Search

We present 3 years of high-contrast imaging of the PDS 70 b and c accreting protoplanets with the new extreme AO system MagAO-X as part of the MaxProtoPlanetS survey of H$\alpha$ protoplanets. In 2023 and 2024 our sharp (25-27 mas FWHM); well AO corrected (20-26% Strehl), deep (2-3.6hr) images detect compact (r~30 mas; r~3 au) circumplanetary disks (CPDs) surrounding both protoplanets. Starlight scattering off the dusty outer edges of these CPDs is the likely source of the bright compact continuum light detected within ~30 mas of both planets in our simultaneously obtained continuum 668 nm filter images. After subtraction of contaminating continuum and PSF residuals with pyKLIP ADI and SDI we obtained high-contrast ASDI H$\alpha$ images of both planets in 2022, 2023 and 2024. We find the H$\alpha$ line flux of planet b fell by (8.1$\pm$1.6)x10$^{-16}$ ergs/s/cm$^2$ a factor of 4.6 drop in flux from 2022 to 2023. In March 2024, planet b continued to be faint with just a slight 1.6x rise to an H$\alpha$ line flux of (3.64$\pm$0.87)x10$^{-16}$ ergs/s/cm$^2$. For c we measure a significant increase of (2.74$\pm$0.51)x10$^{-16}$ ergs/s/cm$^2$ from 2023 to 2024 which is a factor of 2.3x increase. So both protoplanets have recently experienced significant H$\alpha$ variability with ~1 yr sampling. In 2024, planet c is brighter than b: as c is brightening and b generally fading. We also tentatively detect one new point source "CC3" inside the inner disk (~49 mas; at PA~295 deg; 2024) with orbital motion roughly consistent with a ~5.6 au orbit., Comment: 50 pages, 20 figures, published in January 2025 Astronomical Journal. NOTE Figure D1 has correct planetary mass accretion scale compared to published AJ version

Ram pressure stripping is a well-known environmental quenching mechanism that removes gas from galaxies infalling into groups and clusters. In some extreme examples of ram pressure stripping, galaxies with extended gas tails show evidence of enhanced star formation prior to quenching. In this work we use a sample of 5277 local satellite galaxies in which a stripped tail of gas has not necessarily been observed, to quantify the strength of ram pressure-enhanced star formation and compare these results to a control sample of 8360 field galaxies. We use u-band imaging from the Ultraviolet-Near Infrared Northern Survey (UNIONS) as a star formation tracer and several metrics to quantify star formation asymmetry. We compare these results to environmental properties of the galaxy, such as their time since infall and host halo mass, to constrain the degree of ram pressure enhanced star formation as a function of environment. We find no significant differences between the satellite and the field samples. We further restrict our sample to galaxies which we most expect to be experiencing significant ram pressure but find no strong evidence of these galaxies having systematically enhanced star formation. Finally, we investigate the properties of the most asymmetric galaxies in our sample and again find no strong evidence of ram pressure-induced star formation enhancement. We conclude that any star formation enhancement must be small for infalling galaxies, suggesting that this effect is either uncommon or short-lived., Comment: 19 pages, 10 figures, accepted by ApJ

Next-token prediction with the logarithmic loss is a cornerstone of autoregressive sequence modeling, but, in practice, suffers from error amplification, where errors in the model compound and generation quality degrades as sequence length $H$ increases. From a theoretical perspective, this phenomenon should not appear in well-specified settings, and, indeed, a growing body of empirical work hypothesizes that misspecification, where the learner is not sufficiently expressive to represent the target distribution, may be the root cause. Under misspecification -- where the goal is to learn as well as the best-in-class model up to a multiplicative approximation factor $C\geq 1$ -- we confirm that $C$ indeed grows with $H$ for next-token prediction, lending theoretical support to this empirical hypothesis. We then ask whether this mode of error amplification is avoidable algorithmically, computationally, or information-theoretically, and uncover inherent computational-statistical tradeoffs. We show: (1) Information-theoretically, one can avoid error amplification and achieve $C=O(1)$. (2) Next-token prediction can be made robust so as to achieve $C=\tilde O(H)$, representing moderate error amplification, but this is an inherent barrier: any next-token prediction-style objective must suffer $C=\Omega(H)$. (3) For the natural testbed of autoregressive linear models, no computationally efficient algorithm can achieve sub-polynomial approximation factor $C=e^{(\log H)^{1-\Omega(1)}}$; however, at least for binary token spaces, one can smoothly trade compute for statistical power and improve on $C=\Omega(H)$ in sub-exponential time. Our results have consequences in the more general setting of imitation learning, where the widely-used behavior cloning algorithm generalizes next-token prediction., Comment: 75 pages

Reinforcement learning is now widely adopted as the final stage of large language model training, especially for reasoning-style tasks such as maths problems. Typically, models attempt each question many times during a single training step and attempt to learn from their successes and failures. However, we demonstrate that throughout training with two popular algorithms (PPO and VinePPO) on two widely used datasets, many questions are either solved by all attempts - meaning they are already learned - or by none - providing no meaningful training signal. To address this, we adapt a method from the reinforcement learning literature - sampling for learnability - and apply it to the reinforcement learning stage of LLM training. Our curriculum prioritises questions with high variance of success, i.e. those where the agent sometimes succeeds, but not always. Our findings demonstrate that this curriculum consistently boosts training performance across multiple algorithms and datasets, paving the way for more efficient and effective reinforcement learning in LLMs.

