Search

The Every Student Succeeds Act of 2015 requires states to use a variety of indicators, including standardized tests and attendance records, to designate schools for support and improvement based on schoolwide performance and the performance of groups of students within schools. Schoolwide and group-level performance indicators are also diagnostically relevant for district-level and school-level decisionmaking outside the formal accountability context. Like all measurements, performance indicators are subject to measurement error, with some having more random error than others. Measurement error can have an outsized effect for smaller groups of students, rendering their measured performance unreliable, which can lead to misidentification of groups with the greatest needs. Many states address the reliability problem by excluding from accountability student groups smaller than an established threshold, but this approach sacrifices equity, which requires counting students in all relevant groups. With the aim of improving reliability, particularly for small groups of students, this study applied a stabilization model called Bayesian hierarchical modeling to group-level data (with groups assigned according to demographic designations) within schools in New Jersey. Stabilization substantially improved the reliability of test -based indicators, including proficiency rates and median student growth percentiles. The stabilization model used in this study was less effective for non test based indictors, such as chronic absenteeism and graduation rate, for several reasons related to their statistical properties. When stabilization is applied to the indicators best suited for it (such as proficiency and growth), it leads to substantial changes in the lists of schools designated for support and improvement. These results indicate that, applied correctly, stabilization can increase the reliability of performance indicators for processes using these indicators, simultaneously improving accuracy and equity.

While the United States leads the world in Artificial Intelligence (AI) innovation, the schools lag behind in preparing teachers and students for the Fourth Industrial Revolution. The extraordinary pace of technological change, and the potential for both opportunity and risk, may be unprecedented. What leaders in education and education policy do next matters. To help accelerate action in U.S. public education and develop a short-term roadmap for districts and other education leaders, the Center on Reinventing Public Education (CRPE) brought together over 60 state and federal policymakers, edtech innovators, school system leaders, and advocates in Albuquerque, New Mexico, in April 2024 to participate in the Think Forward: Learning with AI forum. This report reflects key learnings and conversations that emerged from CRPE's Think Forward convening, including: (1) how AI can enable needed changes in our schools; (2) how current conditions in the edtech market act as barriers to closing equity gaps; and (3) how policy and practice must adapt for lasting system change. It concludes with a short-term action plan developed by forum participants that provides an immediate path forward and outlines the roles wide-ranging stakeholders must play to address our shared challenges and opportunities.

The Center for Exascale Monte Carlo Neutron Transport is developing Monte Carlo / Dynamic Code (MC/DC) as a portable Monte Carlo neutron transport package for rapid numerical methods exploration on CPU- and GPU-based high-performance computers. In this paper, we describe MC/DC's current event-based GPU algorithm as well as the just-in-time (JIT) compilation scheme we use to enable GPU operability on Nvidia and AMD GPUs from MC/DC's Python source. To analyze performance, we conduct runtime tests of the C5G7 k-eigenvalue benchmark problem and a continuous-energy infinite pin cell on Nvidia Tesla V100 GPU, AMD MI250X GPU, and the AMD MI300A APU and make comparison to a dual-socket Intel Xeon Sapphire Rapid CPU node. We found that for the multi-group C5G7 benchmark problem, we respectively see a 15$\times$, 0.7$\times$, 12$\times$ speedup on a V100, MI250X, and MI300A over 112 Intel Xeon CPU cores. For the continuous-energy infinite pin-cell benchmark, we found speedups of 5$\times$, 3$\times$, 4$\times$ on a V100, MI250X, and MI300A, respectively, over the same CPU node., Comment: 14 pages, 3 figures

Accurate and efficient circuit models are necessary to control the power electronic circuits found on plasma physics experiments. Tuning and controlling the behavior of these circuits is inextricably linked to plasma performance. Linear models are greatly preferred for control applications due to their well-established performance guarantees, but they typically fail to capture nonlinear dynamics and changes in experimental parameters. Data-driven system identification can help mitigate these shortcomings by learning interpretable and accurate reduced-order models of a complex system, in this case the injector circuits of the Helicity Injected Torus - Steady Inductive Upgrade (HIT-SIU) experiment. Specifically, the Bagging Optimized Dynamic Mode Decomposition (BOP-DMD), is leveraged to learn stable, reduced order models of the interaction between the spheromak plasma formed in the confinement volume, and the injector circuits of the device. BOP-DMD is trained and evaluated on an analytic model of the vacuum dynamics of the injector circuits of HIT-SIU, as well as an analytic linear reduced-order model for the injector dynamics when a plasma is present. BOP-DMD is then fit on experimental data, both on shots with and without a plasma in the confinement volume. In doing so, we demonstrate the capability of data-driven methods to produce stable, linear models for control and uncertainty quantification in plasma experiments.

