Search

In this brief, we detail three major findings from our year of interviews and observations: (1) We describe how the external contexts of schooling shifted throughout the pandemic and how many of these outside pressures persist today; (2) We look internally at how district leaders make sense of external pressures and weigh their internal values to set budget priorities, arrive at defining problems, then remedy these concerns. Districts displayed three ways of emerging from the pandemic after the return to in-person learning; and (3) We compare the different solutions or innovations that districts pursued to reengage students and support their well-being. We compare large and small structural innovations as well as pedagogical and organizational innovations.

Public schools buckled under the shock that arrived with the global pandemic, most closing their doors in March 2020. Still fresh in our memories, teachers attempted online instruction, viewing their students each day as small squares on computer screens. We know all too well that learning curves of students flattened or fell. Many kids and teachers experienced death in their families, along with emotional angst that's still reported by local educators. Yet as the COVID-19 virus receded, our research team began visiting a handful of California school districts in early 2021. We asked district leaders and school principals about how they were recovering from this unprecedented jolt, along with the challenges and joys of returning to in-person schooling. These early conversations also revealed a variety of organizational and pedagogical innovations--from digitally enlivened lessons to intense work with small groups of pupils. Teachers and staff, still dealing with health challenges in their own families, were turning to the social and emotional well-being of their students.

This article provides a review of prior empirical work exploring whether and to what extent school district racial composition affects the costs associated with providing equal educational opportunity to achieve a common set of outcomes. This prior work mainly involves education cost function modeling, on several specific states and in an earlier version of our national education cost model. Here, we update the national education cost model and apply a series of tests for selecting the optimal cost model and determining a) whether it is necessary to retain measures of racial composition in the model and b) the effect those measures have on the estimated costs to achieve common outcomes. We find that the optimal model includes an interaction term between % enrollment that is black and population density and that for majority Black enrollment urban districts, the predicted costs per pupil are 20 to 50% higher when using models with this measure than when using models with race neutral alternatives. While changes in cost estimates for these districts are large, aggregate national cost increases from including racial composition are 1.3 to 2.7% in most years.

This study examined the impact of Classworks® Individualized Learning math instruction on second-grade students performing below the 25th percentile in math. The research, conducted during the 2021-2022 school year, involved 1,367 students from 45 schools across seven districts. A quasi-experimental design compared 941 students who had any use of Classworks (treatment group) with 426 students who did not (comparison group). Students in both the treatment and comparison groups completed the Classworks Math Universal Screener assessments at the beginning and end of the school year. The study employed linear modeling with adjustments for clustering, baseline achievement, and district fixed effects. Results showed that Classworks users scored an average of 5.17 points higher on the spring math assessment compared to non-users, with an effect size of 0.13. Treatment group students engaged with the program for an average of 11 minutes per week over 30 instructional weeks. These findings suggest that Classworks Individualized Learning math instruction has a positive impact on the math performance for struggling second-grade students, even with modest time investment. Educators may consider implementing such personalized, technology-based interventions to support students performing below grade level in mathematics. Future research could explore the program's long-term effects and its impact on different student subgroups. The study includes tables presenting baseline equivalence statistics and model-based results, along with references supporting the methodology and context of the research.

Rigid registration aims to determine the translations and rotations necessary to align features in a pair of images. While recent machine learning methods have become state-of-the-art for linear and deformable registration across subjects, they have demonstrated limitations when applied to longitudinal (within-subject) registration, where achieving precise alignment is critical. Building on an existing framework for anatomy-aware, acquisition-agnostic affine registration, we propose a model optimized for longitudinal, rigid brain registration. By training the model with synthetic within-subject pairs augmented with rigid and subtle nonlinear transforms, the model estimates more accurate rigid transforms than previous cross-subject networks and performs robustly on longitudinal registration pairs within and across magnetic resonance imaging (MRI) contrasts., Comment: 5 pages, 4 figures, 1 table, rigid image registration, deep learning, longitudinal analysis, neuroimaging, accepted by the IEEE International Symposium on Biomedical Imaging