The Large Language Model agent workflow enables the LLM to invoke tool functions to increase the performance on specific scientific domain questions. To tackle large scale of scientific research, it requires access to computing resource and parallel computing setup. In this work, we implemented Parsl to the LangChain/LangGraph tool call setup, to bridge the gap between the LLM agent to the computing resource. Two tool call implementations were set up and tested on both local workstation and HPC environment on Polaris/ALCF. The first implementation with Parsl-enabled LangChain tool node queues the tool functions concurrently to the Parsl workers for parallel execution. The second configuration is implemented by converting the tool functions into Parsl ensemble functions, and is more suitable for large task on super computer environment. The LLM agent workflow was prompted to run molecular dynamics simulations, with different protein structure and simulation conditions. These results showed the LLM agent tools were managed and executed concurrently by Parsl on the available computing resource., Comment: 7 pages, 4 figures

Algorithms for reinforcement learning (RL) in large state spaces crucially rely on supervised learning subroutines to estimate objects such as value functions or transition probabilities. Since only the simplest supervised learning problems can be solved provably and efficiently, practical performance of an RL algorithm depends on which of these supervised learning "oracles" it assumes access to (and how they are implemented). But which oracles are better or worse? Is there a minimal oracle? In this work, we clarify the impact of the choice of supervised learning oracle on the computational complexity of RL, as quantified by the oracle strength. First, for the task of reward-free exploration in Block MDPs in the standard episodic access model -- a ubiquitous setting for RL with function approximation -- we identify two-context regression as a minimal oracle, i.e. an oracle that is both necessary and sufficient (under a mild regularity assumption). Second, we identify one-context regression as a near-minimal oracle in the stronger reset access model, establishing a provable computational benefit of resets in the process. Third, we broaden our focus to Low-Rank MDPs, where we give cryptographic evidence that the analogous oracle from the Block MDP setting is insufficient., Comment: 84 pages, 2 figures

We introduce and analyze a variation of the Bertrand game in which the revenue is shared between two players. This game models situations in which one economic agent can provide goods/services to consumers either directly or through an independent seller/contractor in return for a share of the revenue. We analyze the equilibria of this game, and show how they can predict different business outcomes as a function of the players' costs and the transferred revenue shares. Importantly, we identify game parameters for which independent sellers can simultaneously increase the original player's payoff while increasing consumer surplus. We then extend the shared-revenue Bertrand game by considering the shared revenue proportion as an action and giving the independent seller an outside option to sell elsewhere. This work constitutes a first step towards a general theory for how partnership and sharing of resources between economic agents can lead to more efficient markets and improve the outcomes of both agents as well as consumers., Comment: Submitted for review to the 26th ACM Conference on Economics & Computation (EC '25)

Finetuning a Large Language Model (LLM) is crucial for generating results towards specific objectives. This research delves into the realm of drug optimization and introduce a novel reinforcement learning algorithm to finetune a drug optimization LLM-based generative model, enhancing the original drug across target objectives, while retains the beneficial chemical properties of the original drug. This work is comprised of two primary components: (1) DrugImprover: A framework tailored for improving robustness and efficiency in drug optimization. It includes a LLM designed for drug optimization and a novel Structured Policy Optimization (SPO) algorithm, which is theoretically grounded. This algorithm offers a unique perspective for fine-tuning the LLM-based generative model by aligning the improvement of the generated molecule with the input molecule under desired objectives. (2) A dataset of 1 million compounds, each with OEDOCK docking scores on 5 human proteins associated with cancer cells and 24 binding sites from SARS-CoV-2 virus. We conduct a comprehensive evaluation of SPO and demonstrate its effectiveness in improving the original drug across target properties. Our code and dataset will be publicly available at: https://github.com/xuefeng-cs/DrugImproverGPT.

Rydberg atoms remain in the limelight due to their applications in quantum optics and information technologies. In this work, the dynamics of Rydberg atoms stored in a momentum spectrometer by an all-optical trap is studied by ionizing them in the field of a continuous wave optical dipole trap. While the addition of the optical dipole trap allows to further cool the atoms, it comes at the expense of the time of flight information which is required to retrieve photoelectron momentum distributions. Here, we report on a method that extends the standard COLTRIMS (cold target recoil ion momentum spectroscopy) technique, including continuous wave lasers in pump-probe schemes, by utilizing coincidence measurements. In particular, the photoionization of atomic $^6$Li initially in a spin-polarized $2^{2}P_{3/2}$ state is explored. Multi-photon excitation from a tunable-mode femtosecond pulse is exploited to produce Rydberg atoms, which can then be ionized by the dipole field. The resulting ionization rate becomes explicitly time-dependent, and analyzing its structure unlocks the real-time atomic dynamics on nanosecond time scales.

Achieving efficient task parallelism on many-core architectures is an important challenge. The widely used GNU OpenMP implementation of the popular OpenMP parallel programming model incurs high overhead for fine-grained, short-running tasks due to time spent on runtime synchronization. In this work, we introduce and analyze three key advances that collectively achieve significant performance gains. First, we introduce XQueue, a lock-less concurrent queue implementation to replace GNU's priority task queue and remove the global task lock. Second, we develop a scalable, efficient, and hybrid lock-free/lock-less distributed tree barrier to address the high hardware synchronization overhead from GNU's centralized barrier. Third, we develop two lock-less and NUMA-aware load balancing strategies. We evaluate our implementation using Barcelona OpenMP Task Suite (BOTS) benchmarks. Results from the first and second advances demonstrate up to 1522.8$\times$ performance improvement compared to the original GNU OpenMP. Further improvements from lock-less load balancing show up to 4$\times$ improvement compared to GNU OpenMP using XQueue. Through a rich set of profiling and instrumentation tools, we are able to investigate the runtime behavior of GNU OpenMP and improve its performance on fine-grained tasks by many orders of magnitude., Comment: 13 pages, 11 figures, preprint, accepted by IPDPS2025

