open-source parallel, block-adaptive computational framework MPI-AMRVAC, in solving equations of radiation-magnetohydrodynamics (RMHD), and to present benchmark test cases relevant for radiation-dominated magnetized plasmas. Methods. The existing magnetohydrodynamics (MHD) and flux-limited diffusion (FLD) radiative-hydrodynamics physics modules are combined to solve the equations of radiation-magnetohydrodynamics (RMHD) on block-adaptive finite volume Cartesian meshes in any dimensionality. Results. We introduce and validate several benchmark test cases such as steady radiative MHD shocks, radiation-damped linear MHD waves, radiation-modified Riemann problems and a multi-dimensional radiative magnetoconvection case. We recall the basic governing Rankine-Hugoniot relations for shocks and the dispersion relation for linear MHD waves in the presence of optically thick radiation fields where the diffusion limit is reached. The RMHD system allows for 8 linear wave types, where the classical 7-wave MHD picture (entropy and three wave pairs for slow, Alfven and fast) is augmented with a radiative diffusion mode. Conclusions. The MPI-AMRVAC code now has the capability to perform multidimensional RMHD simulations with mesh adaptation making it well-suited for larger scientific applications to study magnetized matter-radiation interactions in solar and stellar interiors and atmospheres., Comment: Accepted to Astronomy and Astrophysics journal

Thermionics and thermophotovoltaics are solid-state technologies that convert high-temperature heat into electricity by utilizing fundamental particles, electrons in thermionics and photons in thermophotovoltaics, as energy carriers. Both systems have the potential to achieve high efficiency and power density, contingent on the optimization of radiative/electronic energy fluxes. A critical factor in enhancing energy flux in these devices is the introduction of microscale (thermionics) or nanoscale (thermophotovoltaics) gaps between the hot thermal emitter and the cooler receiver. In thermionic converters, microscale gaps mitigate space charge effects that create energy barriers to electron flow. For thermophotovoltaic systems, nanoscale gaps facilitate photon tunneling, significantly boosting photon flux towards the thermophotovoltaic cell. Forming these small-scale gaps often necessitates intermediate materials or spacers between the emitter and receiver. Over the past few decades, various spacer designs have been proposed and studied, demonstrating their effectiveness in enhancing energy transfer and conversion. However, challenges remain regarding their reliability and scalability. This article provides a comprehensive overview of spacer technologies for thermionics and thermophotovoltaics and summarizes recent advancements, current capabilities, and persistent challenges.

Light-based detectors have been widely used in fundamental research and industry since their inception in the 1930s. The energy particles deposit in these detectors is converted to optical signals via the Cherenkov and scintillation mechanisms that are then propagated through transparent media to photosensors placed typically on the detector's periphery, sometimes up to tens of metres away. LiquidO is a new technique pioneering the use of opaque media to stochastically confine light around each energy deposition while collecting it with an array of fibres that thread the medium. This approach preserves topological event information otherwise lost in the conventional approach, enabling real-time imaging down to the MeV scale. Our article demonstrates LiquidO's imaging principle with a ten-litre prototype, revealing successful light confinement of 90% of the detected light within a 5 cm radius sphere, using a custom opaque scintillator with a scattering length on the order of a few millimetres. These high-resolution imaging capabilities unlock opportunities in fundamental physics research and applications beyond. The absolute amount of light detected is also studied, including possible data-driven extrapolations to LiquidO-based detectors beyond prototyping limitations. Additionally, LiquidO's timing capabilities are explored through its ability to distinguish Cherenkov light from a slow scintillator., Comment: 22 pages, 9 figures, 2 tables

