Search

This appendix volume supplements report NCEE 2025-005r, describing the effects of two experimental Pell Grant expansions: the first allowed income-eligible students with a bachelor's degree to obtain this form of federal financial aid for short-term occupational training programs and the second allowed income-eligible students to obtain this aid for very short-term programs lasting as little as eight weeks. The report updated earlier results from a rigorous evaluation of the experiments conducted by the U.S. Department of Education's (ED's) Institute of Education Sciences, adding new information about the experiments' effects on employment and earnings outcomes. This appendix volume similarly updates the evaluation's previous published appendix to detail the steps taken to assess these effects. The volume first describes the Pell Grant experiments as overseen by ED's Office of Federal Student Aid, including information about the schools offering the experimental Pell Grants. The next sections document how the study was conducted, including the research questions asked and the approach used to answer the research questions, additional details supporting the key findings of the study, and additional analyses to assess the sensitivity of the key findings.

Pell Grants are the cornerstone of federal financial aid for students with low income who are enrolled in postsecondary education. Currently, these grants are available only to those who seek an initial undergraduate degree or credential requiring at least a typical semester of instruction. Because these rules may restrict access to programs providing skills needed for new or better jobs, in 2011 the U.S. Department of Education (ED) began pilots of two experimental expansions to Pell Grant eligibility. The first experiment allowed income-eligible students with a bachelor's degree to obtain Pell Grants for short-term occupational training programs. The second experiment allowed income-eligible students to obtain Pell Grants for very short-term programs lasting as little as eight weeks. This report updates earlier results from a rigorous evaluation of the experiments conducted by ED's Institute of Education Sciences (IES), adding new information about the experiments' impacts on labor market success. This fuller picture could help Congress as it considers legislation to make Pell Grants for short-term occupational training permanent policy.

The gut-brain axis has emerged as a key player in the regulation of brain function and cognitive health. Gut microbiota dysbiosis has been observed in preclinical models of Alzheimer's disease and patients. Manipulating the composition of the gut microbiota enhances or delays neuropathology and cognitive deficits in mouse models. Accordingly, the health status of the animal facility may strongly influence these outcomes. In the present study, we longitudinally analysed the faecal microbiota composition and amyloid pathology of 5XFAD mice housed in a specific opportunistic pathogen-free (SOPF) and a conventional facility. The composition of the microbiota of 5XFAD mice after aging in conventional facility showed marked differences compared to WT littermates that were not observed when the mice were bred in SOPF facility. The development of amyloid pathology was also enhanced by conventional housing. We then transplanted faecal microbiota (FMT) from both sources into wild-type (WT) mice and measured memory performance, assessed in the novel object recognition test, in transplanted animals. Mice transplanted with microbiota from conventionally bred 5XFAD mice showed impaired memory performance, whereas FMT from mice housed in SOPF facility did not induce memory deficits in transplanted mice. Finally, 18 weeks of housing SOPF-born animals in a conventional facility resulted in the reappearance of specific microbiota compositions in 5XFAD vs WT mice. In conclusion, these results show a strong impact of housing conditions on microbiota-associated phenotypes and question the relevance of breeding preclinical models in specific pathogen-free (SPF) facilities.

The certification of randomness is essential for both fundamental science and information technologies. Unlike traditional random number generators, randomness obtained from nonlocal correlations is fundamentally guaranteed to be unpredictable. However, it is also highly susceptible to noise. Here, we show that extending the conventional bipartite Bell scenario to hybrid quantum networks -- which incorporate both quantum channels and entanglement sources -- enhances the robustness of certifiable randomness. Our protocol even enables randomness to be certified from Bell-local states, broadening the range of quantum states useful for this task. Through both theoretical analysis and experimental validation in a photonic network, we demonstrate enhanced performance and improved noise resilience., Comment: 14 pages, 6 figures

Due to the increase of data volumes expected for the LHC Run 3 and Run 4, the ALICE Collaboration designed and deployed a new, energy efficient, computing model to run Online and Offline O$^2$ data processing within a single software framework. The ALICE O$^2$ Event Processing Nodes (EPN) project performs online data reconstruction using GPUs (Graphic Processing Units) instead of CPUs and applies an efficient, entropy-based, online data compression to cope with PbPb collision data at a 50 kHz hadronic interaction rate. Also, the O$^2$ EPN farm infrastructure features an energy efficient, environmentally friendly, adiabatic cooling system which allows for operational and capital cost savings.