Determining the electronic phase diagram of a quantum material as a function of temperature (T) and applied magnetic field (H) forms the basis for understanding the microscopic origin of transport properties, such as the anomalous Hall effect (AHE) and topological Hall effect (THE). For many magnetic quantum materials, including EuAl$_4$, a THE arises from a topologically protected magnetic skyrmion lattice with a non-zero scalar spin chirality. We identified a square skyrmion lattice (sSkL) peak in Eu(Ga$_{1-x}$Al$_x$)$_4$ ($x$ = 0.9) identical to the peak previously observed in EuAl$_4$ by performing neutron scattering measurements throughout the phase diagram. Comparing these neutron results with transport measurements, we found that in both compounds the maximal THE does not correspond to the sSkL area. Instead of the maximal THE, the maximal magnetocaloric effect (MCE) boundaries better identify the sSkL lattice phase observed by neutron scattering measurements. The maximal THE therefore arises from interactions of itinerant electrons with frustrated spin fluctuations in a topologically trivial magnetic state.

In this paper, we describe the Multi-Band Template Analysis (MBTA) search pipeline dedicated to the detection of compact binary coalescence (CBC) gravitational wave signals from the data obtained by the LIGO-Virgo-KAGRA collaboration (LVK) during the fourth observing run (O4), which started in May 2023. We give details on the configuration of the pipeline and its evolution compared to the third observing run (O3). We focus here on the configuration used for the offline results of the first part of the run (O4a), which are part of the GWTC-4 catalog (in preparation). We also give a brief summary of the online configuration and highlight some of the changes implemented or considered for the second part of O4 (O4b)., Comment: 15 pages, 11 figures

Robotic guide dogs hold significant potential to enhance the autonomy and mobility of blind or visually impaired (BVI) individuals by offering universal assistance over unstructured terrains at affordable costs. However, the design of robotic guide dogs remains underexplored, particularly in systematic aspects such as gait controllers, navigation behaviors, interaction methods, and verbal explanations. Our study addresses this gap by conducting user studies with 18 BVI participants, comprising 15 cane users and three guide dog users. Participants interacted with a quadrupedal robot and provided both quantitative and qualitative feedback. Our study revealed several design implications, such as a preference for a learning-based controller and a rigid handle, gradual turns with asymmetric speeds, semantic communication methods, and explainability. The study also highlighted the importance of customization to support users with diverse backgrounds and preferences, along with practical concerns such as battery life, maintenance, and weather issues. These findings offer valuable insights and design implications for future research and development of robotic guide dogs., Comment: 12 pages, 4 figures, Proceedings of the 2025 ACM/IEEE International Conference on Human-Robot Interaction (HRI'25)

Segment Routing is a recent network technology that helps optimizing network throughput by providing finer control over the routing paths. Instead of routing directly from a source to a target, packets are routed via intermediate waypoints. Between consecutive waypoints, the packets are routed according to traditional shortest path routing protocols. Bottlenecks in the network can be avoided by such rerouting, preventing overloading parts of the network. The associated NP-hard computational problem is Segment Routing: Given a network on $n$ vertices, $d$ traffic demands (vertex pairs), and a (small) number $k$, the task is to find for each demand pair at most $k$ waypoints such that with shortest path routing along these waypoints, all demands are fulfilled without exceeding the capacities of the network. We investigate if special structures of real-world communication networks could be exploited algorithmically. Our results comprise NP-hardness on graphs with constant treewidth even if only one waypoint per demand is allowed. We further exclude (under standard complexity assumptions) algorithms with running time $f(d) n^{g(k)}$ for any functions $f$ and $g$. We complement these lower bounds with polynomial-time solvable special cases.