Incomplete multiview clustering (IMVC) has gained significant attention for its effectiveness in handling missing sample challenges across various views in real-world multiview clustering applications. Most IMVC approaches tackle this problem by either learning consensus representations from available views or reconstructing missing samples using the underlying manifold structure. However, the reconstruction of learned similarity graph tensor in prior studies only exploits the low-tubal-rank information, neglecting the exploration of inter-view correlations. This paper propose a novel joint tensor and inter-view low-rank Recovery (JTIV-LRR), framing IMVC as a joint optimization problem that integrates incomplete similarity graph learning and tensor representation recovery. By leveraging both intra-view and inter-view low rank information, the method achieves robust estimation of the complete similarity graph tensor through sparse noise removal and low-tubal-rank constraints along different modes. Extensive experiments on both synthetic and real-world datasets demonstrate the superiority of the proposed approach, achieving significant improvements in clustering accuracy and robustness compared to state-of-the-art methods., Comment: The paper is under review at IEEE Transactions on Knowledge and Data Engineering

Fragmentation and breakage of rocks is essential to iron ore mining and extraction. The breakage energy requirements and resulting ore particle size and mineral distributions are key to understanding and optimising mining and processing practices. This study combines high-speed and experimental X-ray micro-CT (micro-computed tomography) imaging with 3D image analyses to study the fragmentation of iron ore particles. Here a case study of a hematite-rich iron ore particle is used to illustrate the application and results produced by this imaging procedure. The particle was micro-CT scanned before a high-speed camera was used to image particle breakage in a custom drop weight test, capturing the dynamic processes of particle fracturing and subsequent fragmentation at a resolution of 50 microseconds. Fragments produced were collected, micro-CT scanned and analysed in three-dimension by particle shape, size, and volume., Comment: 9 pages, 7 figures

Non-ergodicity impacts statistical inference in a diverse range of disciplines inside and outside of physics. However the concept of ergodicity is used inconsistently, and may refer to several nonequivalent notions. To help address this, we first identify and clarify the relationship between three major interpretations of ergodicity. We then introduce a method of spectral analysis of non-ergodicity which may be performed using data alone, and so may be applied in both numerical and empirical contexts. This may be used to identify, quantify, and classify non-ergodic populations within an ergodic decomposition. This is demonstrated with an application to the Kob-Andersen kinetically constrained lattice glass model., Comment: 6 pages, 3 figures

Cu is an ubiquitous material used in industry for its great thermal and electrical properties. Studying the high-pressure high temperature properties of copper (Cu) is relevant for nuclear fusion research as projectiles and flyers used in hypervelocity impacts are generally made of copper, where it is used also in the design of the nuclear fusion targets. Recently, a solid (fcc)-solid (bcc) phase transition has been detected in shock compressed Cu with X-Ray Diffraction. Here, we present a study on shock compressed copper up to 300 GPa and 7100 K probed by single pulse (100 ps FWHM) X-ray Absorption Spectroscopy (XAS). Based on the analysis of the XAS spectra, we provide structural identification and bulk temperature measurements along the Hugoniot up to the melting. The collection of XAS spectra under release conditions, i.e. at later times than the breakout time of the shock wave, helped constraining the experimental fcc-bcc and solid-liquid phase boundaries. In particular, we report the first bulk temperature measurement in shock compressed copper on the melting plateau located between 237(40) GPa and 5750(1130) K and 261 (27) GPa and 6240 (1155) K and on liquid copper at 300 GPa and 7100 K.

We present a coherent-population trapping (CPT) microcell atomic clock using symmetric auto-balanced Ramsey (SABR) spectroscopy. The pulsed SABR sequence is applied through direct current-based power modulation of the vertical-cavity surface-emitting laser, eliminating the need for an external optical shutter and enabling compatibility with fully-integrated clocks. The sequence is controlled by a single FPGA-based digital electronics board. A key aspect of proper clock operation was the implementation of a real-time tracking of the atomic signal detection window. The clock frequency dependence on laser power, microwave power, laser frequency, and timing of the detection window has been measured, obtaining sensitivity coefficients lower than those obtained with Ramsey-CPT spectroscopy. The Allan deviation of the SABR-CPT clock, based on a microfabricated cell with low-permeation glass windows, is 1.1~$\times$~10$^{-9}$ at 1~s and averages down to the low 10$^{-12}$ range at 1 day integration time. These results pave the way towards the development of Ramsey-CPT chip-scale atomic clocks with enhanced timekeeping performances., Comment: 9 pages, 7 figures