In this paper we develop a pseudo Markov-chain model to understand time-elapsed flows, over multiple intervals, from time and space aggregated collective inter-location trip data, given as a time-series. Building on the model, we develop measures of mobility that parallel those known for individual mobility data, such as the radius of gyration. We apply these measures to the NetMob 2024 Data Challenge data, and obtain interesting results that are consistent with published statistics and commuting patterns in cities. Besides building a new framework, we foresee applications of this approach to an improved understanding of human mobility in the context of environmental changes and sustainable development., Comment: 27 pages, 11 figures

In this work we investigate the chemical and kinetic nonequilibrium dynamics of the Higgs boson during the primordial Universe QGP (quark-gluon plasma) epoch $130\mathrm{\,GeV}>T>10\mathrm{\,GeV}$. We show that the Higgs bosons is always out of chemical abundance equilibrium with a fugacity $\Upsilon_h = 0.69$ due to virtual decay channels. Additionally, Higgs momentum distribution is found to be ``cold'' for $T<40$\,GeV, since the scattering rate drops below the production rate., Comment: 9 large EPJ pages, 5 figures, contribution to special topics volume on Particles and Plasma 2024

Improving efficiency of electrical machines requires fundamental knowledge on the mechanisms behind magnetic and eddy current losses of the magnetic core materials, with Fe-Si alloy as a prototype. These losses are intrinsically influenced by the microstructure of the materials. This necessitates physics-based, microstructure-informed multiscale simulations. In the present paper, we utilised micromagnetic simulations and computational homogenization methods to calculate the effective hysteresis and effective conductivities of Fe-Si electrical steels. To demonstrate the methodology, binder-jet printed electrical steel material samples with different microstructure were investigated. The microstructure samples were digitized based on both the descriptor-based synthetic reconstruction and SEM-image-based digitization. More samples were generated with varying microstructure features such as grain size and grain boundary phases. The micromagnetic simulations were then performed to investigate the magnetic hysteresis and hysteresis loss. The eddy current loss was also evaluated by using the effective conductivity through computational homogenization. By performing parameter research on a series of synthetic microstructures, effects of average grain size and grain boundary (GB) phase thickness on the hysteresis loss and eddy current loss were unveiled. An average grain size around 120 \si{\micro m} has the lowest hysteresis loss, although the eddy current loss increases with the grain size. Increasing GB-phase thickness helps reduce both losses. Results indicate the potential to decrease loss of magnetic core materials by microstructure optimization.

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

Using a kinetic equation approach and Density Functional Theory, we model the nonequilibrium quasiparticle and phonon dynamics of a thin superconducting film under optical irradiation ab initio. We extend this model to develop a theory for the detection of single photons in superconducting nanowires. In doing so, we create a framework for exploring new superconducting materials for enhanced device performance beyond the state-of-the-art. Though we focus in this study on superconducting nanowire single-photon detectors, these methods are general, and they can be extended to model other superconducting devices, including transition-edge sensors, microwave resonators, and superconducting qubits. Our methods effectively integrate ab initio materials modeling with models of nonequilibrium superconductivity to perform practical modeling of superconducting devices, providing a comprehensive approach that connects fundamental theory with device-level applications., Comment: 9 pages, 4 figures

This paper describes Meta's ACH system for mutation-guided LLM-based test generation. ACH generates relatively few mutants (aka simulated faults), compared to traditional mutation testing. Instead, it focuses on generating currently undetected faults that are specific to an issue of concern. From these currently uncaught faults, ACH generates tests that can catch them, thereby `killing' the mutants and consequently hardening the platform against regressions. We use privacy concerns to illustrate our approach, but ACH can harden code against {\em any} type of regression. In total, ACH was applied to 10,795 Android Kotlin classes in 7 software platforms deployed by Meta, from which it generated 9,095 mutants and 571 privacy-hardening test cases. ACH also deploys an LLM-based equivalent mutant detection agent that achieves a precision of 0.79 and a recall of 0.47 (rising to 0.95 and 0.96 with simple pre-processing). ACH was used by Messenger and WhatsApp test-a-thons where engineers accepted 73% of its tests, judging 36% to privacy relevant. We conclude that ACH hardens code against specific concerns and that, even when its tests do not directly tackle the specific concern, engineers find them useful for their other benefits., Comment: Submitted to FSE 2025 Industry Track

The HALHF collaboration has discussed a new baseline for the project, taking into account comments from the accelerator community on various aspects of the original design. In particular, these concerned the practicality of the dual-purpose linac to accelerate both colliding positron bunches and the drive beams required for the plasma linac. In addition, many other aspects of the project were also considered; the discussion and conclusions are documented in this paper. Finally, a new baseline is outlined that has been optimised and addresses several weaknesses in the original design, has higher luminosity, reduced centre-of-mass energy boost and additional features such as positron polarization as well as electron polarization. Although HALHF has become longer and more expensive, it remains significantly smaller and cheaper than other mature Higgs factory designs currently under discussion., Comment: 32 pages, 13 figures, 2 tables