We measure the cross-correlation between cosmic shear from the third-year release of the Dark Energy Survey, thermal Sunyaev-Zel'dovich (tSZ) maps from Planck, and X-ray maps from ROSAT. We investigate the possibility of developing a physical model able to jointly describe both measurements, simultaneously constraining the spatial distribution and thermodynamic properties of hot gas. We find that a relatively simple model is able to describe both sets of measurements and to make reasonably accurate predictions for other observables (the tSZ auto-correlation, its cross-correlation with X-rays, and tomographic measurements of the bias-weighted mean gas pressure). We show, however, that contamination from X-ray AGN, as well as the impact of non-thermal pressure support, must be incorporated in order to fully resolve tensions in parameter space between different data combinations. We obtain simultaneous constraints on the mass scale at which half of the gas content has been expelled from the halo, $\mathrm{log}_{10}(M_c)=14.83^{+0.16}_{-0.23}$, on the polytropic index of the gas, $\Gamma=1.144^{+0.016}_{-0.013}$, and on the ratio of the central gas temperature to the virial temperature $\alpha_T=1.30^{+0.15}_{-0.28}$., Comment: 20 pages, 10 figures

Wearable sensors offer a non-invasive way to collect physical activity (PA) data, with walking as a key component. Existing models often struggle to detect gait bouts in individuals with neurodegenerative diseases (NDDs) involving involuntary movements. We developed J-Net, a deep learning model inspired by U-Net, which uses a pre-trained self-supervised foundation model fine-tuned with Huntington`s disease (HD) in-lab data and paired with a segmentation head for gait detection. J-Net processes wrist-worn accelerometer data to detect gait during daily living. We evaluated J-Net on in-lab and daily-living data from HD, Parkinson`s disease (PD), and controls. J-Net achieved a 10-percentage point improvement in ROC-AUC for HD over existing methods, reaching 0.97 for in-lab data. In daily-living environments, J-Net estimates showed no significant differences in median daily walking time between HD and controls (p = 0.23), in contrast to other models, which indicated counterintuitive results (p < 0.005). Walking time measured by J-Net correlated with the UHDRS-TMS clinical severity score (r=-0.52; p=0.02), confirming its clinical relevance. Fine-tuning J-Net on PD data also improved gait detection over current methods. J-Net`s architecture effectively addresses the challenges of gait detection in severe chorea and offers robust performance in daily living. The dataset and J-Net model are publicly available, providing a resource for further research into NDD-related gait impairments.

We examine the geometry of neural network training using the Jacobian of trained network parameters with respect to their initial values. Our analysis reveals low-dimensional structure in the training process which is dependent on the input data but largely independent of the labels. We find that the singular value spectrum of the Jacobian matrix consists of three distinctive regions: a "chaotic" region of values orders of magnitude greater than one, a large "bulk" region of values extremely close to one, and a "stable" region of values less than one. Along each bulk direction, the left and right singular vectors are nearly identical, indicating that perturbations to the initialization are carried through training almost unchanged. These perturbations have virtually no effect on the network's output in-distribution, yet do have an effect far out-of-distribution. While the Jacobian applies only locally around a single initialization, we find substantial overlap in bulk subspaces for different random seeds. Our code is available at https://github.com/EleutherAI/training-jacobian

Introduction Lifetime risks quantify health risks from radiation exposure and play an important role in radiation detriment and radon dose conversion. This study considers the lifetime risk of dying from lung cancer related to occupational radon exposure, focusing on lifetime excess absolute risk (LEAR), in addition to other lifetime risk measures. This article derives and discusses uncertainty intervals for these estimates. Methods Uncertainties in two components of lifetime risk calculations are modeled: risk model parameter estimates for excess relative risk of lung cancer and baseline mortality rates. Approximate normality assumption (ANA) methods and Bayesian techniques quantify risk model parameter uncertainty. The methods are applied to risk models from the German "Wismut" uranium miners cohort study (full cohort with follow-up 2018 and the 1960+ sub-cohort of miners hired in 1960 or later). Mortality rate uncertainty is assessed based on WHO data. Monte Carlo simulations yield uncertainty intervals, which are compared across different lifetime risk measures. Results Risk model parameter uncertainty is the largest contributor to lifetime risk uncertainty, with baseline mortality rate uncertainty also significant. For the 1960+ sub-cohort risk model, LEAR was 6.70% (95% uncertainty interval: [3.26, 12.28]) for an exposure of 2 Working Level Months from age 18-64, compared to 3.43% ([2.06, 4.84]) for the full cohort. Differences across lifetime risk measures are minor. Conclusion Here, risk model parameter uncertainty substantially drives lifetime risk uncertainty, supporting the use of ANA methods for practicality. Choice of lifetime risk measures has negligible impact. Derived uncertainty intervals align with the range of lifetime risk estimates from uranium miners studies in the literature and should inform radiation protection policies based on lifetime risks., Comment: 59 pages (25 main and 34 supplement),21 figures (3 in main and 18 in supplement)