CaMn$_2$P$_2$ exemplifies the realization of a frustrated $J_1$-$J_2$-$J_3$ Heisenberg model of a corrugated honeycomb magnetic lattice. Previous studies show that below the N\'eel temperature ($T_{\rm N}$), the system forms a cycloidal $6\times 6$ $ab$-plane magnetic unit cell that conforms with various magnetic space groups. Here, we present single-crystal neutron-diffraction studies across expansive reciprocal-space volumes, confirming the cycloidal magnetic structure while uncovering further distinctive features. We find evidence for three magnetic domains, the analysis of which narrows the possible magnetic model structures. At $T_{\rm N}$, the insulator exhibits a sharp phase transition, above which the spin structure transforms into a spiral spin liquid state, evident via a continuous ring of scattering with degenerate wavevectors corresponding to a collection of short-range spiral spin configurations. These degenerate states emerge as thermal fluctuations effectively reduce the $J_3$ interaction. The integration of experimental, theoretical, and real-space simulation results reveals the intricate balance of exchange interactions ($J_1$-$J_2$-$J_3$) that stabilizes the ground-state magnetic structure and drives the emergence of a sought-after $U$(1)-symmetric spiral spin-liquid state with easy-plane anisotropy above the transition temperature.

The demand for efficient data processing motivates a shift toward in-memory computing architectures. Ferroelectric materials, particularly AlScN, show great promise for next-generation memory devices. However, their widespread application is limited due challenges such as high coercive fields, leakage currents and limited stability. Our work introduces a novel synthesis approach for ferroelectric AlScN thin films using high-power impulse magnetron sputtering (HiPIMS). Through a combinatorial study, we investigate the effect of scandium content and substrate bias on the ferroelectric properties of AlScN films deposited using metal-ion synchronized (MIS) HiPIMS. Leveraging the high ionization rates of HiPIMS and optimally timed substrate bias potentials, we enhance the adatom mobility at low temperatures. Our films exhibit a high degree of texture and crystallinity as well as low roughness at temperatures as low as 250{\deg}C. Most importantly, the films exhibit coercive fields comparable to state-of-the-art values (5 MV/cm) with significantly enhanced remanent polarization (158-172.0 {\mu}C/cm2). Notably, the remanent polarization remains stable across varying scandium concentrations. We further evaluate cycling stability and leakage current to assess suitability for memory applications. This study demonstrates HiPIMS as a scalable and CMOS compatible technique for synthesizing high-quality ferroelectric AlScN films, paving the way for their application in non-volatile memory applications.

This project explores the design and development of a transition edge sensor (TES) spectrometer for resonant soft X- ray scattering (RSXS) measurements developed in collaboration between Argonne National Laboratory (ANL) and the National Institute of Standards and Technology (NIST). Soft X-ray scattering is a powerful technique for studying the electronic and magnetic properties of materials on a microscopic level. However, the lack of high-performance soft X-ray spectrometers has limited the potential of this technique. TES spectrometers have the potential to overcome these limitations due to their high energy resolution, high efficiency, and broad energy range. This project aims to optimize the design of a TES spectrometer for RSXS measurements and more generally soft X-ray spectroscopy at the Advanced Photon Source (APS) 29-ID, leading to improved understanding of advanced materials. We will present a detailed description of the instrument design and implementation. The spectrometer consists of a large array of approximately 250 high-speed and high-resolution pixels. The pixels have saturation energies of approximately 1 keV, sub-ms pulse duration and energy resolution of approximately 1 eV. The array is read out using microwave multiplexing chips with MHz bandwidth per channel, enabling efficient data throughput. To facilitate measurement of samples in situ under ultra-high vacuum conditions at the beamline, the spectrometer is integrated with an approximately 1 m long snout.

The advancement of machine learning and the availability of large-scale reaction datasets have accelerated the development of data-driven models for computer-aided synthesis planning (CASP) in the past decade. Here, we detail the newest version of ASKCOS, an open source software suite for synthesis planning that makes available several research advances in a freely available, practical tool. Four one-step retrosynthesis models form the basis of both interactive planning and automatic planning modes. Retrosynthetic planning is complemented by other modules for feasibility assessment and pathway evaluation, including reaction condition recommendation, reaction outcome prediction, and auxiliary capabilities such as solubility prediction and quantum mechanical descriptor prediction. ASKCOS has assisted hundreds of medicinal, synthetic, and process chemists in their day-to-day tasks, complementing expert decision making. It is our belief that CASP tools like ASKCOS are an important part of modern chemistry research, and that they offer ever-increasing utility and accessibility.