We address the problem of learning to control an unknown nonlinear dynamical system through sequential interactions. Motivated by high-stakes applications in which mistakes can be catastrophic, such as robotics and healthcare, we study situations where it is possible for fast sequential learning to occur. Fast sequential learning is characterized by the ability of the learning agent to incur logarithmic regret relative to a fully-informed baseline. We demonstrate that fast sequential learning is achievable in a diverse class of continuous control problems where the system dynamics depend smoothly on unknown parameters, provided the optimal control policy is persistently exciting. Additionally, we derive a regret bound which grows with the square root of the number of interactions for cases where the optimal policy is not persistently exciting. Our results provide the first regret bounds for controlling nonlinear dynamical systems depending nonlinearly on unknown parameters. We validate the trends our theory predicts in simulation on a simple dynamical system.

Backpropagation is the standard method for achieving state-of-the-art accuracy in neural network training, but it often imposes high memory costs and lacks biological plausibility. In this paper, we introduce the Mono-Forward algorithm, a purely local layerwise learning method inspired by Hinton's Forward-Forward framework. Unlike backpropagation, Mono-Forward optimizes each layer solely with locally available information, eliminating the reliance on global error signals. We evaluated Mono-Forward on multi-layer perceptrons and convolutional neural networks across multiple benchmarks, including MNIST, Fashion-MNIST, CIFAR-10, and CIFAR-100. The test results show that Mono-Forward consistently matches or surpasses the accuracy of backpropagation across all tasks, with significantly reduced and more even memory usage, better parallelizability, and a comparable convergence rate., Comment: 12 pages

The Euler numbers have been widely studied. A signed version of the Euler numbers of even subscript are given by the coefficients of the exponential generating function 1/(1+x^2/2!+x^4/4!+...). Leeming and MacLeod introduced a generalization of the Euler numbers depending on an integer parameter d where one takes the coefficients of the expansion of 1/(1+x^d/d!+x^{2d}/(2d)!+...). These numbers have been shown to have many interesting properties despite being much less studied. And the techniques used have been mainly algebraic. We propose a combinatorial model for them as signed sums over ordered partitions. We show that this approach can be used to prove a number of old and new results including a recursion, integrality, and various congruences. Our methods include sign-reversing involutions and M\"obius inversion over partially ordered sets., Comment: 14 pages, 1 figure

The young disc vertical phase is paramount in our understanding of Galaxy evolution. Analysing the vertical kinematics at different galactic regions provides important information about the space-time variations of the Galactic potential. The vertical phase snail-shell structure found after Gaia DR2 release encompasses a wide range of ages. %\citep{2018Natur.561..360A, Antoja23}. However, the structure of the $V_Z\, vs \, Z$ diagram appears linear when the analysis is limited to studying objects younger than 30 Ma. Based on the vertical velocity and height-over-disc maps obtained for a sample of young open clusters, this method also allows the matter density in the Solar neighbourhood to be estimated using a completely different approach than previously found in the literature. We use two different catalogues of star clusters to confirm the previous result and study new age ranges. The linear pattern between $V_Z$ and $Z$ shows different slopes, $\partial V_Z/\partial Z$, for various age groups. The results fit a simple model (harmonic oscillator) of in-plane decoupled vertical dynamics up to a certain age limit, corresponding to $\sim$ 30 Ma. This work also analyses the relationship between the local volumetric density of matter ($\rho_0$) and the disc vertical kinematics for different age ranges, all below 50 Ma. The best estimates of the effective volumetric mass density in the Solar neighbourhood, 0.09-0.15 M$_\odot$ pc${}^{-3}$, agree with those given by other authors, assessing the reliability of the proposed dynamical model. These values are a minorant of the actual matter density in the region., Comment: 11 pages, 6 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)