We study the longitudinal spin polarization of a relativistic fluid of massive spin-1/2 particles undergoing a boost-invariant expansion in the longitudinal direction and rotating in the transverse plane. We express the polarization vector in terms of spin moments and derive closed equations of motion for the latter using spin kinetic theory with a nonlocal relaxation time approximation. These equations of motion are valid at any time of the evolution, from the free-streaming regime to the hydrodynamic regime. At late time, the polarization features contributions from gradients of the fluid velocity and of the temperature, that emerge from the nonlocal part of the collision term. Our results can be used to explore polarization phenomena in the context of heavy-ion collisions., Comment: 34 pages

We evaluate how well Large Language Models (LLMs) latently recall and compose facts to answer multi-hop queries like "In the year Scarlett Johansson was born, the Summer Olympics were hosted in the country of". One major challenge in evaluating this ability is that LLMs may have developed shortcuts by encounters of the head entity "Scarlett Johansson" and the answer entity "United States" in the same training sequences or merely guess the answer based on frequency-based priors. To prevent shortcuts, we exclude test queries where the head and answer entities co-appear in pretraining corpora. Through careful selection of relations and facts and systematic removal of cases where models might guess answers or exploit partial matches, we construct an evaluation dataset SOCRATES (ShOrtCut-fRee lATent rEaSoning). We observe that LLMs demonstrate promising latent multi-hop reasoning abilities without exploiting shortcuts, but only for certain types of queries. For queries requiring latent recall of countries as the intermediate answer, the best models achieve 80% latent composability, but this drops to just 5% for the recall of years. Comparisons with Chain-of-Thought composability highlight a significant gap between the ability of models to reason latently versus explicitly. Analysis reveals that latent representations of the intermediate answer are constructed more often in queries with higher latent composability, and shows the emergence of latent multi-hop reasoning during pretraining.

We perform a thorough investigation of the universality of the long distance matrix elements (LDMEs) of nonrelativistic QCD factorization based on a next-to-leading order (NLO) fit of $J/\psi$ color octet (CO) LDMEs to high transverse momentum $p_T$ $J/\psi$ and $\eta_c$ production data at the LHC. We thereby apply a novel fit-and-predict procedure to systematically take into account scale variations, and predict various observables never studied in this context before. In particular, the LDMEs can well describe $J/\psi$ hadroproduction up to the highest measured values of $p_T$, as well as $\Upsilon(nS)$ production via potential NRQCD based relations. Furthermore, $J/\psi$ production in $\gamma \gamma$ and $\gamma p$ collisions is surprisingly reproduced down to $p_T=1$ GeV, as long as the region of large inelasticity $z$ is excluded, which may be of significance in future quarkonium studies, in particular at the EIC and the high-luminosity LHC. In addition, our summary reveals an interesting pattern as to which observables still evade a consistent description., Comment: 15 pages, 16 figures

We consider the problem of predicting perturbation effects via causal models. In many applications, it is a priori unknown which mechanisms of a system are modified by an external perturbation, even though the features of the perturbation are available. For example, in genomics, some properties of a drug may be known, but not their causal effects on the regulatory pathways of cells. We propose a generative intervention model (GIM) that learns to map these perturbation features to distributions over atomic interventions in a jointly-estimated causal model. Contrary to prior approaches, this enables us to predict the distribution shifts of unseen perturbation features while gaining insights about their mechanistic effects in the underlying data-generating process. On synthetic data and scRNA-seq drug perturbation data, GIMs achieve robust out-of-distribution predictions on par with unstructured approaches, while effectively inferring the underlying perturbation mechanisms, often better than other causal inference methods.

Two decades ago the $\chi_{c1}\left(3872\right)$ was discovered in the hadron spectrum with two heavy quarks. The discovery fueled a surge in experimental research, uncovering dozens of so called XYZ exotics states lying outside the conventional quark model, as well as theoretical investigations into new forms of matter, such as quark-gluon hybrids, mesonic molecules, and tetraquarks, with the potential of disclosing new information about the fundamental strong force. Among the XYZs, the $\chi_{c1}\left(3872\right)$ and $T_{cc}^+\left(3875\right)$ stand out for their striking characteristics and unlashed many discussions about their nature. Here, we address this question using the Born--Oppenheimer Effective Field Theory (BOEFT) and show how QCD settles the issue of their composition. Not only we describe well the main features of the $\chi_{c1}\left(3872\right)$ and $T_{cc}^+\left(3875\right)$ but obtain also model independent predictions in the bottomonium sector. This opens the way to systematic applications of BOEFT to all XYZs., Comment: 8 pages, 3 figures

Matter-RADiation interaction SIMulation (MRADSIM) is an innovative modular software toolkit developed to simulate the effects of radiation on electronic components, human beings and various materials. It incorporates innovative features aimed at enhancing parametric precision, reducing computational time, and introducing supplementary functions for tailored calculations across diverse applications including the applications required for space missions. Notably, MRADSIM is distinguished as the pioneering simulation toolkit to integrate cutting-edge Artificial Intelligence and Machine Learning (AI/ML) algorithms, with the primary objective of effectively recognizing potential radiation-induced issues and facilitating the implementation of mitigation strategies to avert catastrophic failures in mission-critical systems, whether terrestrial or space-based. The distinctive attributes of MRADSIM, coupled with its early adoption by researchers from the National Institute for Nuclear Physics of Italy (INFN), significantly contribute to the toolkits added value., Comment: 11 pages, 12 figures, 3 tables, 75th International Astronautical Congress (IAC), Milan, Italy, 14-18 October