Sharpness-Aware Minimization (SAM) has emerged as a powerful method for improving generalization in machine learning models by minimizing the sharpness of the loss landscape. However, despite its success, several important questions regarding the convergence properties of SAM in non-convex settings are still open, including the benefits of using normalization in the update rule, the dependence of the analysis on the restrictive bounded variance assumption, and the convergence guarantees under different sampling strategies. To address these questions, in this paper, we provide a unified analysis of SAM and its unnormalized variant (USAM) under one single flexible update rule (Unified SAM), and we present convergence results of the new algorithm under a relaxed and more natural assumption on the stochastic noise. Our analysis provides convergence guarantees for SAM under different step size selections for non-convex problems and functions that satisfy the Polyak-Lojasiewicz (PL) condition (a non-convex generalization of strongly convex functions). The proposed theory holds under the arbitrary sampling paradigm, which includes importance sampling as special case, allowing us to analyze variants of SAM that were never explicitly considered in the literature. Experiments validate the theoretical findings and further demonstrate the practical effectiveness of Unified SAM in training deep neural networks for image classification tasks., Comment: 13th International Conference on Learning Representations (ICLR 2025)

We analyze a class of general nonlinear epidemic models with age and space structure, including a nonlocal infection term depending on age and space. After establishing the well-posedness of the state partial differential equation, we introduce a control parameter interpreted as a vaccination rate. Under certain conditions, we show that an optimal control exists and how it can be characterized by first-order optimality conditions. Finally, we present numerical examples of the optimal control problems governed by these models.

The end points of massive star evolution are poorly known, especially those in interacting binary systems containing compact objects, such as neutron stars or black holes. Such systems are bright in X-rays, and the most luminous among them are called ultra-luminous X-ray sources (ULXs). In this paper, we address the enigmatic NGC 300 ULX-1. It's X-ray activity started in 2010 with the supernova impostor-like event SN 2010da. In the following few years the ULX was powered by persistent super-Eddington accretion but then it dimmed in X-rays. We present the most recent X-ray and optical observations. The Chandra and Swift telescopes confirm that SN 2010da/NGC 300 ULX-1 is not accreting at super-Eddington level anymore. We attribute this switch in accretion regime to the donor star variability and its fast evolution. In order to gain a better understanding of the donor star's nature, we consider its optical light curve on a decade-long time scale and show that the optical counterpart of SN 2010da/NGC 300 ULX-1 dimmed significantly over recent years. The most recent detection in optical by the Gemini telescope reveals that the source is now > 2.5 mag fainter in the r' band compared to the epoch when it was spectroscopically classified as a red supergiant. We discuss the nature of the abrupt changes in the donor star properties, and consider among other possibilities the silent collapse of the donor star into a black hole., Comment: 8 pages, 3 figures, 1 table, updated chronology of events

We study tridiagonal pairs of type II. These involve two linear transformations $A$ and $A^\star$. We define two bases. In the first one, $A$ acts as a diagonal matrix while $A^\star$ acts as a block tridiagonal matrix, and in the second one, $A$ acts as a block tridiagonal matrix while $A^\star$ acts as a diagonal matrix. We obtain the change of basis coefficients between these two bases. The coefficients are special functions that are written as a nested product of polynomials that resemble Racah polynomials but involve shift operators in their expression., Comment: 19 pages

Gravitational waves from individually resolved compact object mergers can be used as standard sirens, offering a novel self-calibrating precision probe of cosmology. While the standard siren method has been well-explored, the gravitational-wave background arising from unresolved mergers offers a novel alternative to probing cosmology. We demonstrate that the combination of resolved binary black hole mergers with the unresolved signals composing the stochastic gravitational-wave background can be used to measure cosmological parameters, including the Hubble constant, $H_0$. We apply this ``stochastic siren'' method to existing gravitational-wave data and find that including the current non-detection of the background increases the accuracy at which $H_0$ can be measured, relative to using resolved mergers alone. We also provide projections for upcoming detectors to highlight their ability to probe cosmology with the background. With the anticipated detection of the astrophysical gravitational-wave background, the stochastic siren approach can be expected to improve future standard siren cosmological measurements.