Background: Healthcare has many manual processes that can benefit from automation and augmentation with Generative Artificial Intelligence (AI), the medical billing and coding process. However, current foundational Large Language Models (LLMs) perform poorly when tasked with generating accurate International Classification of Diseases, 10th edition, Clinical Modification (ICD-10-CM) and Current Procedural Terminology (CPT) codes. Additionally, there are many security and financial challenges in the application of generative AI to healthcare. We present a strategy for developing generative AI tools in healthcare, specifically for medical billing and coding, that balances accuracy, accessibility, and patient privacy. Methods: We fine tune the PHI-3 Mini and PHI-3 Medium LLMs using institutional data and compare the results against the PHI-3 base model, a PHI-3 RAG application, and GPT-4o. We use the post operative surgical report as input and the patients billing claim the associated ICD-10, CPT, and Modifier codes as the target result. Performance is measured by accuracy of code generation, proportion of invalid codes, and the fidelity of the billing claim format. Results: Both fine-tuned models performed better or as well as GPT-4o. The Phi-3 Medium fine-tuned model showed the best performance (ICD-10 Recall and Precision: 72%, 72%; CPT Recall and Precision: 77%, 79%; Modifier Recall and Precision: 63%, 64%). The Phi-3 Medium fine-tuned model only fabricated 1% of ICD-10 codes and 0.6% of CPT codes generated. Conclusions: Our study shows that a small model that is fine-tuned on domain-specific data for specific tasks using a simple set of open-source tools and minimal technological and monetary requirements performs as well as the larger contemporary consumer models., Comment: 21 pages, 3 figures, 2 tables

The aim of this work is focused on linearizing and found the Lax Pairs of the algebraic complete integrability (a.c.i) Toda lattice associated with the twisted affine Lie algebra \(a_4^{\left(2\right)}\). Firstly, we recall that our case of a.c.i is a two-dimensional algebraic completely integrable systems for which the invariant (real) tori can be extended to complex algebraic tori (abelian surfaces). This implies that the geometry can be used to study this system. Secondly, we show that the lattice is related to the Mumford system and we construct an explicit morphism between these systems, leading to a new Poisson structure for the Mumford system. Finally, we give a new Lax equation for this Toda lattice and we construct an explicit linearization of the system., Comment: 10 pages, 1 figure. arXiv admin note: text overlap with arXiv:2404.13688

One of the fundamental results in graph minor theory is that for every planar graph~$H$, there is a minimum integer~$f(H)$ such that graphs with no minor isomorphic to~$H$ have treewidth at most~$f(H)$. The best known bound for an arbitrary planar $H$ is ${O(|V(H)|^9\operatorname{poly~log} |V(H)|)}$. We show that if $H$ is the disjoint union of cycles, then $f(H)$ is $O(|V(H)|\log^2 |V(H)|)$, which is a $\log|V(H)|$ factor away being optimal.

Gamma-ray bursts are known to display various features on top of their canonical behavior. In this short review, we will describe and discuss two of them: the ultra-long gamma-ray bursts, which are defined by an extreme duration of their prompt phase, and the plateau phase, which is defined by a steady phase of large duration at the start of the afterglow. We will review the main properties of those two phenomena, and will discuss their possible origin, in light of the standard fireball model of gamma-ray bursts. A final section will discuss the future missions which could bring new evidences to the study of those objects., Comment: Invited review accepted for publication in Galaxies

One of the long-standing aspirations in conversational AI is to allow them to autonomously take initiatives in conversations, i.e., being proactive. This is especially challenging for multi-party conversations. Prior NLP research focused mainly on predicting the next speaker from contexts like preceding conversations. In this paper, we demonstrate the limitations of such methods and rethink what it means for AI to be proactive in multi-party, human-AI conversations. We propose that just like humans, rather than merely reacting to turn-taking cues, a proactive AI formulates its own inner thoughts during a conversation, and seeks the right moment to contribute. Through a formative study with 24 participants and inspiration from linguistics and cognitive psychology, we introduce the Inner Thoughts framework. Our framework equips AI with a continuous, covert train of thoughts in parallel to the overt communication process, which enables it to proactively engage by modeling its intrinsic motivation to express these thoughts. We instantiated this framework into two real-time systems: an AI playground web app and a chatbot. Through a technical evaluation and user studies with human participants, our framework significantly surpasses existing baselines on aspects like anthropomorphism, coherence, intelligence, and turn-taking appropriateness.