We propose affine concept editing (ACE) as an approach for steering language models' behavior by intervening directly in activations. We begin with an affine decomposition of model activation vectors and show that prior methods for steering model behavior correspond to subsets of terms of this decomposition. We then provide a derivation of ACE and use it to control refusal behavior on ten different models, including Llama 3 70B. ACE combines affine subspace projection and activation addition to reliably control the model's refusal responses across prompt types. We evaluate the results using LLM-based scoring on a collection of harmful and harmless prompts. Our experiments demonstrate that ACE consistently achieves more precise control over model behavior than existing methods and generalizes to models where directional ablation via affine subspace projection alone produces incoherent outputs. Code for reproducing our results is available at https://github.com/EleutherAI/steering-llama3 ., Comment: added plots for results from additional models

Uncertainty Quantification (UQ) is crucial for reliable image segmentation. Yet, while the field sees continual development of novel methods, a lack of agreed-upon benchmarks limits their systematic comparison and evaluation: Current UQ methods are typically tested either on overly simplistic toy datasets or on complex real-world datasets that do not allow to discern true uncertainty. To unify both controllability and complexity, we introduce Arctique, a procedurally generated dataset modeled after histopathological colon images. We chose histopathological images for two reasons: 1) their complexity in terms of intricate object structures and highly variable appearance, which yields challenging segmentation problems, and 2) their broad prevalence for medical diagnosis and respective relevance of high-quality UQ. To generate Arctique, we established a Blender-based framework for 3D scene creation with intrinsic noise manipulation. Arctique contains 50,000 rendered images with precise masks as well as noisy label simulations. We show that by independently controlling the uncertainty in both images and labels, we can effectively study the performance of several commonly used UQ methods. Hence, Arctique serves as a critical resource for benchmarking and advancing UQ techniques and other methodologies in complex, multi-object environments, bridging the gap between realism and controllability. All code is publicly available, allowing re-creation and controlled manipulations of our shipped images as well as creation and rendering of new scenes., Comment: 13 pages, 4 figures

Large language models (LLMs) offer promise in generating educational content, providing instructor feedback, and reducing teacher workload on assessments. While prior studies have focused on studying LLM-powered learning analytics, limited research has examined how effective LLMs are in a bilingual context. In this paper, we study the effectiveness of multilingual large language models (MLLMs) across monolingual (English-only, Spanish-only) and bilingual (Spanglish) student writing. We present a learning analytics use case that details LLM performance in assessing acceptable and unacceptable explanations of Science and Social Science concepts. Our findings reveal a significant bias in the grading performance of pre-trained models for bilingual writing compared to English-only and Spanish-only writing. Following this, we fine-tune open-source MLLMs including Llama 3.1 and Mistral NeMo using synthetic datasets generated in English, Spanish, and Spanglish. Our experiments indicate that the models perform significantly better for all three languages after fine-tuning with bilingual data. This study highlights the potential of enhancing MLLM effectiveness to support authentic language practices amongst bilingual learners. It also aims to illustrate the value of incorporating non-English languages into the design and implementation of language models in education.

In molecular systems, the ultrafast motion of electrons initiates the process of chemical change. Tracking this electronic motion across molecules requires coupling attosecond time resolution to atomic-scale spatial sensitivity. In this work, we employ a pair of attosecond x-ray pulses from an x-ray free-electron laser to follow electron motion resulting from the sudden removal of an electron from a prototypical aromatic system, para-aminophenol. X-ray absorption enables tracking this motion with atomic-site specificity. Our measurements are compared with state-of-the-art computational modeling, reproducing the observed response across multiple timescales. Sub-femtosecond dynamics are assigned to states undergoing non-radiative decay, while few-femtosecond oscillatory motion is associated with electronic wavepacket motion in stable cation states, that will eventually couple to nuclear motion. Our work provides insight on the ultrafast charge motion preceding and initiating chemical transformations in moderately complex systems, and provides a powerful benchmark for computational models of ultrafast charge motion in matter.