We present MOFA, an open-source generative AI (GenAI) plus simulation workflow for high-throughput generation of metal-organic frameworks (MOFs) on large-scale high-performance computing (HPC) systems. MOFA addresses key challenges in integrating GPU-accelerated computing for GPU-intensive GenAI tasks, including distributed training and inference, alongside CPU- and GPU-optimized tasks for screening and filtering AI-generated MOFs using molecular dynamics, density functional theory, and Monte Carlo simulations. These heterogeneous tasks are unified within an online learning framework that optimizes the utilization of available CPU and GPU resources across HPC systems. Performance metrics from a 450-node (14,400 AMD Zen 3 CPUs + 1800 NVIDIA A100 GPUs) supercomputer run demonstrate that MOFA achieves high-throughput generation of novel MOF structures, with CO$_2$ adsorption capacities ranking among the top 10 in the hypothetical MOF (hMOF) dataset. Furthermore, the production of high-quality MOFs exhibits a linear relationship with the number of nodes utilized. The modular architecture of MOFA will facilitate its integration into other scientific applications that dynamically combine GenAI with large-scale simulations., Comment: 13 pages, 10 figures

We present an interactive visualization of the Cell Map for AI Talent Knowledge Graph (CM4AI TKG), a detailed semantic space comprising approximately 28,000 experts and 1,000 datasets focused on the biomedical field. Our tool leverages transformer-based embeddings, WebGL visualization techniques, and generative AI, specifically Large Language Models (LLMs), to provide a responsive and user-friendly interface. This visualization supports the exploration of around 29,000 nodes, assisting users in identifying potential collaborators and dataset users within the health and biomedical research fields. Our solution transcends the limitations of conventional graph visualization tools like Gephi, particularly in handling large-scale interactive graphs. We utilize GPT-4o to furnish detailed justifications for recommended collaborators and dataset users, promoting informed decision-making. Key functionalities include responsive search and exploration, as well as GenAI-driven recommendations, all contributing to a nuanced representation of the convergence between biomedical and AI research landscapes. In addition to benefiting the Bridge2AI and CM4AI communities, this adaptable visualization framework can be extended to other biomedical knowledge graphs, fostering advancements in medical AI and healthcare innovation through improved user interaction and data exploration. The demonstration is available at: https://jiawei-alpha.vercel.app/., Comment: Accepted by GenAI for Health Workshop @ NeurIPS 2024, Vancouver

High-quality biomedical datasets are essential for medical research and disease treatment innovation. The NIH-funded Bridge2AI project strives to facilitate such innovations by uniting top-tier, diverse teams to curate datasets designed for AI-driven biomedical research. We examined 1,699 dataset papers from the Nucleic Acids Research (NAR) database issues and the Bridge2AI Talent Knowledge Graph. By treating each paper's authors as a team, we explored the relationship between team attributes (team power and fairness) and dataset paper quality, measured by scientific impact (Relative Citation Ratio percentile) and clinical translation power (APT, likelihood of citation by clinical trials and guidelines). Utilizing the SHAP explainable AI framework, we identified correlations between team attributes and the success of dataset papers in both citation impact and clinical translation. Key findings reveal that (1) PI (Principal Investigator) leadership and team academic prowess are strong predictors of dataset success; (2) team size and career age are positively correlated with scientific impact but show inverse patterns for clinical translation; and (3) higher female representation correlates with greater dataset success. Although our results are correlational, they offer valuable insights into forming high-performing data generation teams. Future research should incorporate causal frameworks to deepen understanding of these relationships., Comment: Accepted by JCDL 2024

Realizing a shared responsibility between providers and consumers is critical to manage the sustainability of HPC. However, while cost may motivate efficiency improvements by infrastructure operators, broader progress is impeded by a lack of user incentives. We conduct a survey of HPC users that reveals fewer than 30 percent are aware of their energy consumption, and that energy efficiency is among users' lowest priority concerns. One explanation is that existing pricing models may encourage users to prioritize performance over energy efficiency. We propose two transparent multi-resource pricing schemes, Energy- and Carbon-Based Accounting, that seek to change this paradigm by incentivizing more efficient user behavior. These two schemes charge for computations based on their energy consumption or carbon footprint, respectively, rewarding users who leverage efficient hardware and software. We evaluate these two pricing schemes via simulation, in a prototype, and a user study.

The complexity of mathematical models describing respiratory mechanics has grown in recent years, however, parameter identifiability of such models has only been studied in the last decade in the context of observable data. This study investigates parameter identifiability of a nonlinear respiratory mechanics model tuned to the physiology of an extremely preterm infant, using global Morris screening, local deterministic sensitivity analysis, and singular value decomposition-based subset selection. The model predicts airflow and dynamic pulmonary volumes and pressures under varying levels of continuous positive airway pressure, and a range of parameters characterizing both surfactant-treated and surfactant-deficient lung. Sensitivity analyses indicated eleven parameters influence model outputs over the range of continuous positive airway pressure and lung health scenarios. The model was adapted to data from a spontaneously breathing 1 kg infant using gradient-based optimization to estimate the parameter subset characterizing the patient's state of health.