Agent-based modeling (ABM) has become a cornerstone of complexity science, enabling the study of heterogeneous agents interacting within dynamic environments. Among ABM frameworks, John Conway's Game of Life (GoL) stands out for its simplicity and ability to generate emergent macroscopic patterns from basic microscopic rules. In this paper, we propose and implement a novel GoL-based framework to simulate urban migration dynamics. Using a grid-within-a-grid approach, our approach encodes probabilistic tendencies for out-migration due to densification and sparsification, simulating the evolution of population centers. By initializing GoL grids with different distributions and parameterizing migration preferences, we explore how urban structures emerge and stabilize over time. Through a series of experiments, we demonstrate that even with simple rules, this framework shows promise for understanding emergent urban phenomena, providing insights into city growth and structure. Methodologically, our framework offers a versatile and computationally efficient tool for studying urban migration patterns, contributing to the broader application of ABMs in computational urban social science.

Following the concept of creating individual centers, we investigate local electronic states and magnetic ordering in nickel oxide (NiO) induced by substituting oxygen (O) with nitrogen (N). Each N introduces an additional N 2p hole and alters the state of the magnetic moment at a neighboring nickel (Ni) cation site, as the exchange interaction between this hole and the Ni eg electrons dominates the superexchange interaction. This leads to the formation of Ni-N-Ni units consisting of five spins that are decoupled from the rest of the antiferromagnetic (AFM) NiO lattice and exhibit degenerate spin states. These centers are studied using density functional theory and confirmed through high-resolution spectroscopy measurements on N-substituted NiO thin films grown by molecular beam epitaxy. This type of magnetic design could potentially advance quantum technologies based on strongly correlated materials, such as quantum sensors and spin-qubits., Comment: 12 pages, 14 figures. We are resubmitting the publication since the author Ke Zou was missing. No other changes were made

Recent advancements in Multimodal Large Language Models (MLLMs) have demonstrated remarkable progress in visual understanding. This impressive leap raises a compelling question: how can language models, initially trained solely on linguistic data, effectively interpret and process visual content? This paper aims to address this question with systematic investigation across 4 model families and 4 model scales, uncovering a unique class of attention heads that focus specifically on visual content. Our analysis reveals a strong correlation between the behavior of these attention heads, the distribution of attention weights, and their concentration on visual tokens within the input. These findings enhance our understanding of how LLMs adapt to multimodal tasks, demonstrating their potential to bridge the gap between textual and visual understanding. This work paves the way for the development of AI systems capable of engaging with diverse modalities.

Recent observations by Mistele et al. show that the circular velocity curves of isolated galaxies remain flat out to the largest radii probed so far, i.e. $\approx 1$ Mpc. The velocity decline beyond the expected virial radius is not observed. These results imply that the galaxy halo is in thermal equilibrium even at large radii where particles did not have time to relax. The galaxies must have already formed in the isothermal state. How is this possible? In the present note we try to understand the formation of galaxies with warm dark matter in the expanding universe., Comment: 12 pages, 1 figure