We present statistical analysis of video meteor observations for the Perseid and Geminid showers taken with two camera systems operating in Hungary from the end of 2019 through 2023. Zenithal hourly rates (ZHR) and meteor fluxes, determined by MetRec-based analog video cameras HUKON, HUPIS and HUHOD, are inferred and compared with detections of slow fireballs measured at the same sites by a system consisting of automated DSLR cameras (the KoMON system)., Comment: 7 pages, accepted for publication in WGN, Journal of the International Meteor Organization (IMO)

Active labour market policies are widely used by the Swiss government, enrolling more than half of unemployed individuals. This paper analyses whether the Swiss programmes increase future employment and earnings of the unemployed by using causal machine learning methods and leveraging an administrative dataset that captures the population of unemployed and their labour market histories. The findings indicate a small positive average effect on employment and earnings three years after starting a specific Temporary Wage Subsidy programme. In contrast, we find negative effects for Basic Courses, such as job application training, on both outcomes three years after starting the programme. We find no significant effect for Employment Programmes which are conducted outside the regular labour market and Training Courses, such as language and computer courses. The programmes are most effective for individuals with lower education levels and with a migration background from non-EU countries. Last, shallow policy trees provide practical guidance on how the allocation of individuals to programmes could be optimised.

Photometric galaxy surveys, despite their limited resolution along the line of sight, encode rich information about the large-scale structure (LSS) of the Universe thanks to the large number density and extensive depth of the data. However, the complicated selection effects in wide and deep surveys will potentially cause significant bias in the angular two-point correlation function (2PCF) measured from those surveys. In this paper, we measure the 2PCF from the newly published KiDS-Legacy sample. Given an $r$-band $5\sigma$ magnitude limit of $24.8$ and survey footprint of $1347$ deg$^2$, it achieves an excellent combination of sky coverage and depth for such a measurement. We find that complex selection effects, primarily induced by varying seeing, introduce over-estimation of the 2PCF by approximately an order of magnitude. To correct for such effects, we apply a machine learning-based method to recover an ``organised random'' (OR) that presents the same selection pattern as the galaxy sample. The basic idea is to find the selection-induced clustering of galaxies using a combination of self-organising maps (SOM) and hierarchical clustering (HC). This unsupervised machine learning method is able to recover complicated selection effects without specifying their functional forms. We validate this ``SOM+HC'' method on mock deep galaxy samples with realistic systematics and selections derived from the KiDS-Legacy catalogue. Using mock data, we demonstrate that the OR delivers unbiased 2PCF cosmological parameter constraints, removing the $27\sigma$ offset in the galaxy bias parameter that is recovered when adopting uniform randoms. Blinded measurements on the real KiDS-Legacy data show that the corrected 2PCF is robust to the SOM+HC configuration near the optimal setup suggested by the mock tests. Our software is open-source for future usage., Comment: 29 pages, 27 figures, 4 tables

Since the release of ChatGPT in 2022, Generative AI (GenAI) is increasingly being used in higher education computing classrooms across the United States. While scholars have looked at overall institutional guidance for the use of GenAI and reports have documented the response from schools in the form of broad guidance to instructors, we do not know what policies and practices instructors are actually adopting and how they are being communicated to students through course syllabi. To study instructors' policy guidance, we collected 98 computing course syllabi from 54 R1 institutions in the U.S. and studied the GenAI policies they adopted and the surrounding discourse. Our analysis shows that 1) most instructions related to GenAI use were as part of the academic integrity policy for the course and 2) most syllabi prohibited or restricted GenAI use, often warning students about the broader implications of using GenAI, e.g. lack of veracity, privacy risks, and hindering learning. Beyond this, there was wide variation in how instructors approached GenAI including a focus on how to cite GenAI use, conceptualizing GenAI as an assistant, often in an anthropomorphic manner, and mentioning specific GenAI tools for use. We discuss the implications of our findings and conclude with current best practices for instructors.

This paper investigates the mental health penalty for women after childbirth in Switzerland. Leveraging insurance data, we employ a staggered difference-in-difference research design. The findings reveal a substantial mental health penalty for women following the birth of their first child. Approximately four years after childbirth, there is a one percentage point (p.p.) increase in antidepressant prescriptions, representing a 50% increase compared to pre-birth levels. This increase rises to 1.7 p.p. (a 70% increase) six years postpartum. The mental health penalty is likely not only a direct consequence of giving birth but also a consequence of the changed life circumstances and time constraints that accompany it, as the penalty is rising over time and is higher for women who are employed before childbirth.

GPT-4o is an autoregressive omni model that accepts as input any combination of text, audio, image, and video, and generates any combination of text, audio, and image outputs. It's trained end-to-end across text, vision, and audio, meaning all inputs and outputs are processed by the same neural network. GPT-4o can respond to audio inputs in as little as 232 milliseconds, with an average of 320 milliseconds, which is similar to human response time in conversation. It matches GPT-4 Turbo performance on text in English and code, with significant improvement on text in non-English languages, while also being much faster and 50\% cheaper in the API. GPT-4o is especially better at vision and audio understanding compared to existing models. In line with our commitment to building AI safely and consistent with our voluntary commitments to the White House, we are sharing the GPT-4o System Card, which includes our Preparedness Framework evaluations. In this System Card, we provide a detailed look at GPT-4o's capabilities, limitations, and safety evaluations across multiple categories, focusing on speech-to-speech while also evaluating text and image capabilities, and measures we've implemented to ensure the model is safe and aligned. We also include third-party assessments on dangerous capabilities, as well as discussion of potential societal impacts of GPT-4o's text and vision capabilities.