Continuous gravitational waves (CWs) emission from neutron stars carries information about their internal structure and equation of state, and it can provide tests of General Relativity. We present a search for CWs from a set of 45 known pulsars in the first part of the fourth LIGO--Virgo--KAGRA observing run, known as O4a. We conducted a targeted search for each pulsar using three independent analysis methods considering the single-harmonic and the dual-harmonic emission models. We find no evidence of a CW signal in O4a data for both models and set upper limits on the signal amplitude and on the ellipticity, which quantifies the asymmetry in the neutron star mass distribution. For the single-harmonic emission model, 29 targets have the upper limit on the amplitude below the theoretical spin-down limit. The lowest upper limit on the amplitude is $6.4\!\times\!10^{-27}$ for the young energetic pulsar J0537-6910, while the lowest constraint on the ellipticity is $8.8\!\times\!10^{-9}$ for the bright nearby millisecond pulsar J0437-4715. Additionally, for a subset of 16 targets we performed a narrowband search that is more robust regarding the emission model, with no evidence of a signal. We also found no evidence of non-standard polarizations as predicted by the Brans-Dicke theory., Comment: main paper: 12 pages, 6 figures, 4 tables

Generative models hold great potential, but only if one can trust the evaluation of the data they generate. We show that many commonly used quality scores for comparing two-dimensional distributions of synthetic vs. ground-truth data give better results than they should, a phenomenon we call the "grade inflation problem." We show that the correlation score, Jaccard score, earth-mover's score, and Kullback-Leibler (relative-entropy) score all suffer grade inflation. We propose that any score that values all datapoints equally, as these do, will also exhibit grade inflation; we refer to such scores as "equipoint" scores. We introduce the concept of "equidensity" scores, and present the Eden score, to our knowledge the first example of such a score. We found that Eden avoids grade inflation and agrees better with human perception of goodness-of-fit than the equipoint scores above. We propose that any reasonable equidensity score will avoid grade inflation. We identify a connection between equidensity scores and R\'enyi entropy of negative order. We conclude that equidensity scores are likely to outperform equipoint scores for generative models, and for comparing low-dimensional distributions more generally., Comment: 10 pages, 6 figures, 1 table

Molten salts are crucial for clean energy applications, yet exploring their thermophysical properties across diverse chemical space remains challenging. We present the development of a machine learning interatomic potential (MLIP) called SuperSalt, which targets 11-cation chloride melts and captures the essential physics of molten salts with near-DFT accuracy. Using an efficient workflow that integrates systems of one, two, and 11 components, the SuperSalt potential can accurately predict thermophysical properties such as density, bulk modulus, thermal expansion, and heat capacity. Our model is validated across a broad chemical space, demonstrating excellent transferability. We further illustrate how Bayesian optimization combined with SuperSalt can accelerate the discovery of optimal salt compositions with desired properties. This work provides a foundation for future studies that allows easy extensions to more complex systems, such as those containing additional elements. SuperSalt represents a shift towards a more universal, efficient, and accurate modeling of molten salts for advanced energy applications.

The SuperNEMO experiment is searching for neutrinoless double beta decay of \textsuperscript{82}Se, with the unique combination of a tracking detector and a segmented calorimeter. This feature allows to detect the two electrons emitted in the decay and measure their individual energy and angular distribution. The SuperNEMO calorimeter consists of 712 plastic scintillator blocks readout by large PMTs. After the construction of the demonstrator calorimeter underground, we have performed its first commissioning using $\gamma$-particles from calibration sources or from the ambient radioactive background. This article presents the quality assurance tests of the SuperNEMO demonstrator calorimeter and its first time and energy calibrations, with the associated methods.

Clinical trials drive improvements in cancer treatments and outcomes. However, most adults with cancer do not participate in trials, and trials often fail to enroll enough patients to answer their scientific questions. Artificial intelligence could accelerate matching of patients to appropriate clinical trials. Here, we describe the development and evaluation of the MatchMiner-AI pipeline for clinical trial searching and ranking. MatchMiner-AI focuses on matching patients to potential trials based on core criteria describing clinical "spaces," or disease contexts, targeted by a trial. It aims to accelerate the human work of identifying potential matches, not to fully automate trial screening. The pipeline includes modules for extraction of key information from a patient's longitudinal electronic health record; rapid ranking of candidate trial-patient matches based on embeddings in vector space; and classification of whether a candidate match represents a reasonable clinical consideration. Code and synthetic data are available at https://huggingface.co/ksg-dfci/MatchMiner-AI . Model weights based on synthetic data are available at https://huggingface.co/ksg-dfci/TrialSpace and https://huggingface.co/ksg-dfci/TrialChecker . A simple cancer clinical trial search engine to demonstrate pipeline components is available at https://huggingface.co/spaces/ksg-dfci/trial_search_alpha .