Carbonate minerals are important in Earth's system sciences and have been found on Mars and in meteorites and asteroids, highlighting the importance of impacts in planetary processes. While extensively studied under static compression, the behavior of carbonates under shock compression remains underexplored, with no $\textit{in situ}$ X-ray investigations reported so far. Here we investigate natural magnesiosiderite (Fe$_{0.6}$Mg$_{0.4}$CO$_{3}$) under nanosecond laser-driven shock compression at pressures up to 150 GPa, coupled with $\textit{in situ}$ ultrafast synchrotron X-ray absorption spectroscopy (XAS). The interpretation of the experimental spectra is complemented using first-principles absorption cross-section calculations performed on crystalline phases at different pressures and on a dense liquid phase obtained using density functional theory-based molecular dynamics (DFT-MD) simulations. Under laser-driven shock compression, the magnesiosiderite crystal phase remains unchanged up to the melt. Under shock reverberation, the absorption spectra show changes similar to those attributed to a high-spin to low-spin transition observed under static compression. At higher pressures, the laser shock induces the formation of CO$_4$ tetrahedral units in the melt. Upon unloading from the shocked state, only a few nanoseconds later, the original magnesiosiderite phase is recovered., Comment: 42 pages, 4 main text figures, 11 supplementary text figures

The emergence of AI-augmented Data Processing Systems (DPSs) has introduced powerful semantic operators that extend traditional data management capabilities with LLM-based processing. However, these systems face fundamental reliability (a.k.a. trust) challenges, as LLMs can generate erroneous outputs, limiting their adoption in critical domains. Existing approaches to LLM constraints--ranging from user-defined functions to constrained decoding--are fragmented, imperative, and lack semantics-aware integration into query execution. To address this gap, we introduce Semantic Integrity Constraints (SICs), a novel declarative abstraction that extends traditional database integrity constraints to govern and optimize semantic operators within DPSs. SICs integrate seamlessly into the relational model, allowing users to specify common classes of constraints (e.g., grounding and soundness) while enabling query-aware enforcement and optimization strategies. In this paper, we present the core design of SICs, describe their formal integration into query execution, and detail our conception of grounding constraints, a key SIC class that ensures factual consistency of generated outputs. In addition, we explore novel enforcement mechanisms, combining proactive (constrained decoding) and reactive (validation and recovery) techniques to optimize efficiency and reliability. Our work establishes SICs as a foundational framework for trustworthy, high-performance AI-augmented data processing, paving the way for future research in constraint-driven optimizations, adaptive enforcement, and enterprise-scale deployments.

In this work, the $L_p$ version (for $p> 1$) of the dimensional Brunn-Minkowski inequality for the standard Gaussian measure $\gamma_n(\cdot)$ on $\mathbb{R}^n$ is shown. More precisely, we prove that for any $0$-symmetric convex sets with nonempty interior, any $p>1$, and every $\lambda \in (0,1)$, \[ \gamma_n\bigl((1-\lambda)\cdot K+_p \lambda \cdot L\bigr)^{p/n} \geqslant (1-\lambda ) \gamma_n(K)^{p/n} + \lambda \gamma_n(L)^{p/n}, \] with equality, for some $\lambda \in (0,1)$ and $p>1$, if and only if $K=L$. This result, recently established without the equality conditions by Hosle, Kolesnikov and Livshyts, by using a different and functional approach, turns out to be the $L_p$ extension of a celebrated result for the Minkowski sum (that is, for $p=1$) by Eskenazis and Moschidis (2021) on a problem by Gardner and Zvavitch (2010). Moreover, an $L_p$ Brunn-Minkowski type inequality is obtained for the classical Wills functional $\mathcal{W}(\cdot)$ of convex bodies. These results are derived as a consequence of a more general approach, which provides us with other remarkable examples of functionals satisfying $L_p$ Brunn-Minkowski type inequalities, such as different absolutely continuous measures with radially decreasing densities., Comment: Improved presentation. Corrected typos. Main results unchanged