We present a quantum computational framework for SU(2) lattice gauge theory, leveraging continuous variables instead of discrete qubits to represent the infinite-dimensional Hilbert space of the gauge fields. We consider a ladder as well as a two-dimensional grid of plaquettes, detailing the use of gauge fixing to reduce the degrees of freedom and simplify the Hamiltonian. We demonstrate how the system dynamics, ground states, and energy gaps can be computed using the continuous-variable approach to quantum computing. Our results indicate that it is feasible to study non-Abelian gauge theories with continuous variables, providing new avenues for understanding the real-time dynamics of quantum field theories., Comment: 26 pages

While behavioural equivalences among systems of the same type, such as Park/Milner bisimilarity of labelled transition systems, are an established notion, a systematic treatment of relationships between systems of different type is currently missing. We provide such a treatment in the framework of universal coalgebra, in which the type of a system (nondeterministic, probabilistic, weighted, game-based etc.) is abstracted as a set functor: We introduce relational connectors among set functors, which induce notions of heterogeneous (bi)simulation among coalgebras of the respective types. We give a number of constructions on relational connectors. In particular, we identify composition and converse operations on relational connectors; we construct corresponding identity relational connectors, showing that the latter generalize the standard Barr extension of weak-pullback-preserving functors; and we introduce a Kantorovich construction in which relational connectors are induced from relations between modalities. For Kantorovich relational connectors, one has a notion of dual-purpose modal logic interpreted over both system types, and we prove a corresponding Hennessy-Milner-type theorem stating that generalized (bi)similarity coincides with theory inclusion on finitely-branching systems. We apply these results to a number of example scenarios involving labelled transition systems with different label alphabets, probabilistic systems, and input/output conformances.

Generic notions of bisimulation for various types of systems (nondeterministic, probabilistic, weighted etc.) rely on identity-preserving (normal) lax extensions of the functor encapsulating the system type, in the paradigm of universal coalgebra. It is known that preservation of weak pullbacks is a sufficient condition for a functor to admit a normal lax extension (the Barr extension, which in fact is then even strict); in the converse direction, nothing is currently known about necessary (weak) pullback preservation conditions for the existence of normal lax extensions. In the present work, we narrow this gap by showing on the one hand that functors admitting a normal lax extension preserve 1/4-iso pullbacks, i.e. pullbacks in which at least one of the projections is an isomorphism. On the other hand, we give sufficient conditions, showing that a functor admits a normal lax extension if it weakly preserves either 1/4-iso pullbacks and 4/4-epi pullbacks (i.e. pullbacks in which all morphisms are epic) or inverse images. We apply these criteria to concrete examples, in particular to functors modelling neighbourhood systems and weighted systems.

Quasar feedback may play a key role in the evolution of massive galaxies. The dust-reddened quasar, F2M110648.35$+$480712 at $z = 0.4352$ is one of the few cases at its redshift that exhibits powerful quasar feedback through bipolar outflows. Our new observation with the integral field unit mode of Near-infrared Spectrograph onboard JWST opens a new window to examine this spectacular outflow through Pa$\alpha$ emission line with $\sim$3$\times$ better spatial resolution than previous work. The morphology and kinematics of the Pa$\alpha$ nebula confirm the existence of a bipolar outflow extending on a scale of $\sim$17$\times$14 kpc and with a velocity reaching $\sim$1100 km s$^{-1}$. The higher spatial resolution of our new observation leads to more reliable measurements of outflow kinematics. Considering only the spatially resolved outflow and assuming an electron density of 100 cm$^{-2}$, the mass, momentum and kinetic energy outflow rates are $\sim$50-210 M$_{\odot}$ yr$^{-1}$, $\sim$0.3-1.7$\times$10$^{36}$ dynes ($\sim$14-78\% of the quasar photon momentum flux) and $\sim$0.16-1.27$\times$10$^{44}$ erg s$^{-1}$ ($\sim$0.02-0.20\% of the quasar bolometric luminosity), respectively. The local instantaneous outflow rates generally decrease radially. We infer that the quasar is powerful enough to drive the outflow, while stellar processes cannot be overlooked as a contributing energy source. The mass outflow rate is $\sim$0.4-1.5 times the star formation rate, and the ratio of kinetic energy outflow rate to the quasar bolometric luminosity is comparable to the minimum value required for negative quasar feedback in simulations. This outflow may help regulate the star formation activity within the system to some extent., Comment: 14 pages, 6 figures, 1 table, ApJ in review