Radar is a low-cost and ubiquitous automotive sensor, but is limited by array resolution and sensitivity when performing direction of arrival analysis. Synthetic Aperture Radar (SAR) is a class of techniques to improve azimuth resolution and sensitivity for radar. Interferometric SAR (InSAR) can be used to extract elevation from the variations in phase measurements in SAR images. Utilizing InSAR we show that a typical, low-resolution radar array mounted on a vehicle can be used to accurately localize detections in 3D space for both urban and agricultural environments. We generate point clouds in each environment by combining InSAR with a signal processing scheme tailored to automotive driving. This low-compute approach allows radar to be used as a primary sensor to map fine details in complex driving environments, and be used to make autonomous perception decisions., Comment: 9 pages, 6 figures

The stellar age and mass of galaxies have been suggested as the primary determinants for the dynamical state of galaxies, with environment seemingly playing no or only a very minor role. We use a sample of 77 galaxies at intermediate redshift (z~0.3) in the Middle-Ages Galaxies Properties with Integral field spectroscopy (MAGPI) Survey to study the subtle impact of environment on galaxy dynamics. We use a combination of statistical techniques (simple and partial correlations and principal component analysis) to isolate the contribution of environment on galaxy dynamics, while explicitly accounting for known factors such as stellar age, star formation histories and stellar masses. We consider these dynamical parameters: high-order kinematics of the line-of-sight velocity distribution (parametrised by the Gauss-Hermite coefficients $h_3$ and $h_4$), kinematic asymmetries $V_{\rm asym}$ derived using kinemetry and the observational spin parameter proxy $\lambda_{R_e}$. Of these, the mean $h_4$ is the only parameter found to have a significant correlation with environment as parametrised by group dynamical mass. This correlation exists even after accounting for age and stellar mass trends. Finally, we confirm that variations in the spin parameter $\lambda_{R_e}$ are most strongly (anti-)correlated with age as seen in local studies, and show that this dependence is well-established by z~0.3., Comment: 17 pages, 13 figures, 3 tables, accepted for publication in PASA

In this introductory chapter, we lay the groundwork for the rest of the book by providing a more detailed picture of the expected purpose, shape, and architecture of future grid systems. We structure the chapter in terms of six questions that we believe are central to this discussion: Why do we need computational grids? What types of applications will grids be used for? Who will use grids? How will grids be used? What is involved in building a grid? And, what problems must be solved to make grids commonplace? We provide an overview of each of these issues here, referring to subsequent chapters for more detailed discussion., Comment: Chapter 2 of the book, "The Grid: Blueprint for a New Computing Infrastructure", Elsevier, 1998

The ideas of aleatoric and epistemic uncertainty are widely used to reason about the probabilistic predictions of machine-learning models. We identify incoherence in existing discussions of these ideas and suggest this stems from the aleatoric-epistemic view being insufficiently expressive to capture all of the distinct quantities that researchers are interested in. To explain and address this we derive a simple delineation of different model-based uncertainties and the data-generating processes associated with training and evaluation. Using this in place of the aleatoric-epistemic view could produce clearer discourse as the field moves forward., Comment: Presented at the Workshop on Bayesian Decision-Making and Uncertainty (NeurIPS 2024)

The Arcus Probe mission concept provides high-resolution soft X-ray and UV spectroscopy to reveal feedback-driven structure and evolution throughout the universe with an agile response capability ideal for probing the physics of time-dependent phenomena. The X-ray Spectrograph (XRS) utilizes two nearly identical CCD focal planes to detect and record X-ray photons from the dispersed spectra and zero-order of the critical angle transmission gratings. In this paper we describe the Arcus focal plane instrument and the CCDs, including laboratory performance results, which meet observatory requirements., Comment: 11 pages, 7 figures; to be published in JATIS

We report the detection of the [O III] auroral line in 42 galaxies within the redshift range of $3 < z < 10$. These galaxies were selected from publicly available JWST data releases, including the JADES and PRIMALsurveys, and observed using both the low-resolution PRISM/CLEAR configuration and medium-resolution gratings. The measured electron temperatures in the high-ionization regions of these galaxies range from $T_e$([O III]) = 12,000 to 24,000 K, consistent with temperatures observed in local metal-poor galaxies and previous JWST studies. In 10 galaxies, we also detect the [O II] auroral line, allowing us to determine electron temperatures in the low-ionization regions, which range between $T_e$([O II]) = 10,830 and 20,000 K. The direct-$T_e$-based metallicities of our sample span from 12 + log(O/H) = 7.2 to 8.4, indicating these high-redshift galaxies are relatively metal-poor. By combining our sample with 25 galaxies from the literature, we expand the dataset to a total of 67 galaxies within $3 < z < 10$, effectively more than doubling the previous sample size for direct-$T_e$ based metallicity studies. This larger dataset allow us to derive empirical metallicity calibration relations based exclusively on high-redshift galaxies, using six key line ratios: R3, R2, R23, Ne3O2, O32, and O3N2. Notably, we derive a novel metallicity calibration relation for the first time using high-redshift $T_e$-based metallicities: $\hat{R}$ = 0.18log $R2$ + 0.98log $R3$. This new calibration significantly reduces the scatter in high-redshift galaxies compared to the $\hat{R}$ relation previously calibrated for low-redshift galaxies., Comment: 23 pages, 10 figures, submitted to ApJ

We present improvements to the pointing accuracy of the South Pole Telescope (SPT) using machine learning. The ability of the SPT to point accurately at the sky is limited by its structural imperfections, which are impacted by the extreme weather at the South Pole. Pointing accuracy is particularly important during SPT participation in observing campaigns with the Event Horizon Telescope (EHT), which requires stricter accuracy than typical observations with the SPT. We compile a training dataset of historical observations of astronomical sources made with the SPT-3G and EHT receivers on the SPT. We train two XGBoost models to learn a mapping from current weather conditions to two telescope drive control arguments -- one which corrects for errors in azimuth and the other for errors in elevation. Our trained models achieve root mean squared errors on withheld test data of $2.14''$ in cross-elevation and $3.57''$ in elevation, well below our goal of $5''$ along each axis. We deploy our models on the telescope control system and perform further in situ test observations during the EHT observing campaign in 2024 April. Our models result in significantly improved pointing accuracy: for sources within the range of input variables where the models are best trained, average combined pointing error improved 33%, from $15.9''$ to $10.6''$. These improvements, while significant, fall shy of our ultimate goal, but they serve as a proof of concept for the development of future models. Planned upgrades to the EHT receiver on the SPT will necessitate even stricter pointing accuracy which will be achievable with our methods., Comment: 23 pages, 9 figures, to be submitted to Journal of Astronomical Instrumentation (JAI)