For a finite group $A$ with normal subgroup $G$, a subgroup $U$ of $G$ is an $A$-prime-power-covering subgroup if $U$ meets every $A$-conjugacy-class of elements of $G$ of prime power order. It is conjectured that $|G:U|$ is bounded by some function of $|A:G|$, and this conjecture has number theoretic implications for relative Brauer groups of algebraic number fields. We prove the conjecture in the case that the action of $G$ on the set of right cosets of $U$ in $G$ is innately transitive. This includes the case where $U$ is a maximal subgroup of $G$. The proof uses a new bound on the order of a nonabelian finite simple group in terms of its number of classes of elements of prime power order, which in turn depends on the Classification of the Finite Simple Groups.

X-ray attenuation, phase, and dark-field images provide complementary information. Different experimental techniques can capture these contrast mechanisms, and the corresponding images can be retrieved using various theoretical algorithms. Our previous works developed the Multimodal Intrinsic Speckle-Tracking (MIST) algorithm, which is suitable for multimodal image retrieval from speckle-based X-ray imaging (SBXI) data. MIST is based on the X-ray Fokker-Planck equation, requiring the inversion of derivative operators that are often numerically unstable. These instabilities can be addressed by employing regularization techniques, such as Tikhonov regularization. The regularization output is highly sensitive to the choice of the Tikhonov regularization parameter, making it crucial to select this value carefully and optimally. Here, we present an automated iterative algorithm to optimize the regularization of the inverse Laplacian operator in our most recently published MIST variant, addressing the operator's instability near the Fourier-space origin. Our algorithm leverages the inherent stability of the phase solution obtained from the transport-of-intensity equation for SBXI, using it as a reliable ground truth for the more complex Fokker-Planck-based algorithms that incorporate the dark-field signal. We applied the algorithm to an SBXI dataset collected using synchrotron light of a four-rod sample. The four-rod sample's phase and dark-field images were optimally retrieved using our developed algorithm, eliminating the tedious and subjective task of selecting a suitable Tikhonov regularization parameter. The developed regularization-optimization algorithm makes MIST more user-friendly by eliminating the need for manual parameter selection. We anticipate that our optimization algorithm can also be applied to other image retrieval approaches derived from the Fokker-Planck equation.

X-ray dark-field imaging visualises scattering from sample microstructure, and has found application in medical and security contexts. While most X-ray dark-field imaging techniques rely on masks, gratings, or crystals, recent work on the Fokker--Planck model of diffusive imaging has enabled dark-field imaging in the propagation-based geometry. Images captured at multiple propagation distances or X-ray energies can be used to reconstruct dark-field from propagation-based images but have previously required multiple exposures. Here, we show single-exposure dark-field imaging by exploiting the harmonic content in a monochromatised synchrotron beam and utilising an energy-discriminating photon-counting detector to capture dual-energy propagation-based images. The method is validated by filming time-varying samples, showing the advantage of the dark-field contrast in analysing dynamic evolution. We measure and adjust for the impact of detector charge-sharing on the images. This work opens the way for low-dose and dynamic dark-field X-ray imaging without the need for a high-stability set-up and precision optics., Comment: 9 pages, 7 figures

Fact-checking has been promoted as a key method for combating political misinformation. Comparing the spread of election-related misinformation narratives along with their relevant political fact-checks, this study provides the most comprehensive assessment to date of the real-world limitations faced by political fact-checking efforts. To examine barriers to impact, this study extends recent work from laboratory and experimental settings to the wider online information ecosystem present during the 2022 U.S. midterm elections. From analyses conducted within this context, we find that fact-checks as currently developed and distributed are severely inhibited in election contexts by constraints on their i. coverage, ii. speed, and, iii. reach. Specifically, we provide evidence that fewer than half of all prominent election-related misinformation narratives were fact-checked. Within the subset of fact-checked claims, we find that the median fact-check was released a full four days after the initial appearance of a narrative. Using network analysis to estimate user partisanship and dynamics of information spread, we additionally find evidence that fact-checks make up less than 1.2\% of narrative conversations and that even when shared, fact-checks are nearly always shared within,rather than between, partisan communities. Furthermore, we provide empirical evidence which runs contrary to the assumption that misinformation moderation is politically biased against the political right. In full, through this assessment of the real-world influence of political fact-checking efforts, our findings underscore how limitations in coverage, speed, and reach necessitate further examination of the potential use of fact-checks as the primary method for combating the spread of political misinformation.