Pulsar timing arrays (PTAs) detect gravitational waves (GWs) via the correlations they create in the arrival times of pulses from different pulsars. The mean correlation, a function of the angle $\gamma$ between the directions to two pulsars, was predicted in 1983 by Hellings and Downs (HD). Observation of this angular pattern is crucial evidence that GWs are present, so PTAs "reconstruct the HD curve" by estimating the correlation using pulsar pairs separated by similar angles. The angular pattern may be also expressed as a "harmonic sum" of Legendre polynomials ${\rm P}_l(\cos \gamma)$, with coefficients $c_l$. Here, assuming that the GWs and pulsar noise are described by a Gaussian ensemble, we derive optimal estimators for the $c_l$ and compute their variance. We consider two choices for "optimal". The first minimizes the variance of each $c_l$, independent of the values of the others. The second finds the set of $c_l$ which minimizes the (squared) deviation of the reconstructed correlation curve from its mean. These are analogous to the so-called "dirty" and "clean" maps of the electromagnetic and (audio-band) GW backgrounds., Comment: 6 pages, 1 figure, 1 table

We introduce and examine the notion of principal $\mathbb{Z}_2^n$-bundles, i.e., principal bundles in the category of $\mathbb{Z}_2^n$-manifolds. The latter are higher graded extensions of supermanifolds in which a $\mathbb{Z}_2^n$-grading replaces $\mathbb{Z}_2$-grading. These extensions have opened up new areas of research of great interest in both physics and mathematics. In principle, the geometry of $\mathbb{Z}_2^n$-manifolds is essentially different than that of supermanifolds, as for $n>1$ we have formal variables of even parity, so local smooth functions are formal power series. On the other hand, a full version of differential calculus is still valid. We show in this paper that the fundamental properties of classical principal bundles can be generalised to the setting of this `higher graded' geometry, with properly defined frame bundles of $\mathbb{Z}_2^n$-vector bundles as canonical examples. However, formulating these concepts and proving these results relies on many technical upshots established in earlier papers. A comprehensive introduction to $\mathbb{Z}_2^n$-manifolds is therefore included together with basic examples., Comment: 22 pages, editorial changes in one section

We describe the development of the Keck Observatory Archive (KOA) Data Discovery Service, a web-based dashboard that returns metadata for wide-area queries of the entire archive in seconds. Currently in beta, this dashboard will support exploration, visualization, and data access across multiple instruments. This effort is underpinned by an open-source, VO-compliant query infrastructure and will offer services that can be hosted on web pages or in Jupyter notebooks. The effort also informs the design of a new, modern landing page that meets the expectations of accessibility and ease of use. The new query infrastructure is based on nexsciTAP, a component-based, DBMS-agnostic Python implementation of the IVOA Table Access Protocol, developed at NExScI and integrated into the NASA Exoplanet Archive and the NEID Archive, and into the PyKOA Python client. This infrastructure incorporates R-tree spatial indexing, built as memory-mapped files as part of Montage, a software toolkit used to create composite astronomical images. Although R-trees are used most often in geospatial analysis, here they enable searches of the entire KOA archive, an eclectic collection of 100 million records of imaging and spectroscopic data, in 2 seconds, and they speed up spatial searches by x20. The front end is built on the open-source Plotly-Dash framework, which allows users to build an interactive user interface based on a single Python file., Comment: 4 pages. Proceedings of Astronomical Society of the Pacific, Malta, 2024

J. Willard Gibbs published a book in 1902 on statistical mechanics that quickly received significant attention from his contemporaries because of the reputation that he had secured with his prior work on thermodynamics. People reading Gibbs's book were often familiar with Ludwig Boltzmann's work on the kinetic theory of gases. This article looks at the published response to Gibbs's book in the decade following its publication. What did these readers get from reading Elementary Principles in Statistical Mechanics? How did they put that to use? Was it fruitful? Samuel H. Burbury had been strongly critical of the assumptions on which Boltzmann's theory was built and had expressed a need for a theory without them. Burbury was pleased with the foundation and rigor of Gibbs's approach, since in his first three chapters he had built on analytical mechanics in the Hamiltonian formulation, a many-body dynamical system. Others found Gibbs hard to understand and some were critical. Hendrik A. Lorentz began teaching Gibbs's dynamical theory and supervised a thesis by Leonard Ornstein in which Orenstein developed applications built on grand canonical ensembles and correlations, which were contrary to Boltzmann's molecular disorder. Ornstein and Zernike expanded on material in that thesis to develop a theory of critical opalescence without infinities., Comment: 20 pages, 1 figure