We introduce FarExStance, a new dataset for explainable stance detection in Farsi. Each instance in this dataset contains a claim, the stance of an article or social media post towards that claim, and an extractive explanation which provides evidence for the stance label. We compare the performance of a fine-tuned multilingual RoBERTa model to several large language models in zero-shot, few-shot, and parameter-efficient fine-tuned settings on our new dataset. On stance detection, the most accurate models are the fine-tuned RoBERTa model, the LLM Aya-23-8B which has been fine-tuned using parameter-efficient fine-tuning, and few-shot Claude-3.5-Sonnet. Regarding the quality of the explanations, our automatic evaluation metrics indicate that few-shot GPT-4o generates the most coherent explanations, while our human evaluation reveals that the best Overall Explanation Score (OES) belongs to few-shot Claude-3.5-Sonnet. The fine-tuned Aya-32-8B model produced explanations most closely aligned with the reference explanations., Comment: Accepted in COLING 2025

The quantum chromodynamics (QCD) axion arises as the pseudo-Goldstone mode of a spontaneously broken abelian Peccei-Quinn (PQ) symmetry. If the scale of PQ symmetry breaking occurs below the inflationary reheat temperature and the domain wall number is unity, then there is a unique axion mass that gives the observed dark matter (DM) abundance. Computing this mass has been the subject of intensive numerical simulations for decades since the mass prediction informs laboratory experiments. Axion strings develop below the PQ symmetry-breaking temperature, and as the string network evolves it emits axions that go on to become the DM. A key ingredient in the axion mass prediction is the spectral index of axion radiation emitted by the axion strings. We compute this index in this work using the most precise and accurate large-scale simulations to date of the axion-string network leveraging adaptive mesh refinement to achieve the precision that would otherwise require a static lattice with 262,144$^3$ lattice sites. We find a scale-invariant axion radiation spectrum to within 1% precision. Accounting for axion production from strings prior to the QCD phase transition leads us to predict that the axion mass should be approximately $m_a\in(45,65)$ $\mu \mathrm{eV}$. However, we provide preliminary evidence that axions are produced in greater quantities from the string-domain-wall network collapse during the QCD phase transition, potentially increasing the mass prediction to as much as 300 $\mu$eV., Comment: 7 + 25 pages, 3 + 18 figures, 0 + 4 tables, supplementary animations at https://tinyurl.com/AxionStringsAMR, video abstract at https://tinyurl.com/rxfnbt9c

GEANT4 is a particle physics simulation tool used to develop and optimize radiation detectors. While C++ based examples exist, Python's growing popularity necessitates the development of a more accessible Python bindings interface. This work demonstrates the use of cppyy, the automated C++-Python binding package, to provide an accessible interface for developing applications with GEANT4. Coupled with newly developed Python visualization tools and a Python-specific helper layer, we demonstrate the suitability of the interface for use in constructing simplistic simulation scenarios showing some initial benchmarking studies when compared to a pure C++ equivalent simulation example., Comment: 14 pages, 4 figures

Self-improvement is a mechanism in Large Language Model (LLM) pre-training, post-training and test-time inference. We explore a framework where the model verifies its own outputs, filters or reweights data based on this verification, and distills the filtered data. Despite several empirical successes, a fundamental understanding is still lacking. In this work, we initiate a comprehensive, modular and controlled study on LLM self-improvement. We provide a mathematical formulation for self-improvement, which is largely governed by a quantity which we formalize as the generation-verification gap. Through experiments with various model families and tasks, we discover a scaling phenomenon of self-improvement -- a variant of the generation-verification gap scales monotonically with the model pre-training flops. We also examine when self-improvement is possible, an iterative self-improvement procedure, and ways to improve its performance. Our findings not only advance understanding of LLM self-improvement with practical implications, but also open numerous avenues for future research into its capabilities and boundaries., Comment: 41 pages, 19 figures

Recent work in language modeling has raised the possibility of self-improvement, where a language models evaluates and refines its own generations to achieve higher performance without external feedback. It is impossible for this self-improvement to create information that is not already in the model, so why should we expect that this will lead to improved capabilities? We offer a new perspective on the capabilities of self-improvement through a lens we refer to as sharpening. Motivated by the observation that language models are often better at verifying response quality than they are at generating correct responses, we formalize self-improvement as using the model itself as a verifier during post-training in order to ``sharpen'' the model to one placing large mass on high-quality sequences, thereby amortizing the expensive inference-time computation of generating good sequences. We begin by introducing a new statistical framework for sharpening in which the learner aims to sharpen a pre-trained base policy via sample access, and establish fundamental limits. Then we analyze two natural families of self-improvement algorithms based on SFT and RLHF. We find that (i) the SFT-based approach is minimax optimal whenever the initial model has sufficient coverage, but (ii) the RLHF-based approach can improve over SFT-based self-improvement by leveraging online exploration, bypassing the need for coverage. Finally, we empirically validate the sharpening mechanism via inference-time and amortization experiments. We view these findings as a starting point toward a foundational understanding that can guide the design and evaluation of self-improvement algorithms.