The X-Ray Imaging and Spectroscopy Mission (XRISM) is the 7th Japanese X-ray observatory, whose development and operation are in collaboration with universities and research institutes in Japan, U.S., and Europe, including JAXA, NASA, and ESA. The telemetry data downlinked from the satellite are reduced to scientific products by the pre-pipeline (PPL) and pipeline (PL) software running on standard Linux virtual machines on the JAXA and NASA sides, respectively. We ported the PPL to the JAXA "TOKI-RURI" high-performance computing (HPC) system capable of completing $\simeq 160$ PPL processes within 24 hours by utilizing the container platform of Singularity and its "--bind" option. In this paper, we briefly show the data processing in XRISM and present our porting strategy of PPL to the HPC environment in detail., Comment: 4 pages, 2 figures, proceedings of the Astronomical Data Analysis Software and Systems XXXIV

Over 2000 Gamma ray sources identified by the Large Area Telescope (LAT) on NASA's Fermi Gamma-ray Space Telescope are considered unassociated, meaning that they have no known counterparts in any other frequency regime. We have carried out an image-based search for steep spectrum radio sources, with in-band spectral index less than -1.4, within the error regions of Fermi unassociated sources using 1 to 1.4 GHz radio data from the MeerKAT Absorption Line Survey (MALS) Data Release. MALS DR1 with a median rms noise of 22 to 25 microJy and 735,649 sources is a significant advance over past image-based searches with improvements in sensitivity, resolution and bandwidth. Steep spectrum candidates were identified using a combination of in-band spectral indices from MALS and existing radio surveys. We developed an optical and infrared source classification scheme in order to distinguish between galactic pulsars and radio galaxies. In total, we identify nine pulsar candidates towards six Fermi sources that are worthy of follow-up for pulsation searches. We also report 41 steep spectrum radio galaxy candidates that may be of interest in searches for high-redshift radio galaxies. We show that MALS due to its excellent continuum sensitivity can detect 80 percent of the known pulsar population. This exhibits the promise of identifying exotic pulsar candidates with future image-based surveys with the Square Kilometre and its precursors., Comment: 13 pages, 5 figures, Accepted for publication in ApJ

This study investigates students' perceptions of Generative Artificial Intelligence (AI), with a focus on Higher Education institutions in Northern Ireland and India. We collect quantitative Likert ratings and qualitative comments from 1,211 students on their awareness and perceptions of AI and investigate variations in attitudes toward AI across institutions and subject areas, as well as interactions between these variables with demographic variables (focusing on gender). We find that: (a) while perceptions varied across institutions, responses for Computer Sciences students were similar; and (b) after controlling for institution and subject area, we observed no effect of gender. These results are consistent with previous studies, which find that students' perceptions are predicted by prior experience. We consider the implications of this relation and some considerations for the role of experience.

Performance of large language models (LLMs) on medical tasks has traditionally been evaluated using multiple choice question benchmarks. However, such benchmarks are highly constrained, saturated with repeated impressive performance by LLMs, and have an unclear relationship to performance in real clinical scenarios. Clinical reasoning, the process by which physicians employ critical thinking to gather and synthesize clinical data to diagnose and manage medical problems, remains an attractive benchmark for model performance. Prior LLMs have shown promise in outperforming clinicians in routine and complex diagnostic scenarios. We sought to evaluate OpenAI's o1-preview model, a model developed to increase run-time via chain of thought processes prior to generating a response. We characterize the performance of o1-preview with five experiments including differential diagnosis generation, display of diagnostic reasoning, triage differential diagnosis, probabilistic reasoning, and management reasoning, adjudicated by physician experts with validated psychometrics. Our primary outcome was comparison of the o1-preview output to identical prior experiments that have historical human controls and benchmarks of previous LLMs. Significant improvements were observed with differential diagnosis generation and quality of diagnostic and management reasoning. No improvements were observed with probabilistic reasoning or triage differential diagnosis. This study highlights o1-preview's ability to perform strongly on tasks that require complex critical thinking such as diagnosis and management while its performance on probabilistic reasoning tasks was similar to past models. New robust benchmarks and scalable evaluation of LLM capabilities compared to human physicians are needed along with trials evaluating AI in real clinical settings.