While the activations of neurons in deep neural networks usually do not have a simple human-understandable interpretation, sparse autoencoders (SAEs) can be used to transform these activations into a higher-dimensional latent space which may be more easily interpretable. However, these SAEs can have millions of distinct latent features, making it infeasible for humans to manually interpret each one. In this work, we build an open-source automated pipeline to generate and evaluate natural language explanations for SAE features using LLMs. We test our framework on SAEs of varying sizes, activation functions, and losses, trained on two different open-weight LLMs. We introduce five new techniques to score the quality of explanations that are cheaper to run than the previous state of the art. One of these techniques, intervention scoring, evaluates the interpretability of the effects of intervening on a feature, which we find explains features that are not recalled by existing methods. We propose guidelines for generating better explanations that remain valid for a broader set of activating contexts, and discuss pitfalls with existing scoring techniques. We use our explanations to measure the semantic similarity of independently trained SAEs, and find that SAEs trained on nearby layers of the residual stream are highly similar. Our large-scale analysis confirms that SAE latents are indeed much more interpretable than neurons, even when neurons are sparsified using top-$k$ postprocessing. Our code is available at https://github.com/EleutherAI/sae-auto-interp, and our explanations are available at https://huggingface.co/datasets/EleutherAI/auto_interp_explanations.

Scalable oversight studies methods of training and evaluating AI systems in domains where human judgment is unreliable or expensive, such as scientific research and software engineering in complex codebases. Most work in this area has focused on methods of improving the quality of labels. Recent work by Burns et al. (2023) considers the complementary problem of training models with low-quality labels, finding that large pretrained models often have an inductive bias towards producing correct answers. In practice, however, neither label quantity nor quality is fixed: practitioners face a quantity-quality tradeoff. In this paper, we explore the microeconomics of the quantity-quality tradeoff on binary NLP classification tasks used in Burns et al. (2023). While sample-efficient learning has been studied extensively, little public research has focused on scalable elicitation: eliciting capabilities from pretrained models subject to labeling cost constraints. We find that this setting has novel dynamics caused by the tradeoff between label quantity and quality, as well as the model's existing latent capabilities. We observe three regimes of eliciting classification knowledge from pretrained models using supervised finetuning: quantity-dominant, quality-dominant, and a mixed regime involving the use of low- and high-quality data together to attain higher accuracy at a lower cost than using either alone. We explore sample-efficient elicitation methods that make use of two datasets of differing qualities, and establish a Pareto frontier of scalable elicitation methods that optimally trade off labeling cost and classifier performance. We find that the accuracy of supervised fine-tuning can be improved by up to 5 percentage points at a fixed labeling budget by adding a few-shot prompt to make use of the model's existing knowledge of the task.

In face of the climate emergency and growing challenges ranging from pollution to traffic jams, ride pooling has been floated as a potential solution for less congested, low-carbon and more space-efficient urban transport. However, it is unclear which system configurations enable an economically viable case for shared pooled mobility. To gain a better understanding of mechanisms behind this, we develop a simplified model to analyze the switching potential and CO2 emissions of ride pooling systems for a given number of transport users, street network topology and other system parameter values with different hypothetical switch rate functions. We find that CO2 emissions of local transport can be reduced by 39 to 45 % depending on the assumed switch rate function with other system parameters only having a secondary effect of a few percentage points. We call for empirically gauged analyses that translate our model into scenarios for metropolitan low-carbon and smart para-transit.

Simulating open quantum systems, which interact with external environments, presents significant challenges on noisy intermediate-scale quantum (NISQ) devices due to limited qubit resources and noise. In this paper, we propose an efficient framework for simulating open quantum systems on NISQ hardware by leveraging a time-perturbative Kraus operator representation of the system's dynamics. Our approach avoids the computationally expensive Trotterization method and exploits the Lindblad master equation to represent time evolution in a compact form, particularly for systems satisfying specific commutation relations. We demonstrate the efficiency of our method by simulating quantum channels, such as the continuous-time Pauli channel and damped harmonic oscillators, on NISQ trapped-ion hardware, including IonQ Harmony and Quantinuum H1-1. Additionally, we introduce hardware-agnostic error mitigation techniques, including Pauli channel fitting and quantum depolarizing channel inversion, to enhance the fidelity of quantum simulations. Our results show strong agreement between the simulations on real quantum hardware and exact solutions, highlighting the potential of Kraus-based methods for scalable and accurate simulation of open quantum systems on NISQ devices. This framework opens pathways for simulating more complex systems under realistic conditions in the near term., Comment: 18 pages, 12 figures