Machine unlearning is gaining increasing attention as a way to remove adversarial data poisoning attacks from already trained models and to comply with privacy and AI regulations. The objective is to unlearn the effect of undesired data from a trained model while maintaining performance on the remaining data. This paper introduces HyperForget, a novel machine unlearning framework that leverages hypernetworks - neural networks that generate parameters for other networks - to dynamically sample models that lack knowledge of targeted data while preserving essential capabilities. Leveraging diffusion models, we implement two Diffusion HyperForget Networks and used them to sample unlearned models in Proof-of-Concept experiments. The unlearned models obtained zero accuracy on the forget set, while preserving good accuracy on the retain sets, highlighting the potential of HyperForget for dynamic targeted data removal and a promising direction for developing adaptive machine unlearning algorithms., Comment: AdvML-Frontiers'24: The 3rd Workshop on New Frontiers in Adversarial Machine Learning@NeurIPS'24, Vancouver, CA

We investigate the internal and external mechanisms that regulate and quench star formation (SF) in galaxies at $z \sim 0.3$ using MAGPI observations and the EAGLE, Magneticum, and IllustrisTNG cosmological simulations. Using SimSpin to generate mock observations of simulated galaxies, we match detection/resolution limits in star formation rates and stellar mass, along with MAGPI observational details including the average point spread function and pixel scale. While we find a good agreement in the slope of the global star-forming main sequence (SFMS) between MAGPI observations and all three simulations, the slope of the resolved SFMS does not agree within 1 $-$ 2$\sigma$. Furthermore, in radial SF trends, good agreement between observations and simulations exists only for galaxies far below the SFMS, where we capture evidence for inside-out quenching. The simulations overall agree with each other between $\sim1.5-4 \ R_{\rm e}$ but show varying central suppression within $R \sim 1.5 \ R_{\rm e}$ for galaxies on and below the SFMS, attributable to different AGN feedback prescriptions. All three simulations show similar dependencies of SF radial trends with environment. Central galaxies are subject to both internal and external mechanisms, showing increased SF suppression in the centre with increasing halo mass, indicating AGN feedback. Satellite galaxies display increasing suppression in the outskirts as halo mass increases, indicative of environmental processes. These results demonstrate the power of spatially resolved studies of galaxies; while global properties align, radial profiles reveal discrepancies between observations and simulations and their underlying physics., Comment: 20 pages, 10 figures, submitted to MNRAS

Algorithmic agents are used in a variety of competitive decision settings, notably in making pricing decisions in contexts that range from online retail to residential home rentals. Business managers, algorithm designers, legal scholars, and regulators alike are all starting to consider the ramifications of "algorithmic collusion." We study the emergent behavior of multi-armed bandit machine learning algorithms used in situations where agents are competing, but they have no information about the strategic interaction they are engaged in. Using a general-form repeated Prisoner's Dilemma game, agents engage in online learning with no prior model of game structure and no knowledge of competitors' states or actions (e.g., no observation of competing prices). We show that these context-free bandits, with no knowledge of opponents' choices or outcomes, still will consistently learn collusive behavior - what we call "naive collusion." We primarily study this system through an analytical model and examine perturbations to the model through simulations. Our findings have several notable implications for regulators. First, calls to limit algorithms from conditioning on competitors' prices are insufficient to prevent algorithmic collusion. This is a direct result of collusion arising even in the naive setting. Second, symmetry in algorithms can increase collusion potential. This highlights a new, simple mechanism for "hub-and-spoke" algorithmic collusion. A central distributor need not imbue its algorithm with supra-competitive tendencies for apparent collusion to arise; it can simply arise by using certain (common) machine learning algorithms. Finally, we highlight that collusive outcomes depend starkly on the specific algorithm being used, and we highlight market and algorithmic conditions under which it will be unknown a priori whether collusion occurs., Comment: To be published in proceedings of International Conference on Information Systems 2024

Here, we present the outcomes from the second Large Language Model (LLM) Hackathon for Applications in Materials Science and Chemistry, which engaged participants across global hybrid locations, resulting in 34 team submissions. The submissions spanned seven key application areas and demonstrated the diverse utility of LLMs for applications in (1) molecular and material property prediction; (2) molecular and material design; (3) automation and novel interfaces; (4) scientific communication and education; (5) research data management and automation; (6) hypothesis generation and evaluation; and (7) knowledge extraction and reasoning from scientific literature. Each team submission is presented in a summary table with links to the code and as brief papers in the appendix. Beyond team results, we discuss the hackathon event and its hybrid format, which included physical hubs in Toronto, Montreal, San Francisco, Berlin, Lausanne, and Tokyo, alongside a global online hub to enable local and virtual collaboration. Overall, the event highlighted significant improvements in LLM capabilities since the previous year's hackathon, suggesting continued expansion of LLMs for applications in materials science and chemistry research. These outcomes demonstrate the dual utility of LLMs as both multipurpose models for diverse machine learning tasks and platforms for rapid prototyping custom applications in scientific research., Comment: Updating author information, the submission remains largely unchanged. 98 pages total