Recent observations of interplanetary scintillation (IPS) at radio frequencies have proved to be a powerful tool for probing the solar environment from the ground. But how far back does this tradition really extend? Our survey of the literature to date has revealed a long history of scintillating observations, beginning with the oral traditions of Indigenous peoples from around the globe, encompassing the works of the Ancient Greeks and Renaissance scholars, and continuing right through into modern optics, astronomy and space science. We outline here the major steps that humanity has taken along this journey, using scintillation as a tool for predicting first terrestrial, and then space weather without ever having to leave the ground., Comment: 4 pages, 1 figure. Submitted for publication in proceedings of IAU Symposium 390: A Multi-Point view of the Sun: Advances in Solar Observations and in Space Weather Understanding. From a talk given at the XXXII IAU General Assembly, Cape Town

The currently leading artificial neural network (ANN) models of the visual ventral stream -- which are derived from a combination of performance optimization and robustification methods -- have demonstrated a remarkable degree of behavioral alignment with humans on visual categorization tasks. Extending upon previous work, we show that not only can these models guide image perturbations that change the induced human category percepts, but they also can enhance human ability to accurately report the original ground truth. Furthermore, we find that the same models can also be used out-of-the-box to predict the proportion of correct human responses to individual images, providing a simple, human-aligned estimator of the relative difficulty of each image. Motivated by these observations, we propose to augment visual learning in humans in a way that improves human categorization accuracy at test time. Our learning augmentation approach consists of (i) selecting images based on their model-estimated recognition difficulty, and (ii) using image perturbations that aid recognition for novice learners. We find that combining these model-based strategies gives rise to test-time categorization accuracy gains of 33-72% relative to control subjects without these interventions, despite using the same number of training feedback trials. Surprisingly, beyond the accuracy gain, the training time for the augmented learning group was also shorter by 20-23%. We demonstrate the efficacy of our approach in a fine-grained categorization task with natural images, as well as tasks in two clinically relevant image domains -- histology and dermoscopy -- where visual learning is notoriously challenging. To the best of our knowledge, this is the first application of ANNs to increase visual learning performance in humans by enhancing category-specific features.

The population dynamics of the 229Th isomeric state is studied in a solid-state host under laser illumination. A photoquenching process is observed, where off-resonant vacuum-ultraviolet (VUV) radiation leads to relaxation of the isomeric state. The cross-section for this photoquenching process is measured and a model for the decay process, where photoexcitation of electronic states within the material bandgap opens an internal conversion decay channel, is presented and appears to reproduce the measured cross-section., Comment: 7 pages, 6 figures

In materials with one-dimensional electronic bands, electron-electron interactions can produce intriguing quantum phenomena, including spin-charge separation and charge density waves (CDW). Most of these systems, however, are non-magnetic, motivating a search for anisotropic materials where the coupling of charge and spin may affect emergent quantum states. Here, electron doping the van der Waals magnetic semiconductor CrSBr induces an electronically driven quasi-1D CDW, which survives above room temperature. Lithium intercalation also increases the magnetic ordering temperature to 200 K and changes its interlayer magnetic coupling from antiferromagnetic to ferromagnetic. The spin-polarized nature of the anisotropic bands that give rise to this CDW enforces an intrinsic coupling of charge and spin. The coexistence and interplay of ferromagnetism and charge modulation in this exfoliatable material provides a promising platform for studying tunable quantum phenomena across a range of temperatures and thicknesses., Comment: Main text: 30 pages, 4 figures, 1 table. Supplementary Information: 21 pages, 11 figures, 2 tables

Traditional robotic manipulation mostly focuses on collision-free tasks. In practice, however, many manipulation tasks (e.g., occluded object grasping) require the robot to intentionally collide with the environment to reach a desired task configuration. By enabling compliant robot motions, collisions between the robot and the environment are allowed and can thus be exploited, but more physical uncertainties are introduced. To address collision-rich problems such as occluded object grasping while handling the involved uncertainties, we propose a collision-inclusive planning framework that can transition the robot to a desired task configuration via roughly modeled collisions absorbed by Cartesian impedance control. By strategically exploiting the environmental constraints and exploring inside a manipulation funnel formed by task repetitions, our framework can effectively reduce physical and perception uncertainties. With real-world evaluations on both single-arm and dual-arm setups, we show that our framework is able to efficiently address various realistic occluded grasping problems where a feasible grasp does not initially exist., Comment: This work has been submitted to the IEEE for possible publication

pyAMPACT (Python-based Automatic Music Performance Analysis and Comparison Toolkit) links symbolic and audio music representations to facilitate score-informed estimation of performance data in audio as well as general linking of symbolic and audio music representations with a variety of annotations. pyAMPACT can read a range of symbolic formats and can output note-linked audio descriptors/performance data into MEI-formatted files. The audio analysis uses score alignment to calculate time-frequency regions of importance for each note in the symbolic representation from which to estimate a range of parameters. These include tuning-, dynamics-, and timbre-related performance descriptors, with timing-related information available from the score alignment. Beyond performance data estimation, pyAMPACT also facilitates multi-modal investigations through its infrastructure for linking symbolic representations and annotations to audio., Comment: International Society for Music Information Retrieval, Late Breaking Demo