The Montage Image Mosaic Engine, first released in 2002, has found applicability across the electromagnetic spectrum to support data processing and visualization. This broad applicability has come about through its design as an Open Source ANSI-C toolkit (and Python binary extensions), with independent components to perform each step in the creation of a mosaic and with support for all WCS extensions. This design enables easy integration into custom environments, workflows and pipelines, and is the principal reason for its long lifetime. Here we emphasize the growing use of Montage in radio astronomy (37 peer-reviewed papers since 2020), and will focus on three high-profile applications: (1) Analysis of observations made with SKA precursor experiments, such as MeerKAT and the Murchison Wide-field Array, (2) Faraday tomography of LOFAR Two-Metre Sky Survey data (LoTSS-DR2), which explores the structure of the local interstellar medium, and (3) Identification of fast radio bursts., Comment: 4 pages, 1 figure. Proceedings of ADASS 2024, Valletta, Malta

Appearing in the modular transformation formula for the classical theta function $\theta_3(z)$ is the sum $S(h,k):=\sum_{j=1}^{k-1}(-1)^{j+1+[hj/k]}$, which is an analogue of the classical Dedekind sum $s(h,k).$ We establish several properties for $S(h,k)$ and $S(k) := \sum_{h=1}^{k-1}S(h,k).$ Several conjectures about the values of $S(k)$ are given.

We probe the spectrum of elementary excitations in SrIrO$_3$ by using heterostructured [(SrIrO$_3$)$_m$/(SrTiO$_3$)$_l$] samples to approach the bulk limit. Our resonant inelastic x-ray scattering (RIXS) measurements at the Ir $L_3$-edge reveal a robust low-lying collective magnetic mode with an antiferromagnetic (AF) dispersion similar to the insulators Sr$_2$IrO$_4$ and Sr$_3$Ir$_2$O$_7$, albeit with a large gap and much larger linewidth. At higher energies we find the spin-orbit exciton, also strongly broadened, but with an inverted dispersion and doubled periodicity that are controlled by the charge hopping. These results demonstrate that the AF paramagnon persists, somewhat counterintuitively, far into the metallic regime of the insulator-metal transition driven by the degree of confinement in the heterostructure. We conclude that these two excitations, which are contrasting but coexisting hallmarks of strong AF pseudospin and charge fluctuations in a spin-orbit-coupled Mott-Slater material, are properties intrinsic to the ground state of semimetallic perovskite SrIrO$_3$.

The co-occurrence of multiple long-term conditions (MLTC), or multimorbidity, in an individual can reduce their lifespan and severely impact their quality of life. Exploring the longitudinal patterns, e.g. clusters, of disease accrual can help better understand the genetic and environmental drivers of multimorbidity, and potentially identify individuals who may benefit from early targeted intervention. We introduce $\textit{probabilistic modelling of onset times}$, or $\texttt{ProMOTe}$, for clustering and forecasting MLTC trajectories. $\texttt{ProMOTe}$ seamlessly learns from incomplete and unreliable disease trajectories that is commonplace in Electronic Health Records but often ignored in existing longitudinal clustering methods. We analyse data from 150,000 individuals in the UK Biobank and identify 50 clusters showing patterns of disease accrual that have also been reported by some recent studies. We further discuss the forecasting capabilities of the model given the history of disease accrual.