Observations of the Milky Way and external galaxies support the idea that large-scale magnetic fields are concentrated in galactic disks, with halo magnetic fields at least an order of magnitude weaker. However, very little is known about the transition between the two. We present the discovery of linearly polarized radio emission at the interface between interacting shells of gas within a well-known grouping of high-velocity clouds (HVCs), the Anti-Center Shell. Faraday rotation of diffuse emission and of background extragalactic compact sources demonstrates an enhancement of the field at the interface. This is the clearest observed example of an HVC altering the large-scale magnetic field at the disk-halo interface and is the first image of magnetic field effects in an HVC. These results demonstrate the possibility of future three-dimensional reconstruction of the Galactic magnetic field and showcase the versatility of the Synthesis Telescope at the Dominion Radio Astrophysical Observatory as one of the few existing telescopes which can exploit this new method of probing Galactic magnetism., Comment: Accepted for publication in the Astrophysical Journal Nov. 13, 2024

In this study, we explore the use of Co-55 as a radioisotope for multiplexed PET (mPET) by utilizing its emission of a prompt gamma-ray in cascade with a positron during decay. We leverage the prompt-gamma signal to generate triple coincidences for a Co-55-labeled tracer, allowing us to distinguish it from a tracer labeled with a pure positron emitter, such as F-18. By employing triple versus double coincidence detection and signal processing methodology, we successfully separate the Co-55 signal from that of F-18. Phantom studies were conducted to establish the correlation between Co-55 double and triple coincidence counts and Co-55 activity. Additionally, we demonstrate the potential for quantifying hot spots within a warm background produced by both Co-55 and F-18 signals in a single PET scan. Finally, we showcase the ability to simultaneously image two tracers in vivo in a single PET session with mouse models of cancer., Comment: 5 pages, 4 figures, submitted IEEE ISBI 2025; v2: clarification on unstable Co-55-labeled antibody

Governments are increasingly employing funding for open source software (OSS) development as a policy lever to support the security of software supply chains, digital sovereignty, economic growth, and national competitiveness in science and innovation, among others. However, the impacts of public funding on OSS development remain poorly understood, with a lack of consensus on how to meaningfully measure them. This gap hampers assessments of the return on public investment and impedes the optimisation of public-interest funding strategies. We address this gap with a toolkit of methodological considerations that may inform such measurements, drawing on prior work on OSS valuations and community health metrics by the Community Health Analytics Open Source Software (CHAOSS) project as well as our first-hand learnings as practitioners tasked with evaluating funding programmes by the Next Generation Internet initiative and the Sovereign Tech Agency. We discuss salient considerations, including the importance of accounting for funding objectives, project life stage and social structure, and regional and organisational cost factors. Next, we present a taxonomy of potential social, economic, and technological impacts that can be both positive and negative, direct and indirect, internal (i.e. within a project) and external (i.e. among a project's ecosystem of dependents and users), and manifest over various time horizons. Furthermore, we discuss the merits and limitations of qualitative, quantitative, and mixed-methods approaches, as well as options for and hazards of estimating multiplier effects. With this toolkit, we contribute to the multi-stakeholder conversation about the value and impacts of funding on OSS developers and society at large., Comment: 23 pages, 3 tables

From CMB polarization data alone we reconstruct the CMB lensing power spectrum, comparable in overall constraining power to previous temperature-based reconstructions, and an unlensed E-mode power spectrum. The observations, taken in 2019 and 2020 with the South Pole Telescope (SPT) and the SPT-3G camera, cover 1500 deg$^2$ at 95, 150, and 220 GHz with arcminute resolution and roughly 4.9$\mu$K-arcmin coadded noise in polarization. The power spectrum estimates, together with systematic parameter estimates and a joint covariance matrix, follow from a Bayesian analysis using the Marginal Unbiased Score Expansion (MUSE) method. The E-mode spectrum at $\ell>2000$ and lensing spectrum at $L>350$ are the most precise to date. Assuming the $\Lambda$CDM model, and using only these SPT data and priors on $\tau$ and absolute calibration from Planck, we find $H_0=66.81\pm0.81$ km/s/Mpc, comparable in precision to the Planck determination and in 5.4$\sigma$ tension with the most precise $H_0$ inference derived via the distance ladder. We also find $S_8=0.850\pm0.017$, providing further independent evidence of a slight tension with low-redshift structure probes. The $\Lambda$CDM model provides a good simultaneous fit to the combined Planck, ACT, and SPT data, and thus passes a powerful test. Combining these CMB datasets with BAO observations, we find that the effective number of neutrino species, spatial curvature, and primordial helium fraction are consistent with standard model values, and that the 95% confidence upper limit on the neutrino mass sum is 0.075 eV. The SPT data are consistent with the somewhat weak preference for excess lensing power seen in Planck and ACT data relative to predictions of the $\Lambda$CDM model. We also detect at greater than 3$\sigma$ the influence of non-linear evolution in the CMB lensing power spectrum and discuss it in the context of the $S_8$ tension.(abridged), Comment: 28 pages, 21 figures + appendices

Many popular open source projects are owned and driven by vendors, and in today's difficult economic climate, those vendors are under increasing pressure from investors to deliver a strong return on their investments. One response to this pressure has been the relicensing of popular open source projects to more restrictive licenses in the hopes of generating more revenue, disrupting the idea of open source as a digital commons. In some cases, relicensing has resulted in a hard fork of the original project. These relicensing events and resulting forks can be disruptive to the organizations and individuals using these open source projects. This research compares and contrasts organizational affiliation data from three case studies based on license changes that resulted in forks: Elasticsearch / OpenSearch, Redis / Valkey, and Terraform / OpenTofu. The research indicates that the forks resulting from these relicensing events have more organizational diversity than the original projects, especially when the forks are created under a neutral foundation, like the Linux Foundation, rather than by a single company., Comment: Presented at OpenForum Academy (OFA) Symposium in Boston, Massachusetts, November 13 & 14, 2024