Retrieval in 3D point clouds is a challenging task that consists in retrieving the most similar point clouds to a given query within a reference of 3D points. Current methods focus on comparing descriptors of point clouds in order to identify similar ones. Due to the complexity of this latter step, here we focus on the acceleration of the retrieval by adapting the Differentiable Search Index (DSI), a transformer-based approach initially designed for text information retrieval, for 3D point clouds retrieval. Our approach generates 1D identifiers based on the point descriptors, enabling direct retrieval in constant time. To adapt DSI to 3D data, we integrate Vision Transformers to map descriptors to these identifiers while incorporating positional and semantic encoding. The approach is evaluated for place recognition on a public benchmark comparing its retrieval capabilities against state-of-the-art methods, in terms of quality and speed of returned point clouds., Comment: 8 pages, 1 figures

We consider the weak-error rate of the SPDE approximation by regularized Dean-Kawasaki equation with It\^o noise for particle systems with mean-field interactions both on the drift and the noise. The global existence and uniqueness of the corresponding SPDEs are established using the variational approach to SPDEs, and the weak-error rate is estimated using the technique of Kolmogorov equations on the space of probability measures. In particular, the rate derived in this paper coincides with that is the previous work arXiv:2212.11714, which considered free Brownian particles using Laplace duality., Comment: 54 pages

Self-supervised learning has become a key method for training deep learning models when labeled data is scarce or unavailable. While graph machine learning holds great promise across various domains, the design of effective pretext tasks for self-supervised graph representation learning remains challenging. Contrastive learning, a popular approach in graph self-supervised learning, leverages positive and negative pairs to compute a contrastive loss function. However, current graph contrastive learning methods often struggle to fully use structural patterns and node similarities. To address these issues, we present a new method called Fused Gromov Wasserstein Subgraph Contrastive Learning (FOSSIL). Our model integrates node-level and subgraph-level contrastive learning, seamlessly combining a standard node-level contrastive loss with the Fused Gromov-Wasserstein distance. This combination helps our method capture both node features and graph structure together. Importantly, our approach works well with both homophilic and heterophilic graphs and can dynamically create views for generating positive and negative pairs. Through extensive experiments on benchmark graph datasets, we show that FOSSIL outperforms or achieves competitive performance compared to current state-of-the-art methods.

We investigate $L_2$-discrepancies of what we call weak Latin hypercubes. In this case it turns out that there is a precise equivalence between the extreme and periodic $L_2$-discrepancy which follows from a much broader result about generalized energies for weighted point sets. Motivated by this we study the asymptotics of the optimal $L_2$-discrepancy of weak Latin hypercubes. We determine asymptotically tight bounds for $d \geq 3$ and even the precise (dimension dependent) constant in front of the dominating term for $d \geq 4$., Comment: To appear in Monatshefte f\"ur Mathematik

We present polarization pulse profiles for 56 millisecond pulsars (MSPs) monitored by the Chinese Pulsar Timing Array (CPTA) collaboration using the Five-hundred-meter Aperture Spherical radio Telescope (FAST). The observations centered at 1.25 GHz with a raw bandwidth of 500 MHz. Due to the high sensitivity ($\sim$16 K/Jy) of the FAST telescope and our long integration time, the high signal-to-noise ratio polarization profiles show features hardly detected before. Among 56 pulsars, the polarization profiles of PSRs J0406$+$3039, J1327$+$3423, and J2022$+$2534 were not previously reported. 80\% of MSPs in the sample show weak components below 3\% of peak flux, 25\% of pulsars show interpulse-like structures, and most pulsars show linear polarization position angle jumps. Six pulsars seem to be emitting for full rotation phase, with another thirteen pulsars being good candidates for such a 360$^\circ$ radiator. We find that the distribution of the polarization percentage in our sample is compatible with the normal pulsar distribution. Our detailed evaluation of the MSP polarization properties suggests that the wave propagation effects in the pulsar magnetosphere are important in shaping the MSP polarization pulse profiles., Comment: 17 pages, 10 figures, 2 tables, accepted by A&A