The Virtual Observatory (VO) is a global ecosystem of interoperating services that connect worldwide data archives. The VO is implemented in all major astronomy archives through common interfaces developed by the 22 members of the International Virtual Observatory Alliance (IVOA). It was founded in 2002, and the newest members, the SKA Observatory and the Kazakhstan Virtual Observatory, joined in 2022. The VO offers access to data on FAIR principles and from its inception has supported Open Science. The VO acts as a democratizing influence in astronomy: it provides equal access to worldwide public data sets to underserved communities as well as to large data centers, and it enables international participation in scientific research and education. Thus, astronomers from many different communities are positioned to participate in the big science questions emerging in astronomy in the 2020s, such as interpreting transient sources that will be measured in forthcoming missions such as Rubin. In addition, the IVOA has signed an MoU with the IAU Office of Astronomy for Development (OAD). Under this MoU, IVOA members participated in "Astronomy from Archival Data," which involved educational activities for undergraduate and post-graduate students organized by Dr. Priya Hasan. The IVOA plans to participate in future such educational events. The presentation describes how new communities may participate in Virtual Observatory science and educational activities, including practices for developing VO-compliant data centers and archives and education and training for developers and end users., Comment: 4 pages. Proceedings of IAU General Assembly, Cape Town, South Africa, 2024

One of the bottlenecks preventing Deep Reinforcement Learning algorithms (DRL) from real-world applications is how to explore the environment and collect informative transitions efficiently. The present paper describes bounded exploration, a novel exploration method that integrates both 'soft' and intrinsic motivation exploration. Bounded exploration notably improved the Soft Actor-Critic algorithm's performance and its model-based extension's converging speed. It achieved the highest score in 6 out of 8 experiments. Bounded exploration presents an alternative method to introduce intrinsic motivations to exploration when the original reward function has strict meanings., Comment: 8 pages, 7 figures. Accepted as a poster presentation in the Australian Robotics and Automation Association (2023)

These are expanded notes for a short series of lectures, presented at the University of Luxembourg in 2017, giving an introduction to some of the ideas of supersymmetry and supergeometry. In particular, we start from some motivating facts in physics, pass to the theory of supermanifolds, then to spinors, ending up at super-Minkowski space-times. We examine some salient mathematical issues with understanding supersymmetry in a classical setting and make no attempt to discuss phenomenologically interesting models. Moreover, the presentation is "light" in the sense that nothing is carefully proved. The audience of the seminars ranged from Ph.D. students to more experienced postdocs in mathematics. The audience was assumed to have a working knowledge of differential geometry, elementary category theory, and basic ideas from classical & quantum mechanics. No knowledge of supersymmetry and supergeometry is assumed., Comment: 24 pages, notes based on a series of lectures. Further reference added for version II

High performance computing (HPC) architectures have undergone rapid development in recent years. As a result, established software suites face an ever increasing challenge to remain performant on and portable across modern systems. Many of the widely adopted atmospheric modeling codes cannot fully (or in some cases, at all) leverage the acceleration provided by General-Purpose Graphics Processing Units (GPGPUs), leaving users of those codes constrained to increasingly limited HPC resources. Energy Research and Forecasting (ERF) is a regional atmospheric modeling code that leverages the latest HPC architectures, whether composed of only Central Processing Units (CPUs) or incorporating GPUs. ERF contains many of the standard discretizations and basic features needed to model general atmospheric dynamics as well as flows relevant to renewable energy. The modular design of ERF provides a flexible platform for exploring different physics parameterizations and numerical strategies. ERF is built on a state-of-the-art, well-supported, software framework (AMReX) that provides a performance portable interface and ensures ERF's long-term sustainability on next generation computing systems. This paper details the numerical methodology of ERF and presents results for a series of verification and validation cases.