We prove the Gap Theorem for the spectrum of topological modular forms $\mathrm{Tmf}$. This removes a longstanding circularity in the literature, thereby confirming the computation of $\pi_\ast \mathrm{tmf}$ from over two decades ago by Hopkins and Mahowald. Our approach is crucially a modern one, developing and refining many techniques in synthetic spectra., Comment: 87 pages, plus appendix containing 6 tables and 6 figures. Comments welcome!

Large language models (LLMs) gained immense popularity due to their impressive capabilities in unstructured conversations. However, they underperform compared to previous approaches in task-oriented dialogue (TOD), wherein reasoning and accessing external information are crucial. Empowering LLMs with advanced prompting strategies such as reasoning and acting (ReAct) has shown promise in solving complex tasks traditionally requiring reinforcement learning. In this work, we apply the ReAct strategy to guide LLMs performing TOD. We evaluate ReAct-based LLMs (ReAct-LLMs) both in simulation and with real users. While ReAct-LLMs seem to underperform state-of-the-art approaches in simulation, human evaluation indicates higher user satisfaction rate compared to handcrafted systems despite having a lower success rate.

Solvation free energy is an important design parameter in reaction kinetics and separation processes, making it a critical property to predict during process development. In previous research, directed message passing neural networks (D-MPNN) have successfully been used to predict solvation free energies and enthalpies in organic solvents. However, solvent mixtures provide greater flexibility for optimizing solvent interactions than monosolvents. This work aims to extend our previous models to mixtures. To handle mixtures in a permutation invariant manner we propose a pooling function; MolPool. With this pooling function, the machine learning models can learn and predict properties for an arbitrary number of molecules. The novel SolProp-mix software that applies MolPool to D-MPNN was compared to state-of-the-art architectures for predicting mixture properties and validated with our new database of COSMOtherm calculations; BinarySolv-QM. To improve predictions towards experimental accuracy, the network was then fine-tuned on experimental data in monosolvents. To demonstrate the benefit of this transfer learning methodology, experimental datasets of solvation free energies in binary (BinarySolv-Exp) and ternary (TernarySolv-Exp) solvent mixtures were compiled from data on vapor-liquid equilibria and activity coefficients. The neural network performed better than COSMOtherm calculations with an MAE of 0.25 kcal/mol and an RMSE of 0.37 kcal/mol for non-aqueous mixed solvents. Additionally, the ability to capture trends for a varying mixture composition was validated successfully. Our model's ability to accurately predict mixture properties from the combination of in silico data and pure component experimental data is promising given the scarcity of experimental data for mixtures in many fields.

The downstream use cases, benefits, and risks of AI systems depend significantly on the access afforded to the system, and to whom. However, the downstream implications of different access styles are not well understood, making it difficult for decision-makers to govern model access responsibly. Consequently, we spotlight Model Access Governance, an emerging field focused on helping organisations and governments make responsible, evidence-based access decisions. We outline the motivation for developing this field by highlighting the risks of misgoverning model access, the limitations of existing research on the topic, and the opportunity for impact. We then make four sets of recommendations, aimed at helping AI evaluation organisations, frontier AI companies, governments and international bodies build consensus around empirically-driven access governance.

For an element $x$ of a finite group $T$, the $\mathrm{Aut}(T)$-class of $x$ is the set $\{ x^\sigma\mid \sigma\in \mathrm{Aut}(T)\}$. We prove that the order $|T|$ of a finite nonabelian simple group $T$ is bounded above by a function of the parameter $m(T)$, where $m(T)$ is the maximum, over all primes $p$, of the number of $\mathrm{Aut}(T)$-classes of elements of $T$ of $p$-power order. This bound is a substantial generalisation of results of Pyber, and of H\'ethelyi and K\"ulshammer, and it has implications for relative Brauer groups of finite extensions of global fields.