Current cosmological data are well-described by the Lambda-Cold Dark Matter ($\Lambda$CDM) model, which assumes adiabatic initial conditions for the primordial density perturbations. This agreement between data and theory enables strong constraints on new physics that generates isocurvature perturbations. Existing constraints typically assume a simple power law form for the isocurvature power spectrum. However, many new physics scenarios -- such as cosmological phase transitions and gravitational particle production -- can deviate from this assumption. To derive general constraints which apply to a wide variety of new physics scenarios, we consider four types of isocurvature modes (dark matter, baryon, dark radiation and neutrino density isocurvature) and parametrize the isocurvature power spectrum using two general forms: a delta function and a broken power law. Using data from the cosmic microwave background (CMB), baryon acoustic oscillations, the Lyman-$\alpha$ forest, and CMB spectral distortions, we place constraints on the isocurvature power spectrum across a wide range of scales, from $10^{-4}\,\textrm{Mpc}^{-1}$ to $10^{4}\,\textrm{Mpc}^{-1}$., Comment: 13 pages, 4 figures

Deepfakes - manipulated or forged audio and video media - pose significant security risks to individuals, organizations, and society at large. To address these challenges, machine learning-based classifiers are commonly employed to detect deepfake content. In this paper, we assess the robustness of such classifiers through a systematic penetration testing methodology, which we introduce as DeePen. Our approach operates without prior knowledge of or access to the target deepfake detection models. Instead, it leverages a set of carefully selected signal processing modifications - referred to as attacks - to evaluate model vulnerabilities. Using DeePen, we analyze both real-world production systems and publicly available academic model checkpoints, demonstrating that all tested systems exhibit weaknesses and can be reliably deceived by simple manipulations such as time-stretching or echo addition. Furthermore, our findings reveal that while some attacks can be mitigated by retraining detection systems with knowledge of the specific attack, others remain persistently effective. We release all associated code.

Evaluation has traditionally focused on ranking candidates for a specific skill. Modern generalist models, such as Large Language Models (LLMs), decidedly outpace this paradigm. Open-ended evaluation systems, where candidate models are compared on user-submitted prompts, have emerged as a popular solution. Despite their many advantages, we show that the current Elo-based rating systems can be susceptible to and even reinforce biases in data, intentional or accidental, due to their sensitivity to redundancies. To address this issue, we propose evaluation as a 3-player game, and introduce novel game-theoretic solution concepts to ensure robustness to redundancy. We show that our method leads to intuitive ratings and provide insights into the competitive landscape of LLM development., Comment: Published at ICLR 2025

The role of complex quantities in quantum theory has been puzzling physicists since the beginnings. Recently, it was shown that they are inevitable in network scenarios with independent sources. Indeed, quantum theory based on real Hilbert spaces cannot explain the predictions of quantum theory over complex Hilbert spaces [Renou et al., Nature 600, 2021]. Here, we revisit the independence assumption underlying this work. We show that assuming partial independence is sufficient for showing the inadequacy of quantum theory over real Hilbert spaces. We derive a trade-off between source independence and the Bell value achievable in real quantum theory, which also lower bounds the source correlations required to explain previous experiments with real quantum systems. We further show that 1 bit of entanglement is necessary and sufficient for recovering the complex quantum correlations by means of real quantum theory in the scenario from [Renou et al., Nature 600, 2021]. Finally, building on [McKague et al., PRL 102, 2009], we provide a construction to simulate any complex quantum setup with m independent sources by means of real quantum theory, by allowing the sources to share a m real-qubit entangled state in the first round of the experiment., Comment: 16 pages, 1 figure

The origin of the second harmonic generation (SHG) from plasmonic structures remains a subject of debate. Here, we investigate SHG from gold plasmonic nanostructures by scanning nanostructures at the single-particle level to acquire two-dimensional (2D) SHG maps which are qualitatively and quantitatively compared with numerically simulated SHG maps. This approach allows us to discriminate the role played by the surface and bulk contributions to SHG. Our findings reveal that the tangential surface source contribution, often neglected in the literature, plays a dominant role and is affected by surface roughness. This is evident when comparing the SHG maps obtained from identically sized gold nanoparticles that are fabricated using physical and chemical techniques. Additionally, the normal surface source is not found to significantly impact SHG in these configurations.