Ultrafast X-ray spectroscopy has been revolutionized in recent years due to the advent of fourth-generation X-ray facilities. In solid-state materials, core excitons determine the energy and line shape of absorption features in core-level spectroscopies such as X-ray absorption spectroscopy. The screening of core excitons is an inherent many-body process that can reveal insight into charge-transfer excitations and electronic correlations. Under non-equilibrium conditions such as after photoexcitation, however, core-exciton screening is still not fully understood. Here we demonstrate the dynamic Coulomb screening of core excitons induced by photoexcited carriers by employing X-ray transient absorption (XTA) spectroscopy with picosecond time resolution. Our interpretation is supported by state-of-the-art ab initio theory, combining constrained and real-time time-dependent density functional theory with many-body perturbation theory. Using ZnO as an archetypal wide band-gap semiconductor, we show that the Coulomb screening by photoexcited carriers at the Zn K-edge leads to a decrease in the core-exciton binding energy, which depends nonlinearly on both the excitation density and the distribution of photoexcited carriers in reciprocal space. The effect of Coulomb screening dominates over Pauli blocking in the XTA spectra. We show that dynamic core-exciton screening is also observed at other X-ray absorption edges and theoretically predict the effect of core-exciton screening on the femtosecond time scale for the case of ZnO, a major step towards hard X-ray excitonics. The results have implications for the interpretation of ultrafast X-ray spectra in general and their use in tracking charge carrier dynamics in complex materials on atomic length scales., Comment: 43 pages, 37 figures

In this paper, we evaluate the capability of large language models to conduct personalized phishing attacks and compare their performance with human experts and AI models from last year. We include four email groups with a combined total of 101 participants: A control group of arbitrary phishing emails, which received a click-through rate (recipient pressed a link in the email) of 12%, emails generated by human experts (54% click-through), fully AI-automated emails 54% (click-through), and AI emails utilizing a human-in-the-loop (56% click-through). Thus, the AI-automated attacks performed on par with human experts and 350% better than the control group. The results are a significant improvement from similar studies conducted last year, highlighting the increased deceptive capabilities of AI models. Our AI-automated emails were sent using a custom-built tool that automates the entire spear phishing process, including information gathering and creating personalized vulnerability profiles for each target. The AI-gathered information was accurate and useful in 88% of cases and only produced inaccurate profiles for 4% of the participants. We also use language models to detect the intention of emails. Claude 3.5 Sonnet scored well above 90% with low false-positive rates and detected several seemingly benign emails that passed human detection. Lastly, we analyze the economics of phishing, highlighting how AI enables attackers to target more individuals at lower cost and increase profitability by up to 50 times for larger audiences.

We derive a formula for the product of two Dirichlet series satisfying Hecke's functional equation. Several examples are given. At the conclusion of our paper, we address prior work on similar formulas., Comment: 11 pages

For decades, researchers have developed task-specific models to address scientific challenges across diverse disciplines. Recently, large language models (LLMs) have shown enormous capabilities in handling general tasks; however, these models encounter difficulties in addressing real-world scientific problems, particularly in domains involving large-scale numerical data analysis, such as experimental high energy physics. This limitation is primarily due to BPE tokenization's inefficacy with numerical data. In this paper, we propose a task-agnostic architecture, BBT-Neutron, which employs a binary tokenization method to facilitate pretraining on a mixture of textual and large-scale numerical experimental data. We demonstrate the application of BBT-Neutron to Jet Origin Identification (JoI), a critical categorization challenge in high-energy physics that distinguishes jets originating from various quarks or gluons. Our results indicate that BBT-Neutron achieves comparable performance to state-of-the-art task-specific JoI models. Furthermore, we examine the scaling behavior of BBT-Neutron's performance with increasing data volume, suggesting the potential for BBT-Neutron to serve as a foundational model for particle physics data analysis, with possible extensions to a broad spectrum of scientific computing applications for Big Science experiments, industrial manufacturing and spacial computing. The project code is available at https://github.com/supersymmetry-technologies/bbt-neutron.

New lattice quantizers with lower normalized second moments than previously reported are constructed in 13 and 14 dimensions and conjectured to be optimal. Our construction combines an initial numerical optimization with a subsequent analytical optimization of families of lattices, whose Voronoi regions are constructed exactly. The new lattices are constructed from glued products of previously known lattices, by scaling the component lattices and then optimizing the scale factors. A two-parameter family of lattices in 13 dimensions reveals an intricate landscape of phase changes as the parameters are varied., Comment: 16 pages, 7 figures