"Fluffy" hydrogenated amorphouscarbon(a-C:H)wassynthesizedusingadielectric barrier discharge plasma, driven by nanosec ond high voltage pulses at 1 kHz frequency in a helium-butane mixture. The a-C:H samples were characterized by scanning and transmission electron microscopy, laser-assisted and secondary ion mass spectrometry, and Raman and Fourier-transform infrared spectroscopy. We find that a-C:H samples exhibit infrared absorption features in good agreement with those observed for carbonaceous dust in IRAS 08572+3915 galaxy. We discuss their nano to microscale structure and derive their hydrogen to carbon (H/C) ratios from the results obtained by three distinct experimental characterization techniques. Relying on the average H/C value determined by mass spectrometry and Raman spectroscopy, we can then constrain the absorption strengths values to those best corresponding to our dust analogue, and calculate the H/C ratio from the infrared spectra. Altogether, we find that our dust analogue consists of a dominant hydrogen-rich aliphatic network, with small, isolated, aromatic regions. The a-C:H dust analogue was then irradiated with 3 MeV H+ and subsequently analyzed ex situ. Morphological and chemical changes, including the evolution of H/C, CH2/CH3, and sp2/sp3 ratios, were observed with increasing proton fluence, indicating dehydrogenation and graphitization. Proton bombardment shifted the initial location of a-C:H in the hydrocarbon ternary phase diagram toward the central region defined by IRAS 08572+3915 observations. The decay of the 3.4 {\mu}m band with proton fluence was used to calculate CH destruction cross-sections, results consistent with a direct effect of cosmic rays on the disappearance of the 3.4 {\mu}m band.

Used for centuries to enhance mechanical properties of materials, solid solution strengthening (SSS) is a classical metallurgical method in which small amounts of impurity elements are added to a base metal. Developed for dilute alloys, classical theories of SSS are presently challenged by the ongoing explosive development of complex concentrated alloys (CCA) in which all component elements are present in nearly equal fractions. Here we develop a method of computational alchemy in which interatomic interactions are modified to continuously and systematically vary two key parameters defining SSS, atomic size misfit and elastic stiffness misfit, over a maximally wide range of misfit values. The resulting model alloys are subjected to massive Molecular Dynamics simulations reproducing full complexity of plastic strength response in concentrated single-phase body-centered cubic solid solutions. At variance with views prevailing in the literature, our computational experiments show that stiffness misfit can contribute to SSS on par if not more than size misfit. Furthermore, depending on exactly how they are combined, the two misfits can result in synergistic or antagonistic effect on alloy strengthening. In contrast to real CCAs in which every constituent element comes with its specific combination of atomic size and elastic stiffness, our alchemical model alloys sample the space of misfit parameters continuously thus augmenting the much more constrained and inevitably spotty experimental exploration of the CCA design space. Taking advantage of unique to our approach ability to define alloy misfit parameters, our computational study demonstrates how useful insights can be gained from intentionally unrealistic alchemical models. Rather than practical recommendation for alloy design, our computational experiments should be regarded as a proving ground for further SSS theory development., Comment: 23 pages, 9 figures, submitted to Matter

Spontaneous symmetry breaking, arising from the competition of interactions and quantum fluctuations, is fundamental to understanding ordered electronic phases. Although electrically neutral, optical excitations like excitons can interact through their dipole moment, raising the possibility of optically active ordered phases. The effects of spontaneous ordering on optical properties remain largely unexplored. Recent observations of the excitonic Mott insulating state in semiconducting moir\'e crystals make them promising for addressing this question. Here, we present evidence for an in-plane ferroelectric phase of dipolar moir\'e excitons driven by strong exciton-exciton interactions. We discover a surprising speed-up of photon emission at late times and low densities in excitonic decay. This counterintuitive behavior is attributed to collective radiance, linked to the transition between disordered and symmetry-broken ferroelectric phases of moir\'e excitons. Our findings provide first evidence for strong dipolar inter-site interactions in moir\'e lattices, demonstrate collective photon emission as a probe for moir\'e quantum materials, and pave the way for exploring cooperative optical phenomena in strongly correlated systems.