In a hierarchically formed Universe, galaxies accrete smaller systems that tidally disrupt as they evolve in the host's potential. We present a complete catalogue of disrupting galaxies accreted onto Milky Way-mass haloes from the Auriga suite of cosmological magnetohydrodynamic zoom-in simulations. We classify accretion events as intact satellites, stellar streams, or phase-mixed systems based on automated criteria calibrated to a visually classified sample, and match accretions to their counterparts in haloes re-simulated at higher resolution. Most satellites with a bound progenitor at the present day have lost substantial amounts of stellar mass -- 67 per cent have $f_\text{bound} < 0.97$ (our threshold to no longer be considered intact), while 53 per cent satisfy a more stringent $f_\text{bound} < 0.8$. Streams typically outnumber intact systems, contribute a smaller fraction of overall accreted stars, and are substantial contributors at intermediate distances from the host centre ($\sim$0.1 to $\sim$0.7$R_\text{200m}$, or $\sim$35 to $\sim$250 kpc for the Milky Way). We also identify accretion events that disrupt to form streams around massive intact satellites instead of the main host. Streams are more likely than intact or phase-mixed systems to have experienced preprocessing, suggesting this mechanism is important for setting disruption rates around Milky Way-mass haloes. All of these results are preserved across different simulation resolutions, though we do find some hints that satellites disrupt more readily at lower resolution. The Auriga haloes suggest that disrupting satellites surrounding Milky Way-mass galaxies are the norm and that a wealth of tidal features waits to be uncovered in upcoming surveys., Comment: 16+4 pages, 13+2 figures, 1+1 tables. Submitted to MNRAS, comments are welcome

Galaxies like the Milky Way are surrounded by complex populations of satellites at all stages of tidal disruption. In this paper, we present a dynamical study of the disrupting satellite galaxies in the Auriga simulations that are orbiting 28 distinct Milky Way-mass hosts across three resolutions. We find that the satellite galaxy populations are highly disrupted. The majority of satellites that remain fully intact at present day were accreted recently without experiencing more than one pericentre ($n_{\rm peri} \lesssim 1$) and have large apocentres ($r_{\rm apo} \gtrsim 200$ kpc) and pericentres ($r_{\rm peri} \gtrsim 50$ kpc). The remaining satellites have experienced significant tidal disruption and, given full knowledge of the system, would be classified as stellar streams. We find stellar streams in Auriga across the range of pericentres and apocentres of the known Milky Way dwarf galaxy streams and, interestingly, overlapping significantly with the Milky Way intact satellite population. We find no significant change in satellite orbital distributions across resolution. However, we do see substantial halo-to-halo variance of $(r_\text{peri}, r_\text{apo})$ distributions across host galaxies, as well as a dependence of satellite orbits on host halo mass - systems disrupt at larger pericentres and apocentres in more massive hosts. Our results suggest that either cosmological simulations (including, but not limited to, Auriga) are disrupting satellites far too readily, or that the Milky Way's satellites are more disrupted than current imaging surveys have revealed. Future observing facilities and careful mock observations of these systems will be key to revealing the nature of this apparent discrepancy., Comment: 16 pages, 10 figures, 1 table, submitted to MNRAS

Technological progress related to astronomical observatories such as the recently launched James Webb Space Telescope (JWST) allows searching for signs of life beyond our Solar System, namely in the form of unambiguous biosignature gases in exoplanetary atmospheres. The tentative assignment of a $1-2.4\sigma$ spectral feature observed with JWST in the atmosphere of exoplanet K2-18b to the biosignature gas dimethyl sulfide (DMS; sum formula C$_2$H$_6$S) raised hopes that, although controversial, a second genesis had been found. Terrestrial atmospheric DMS is exclusively stemming from marine biological activity and no natural abiotic source has been identified - neither on Earth nor in space. Therefore, DMS is considered a robust biosignature. Since comets possess a pristine inventory of complex organic molecules of abiotic origin, we have searched high-resolution mass spectra collected at comet 67P/Churyumov-Gerasimenko, target of the European Space Agency's Rosetta mission, for the signatures of DMS. Previous work reported the presence of a C2H6S signal when the comet was near its equinox but distinction of DMS from its structural isomer ethanethiol remained elusive. Here we reassess these and evaluate additional data. Based on differences in the electron ionization induced fragmentation pattern of the two isomers, we show that DMS is significantly better compatible with the observations. Deviations between expected and observed signal intensities for DMS are $<1\sigma$, while for ethanethiol they are $2-4\sigma$. The local abundance of DMS relative to methanol deduced from these data is (0.13$\pm$0.04)%. Our results provide the first evidence for the existence of an abiotic synthetic pathway to DMS in pristine cometary matter and hence motivate more detailed studies of the sulfur chemistry in such matter and its analogs. [...], Comment: 11 pages, 4 figures

In organic semiconductors charge transport typically takes place via slow hopping processes. Molecular aggregation can lead to enhanced exciton and charge transport through coupling of the transition dipole moments. In this work, we investigate the optical, morphological, and electronic properties of thin films of a merocyanine dye, that aggregates due to its high ground-state dipole moment. The degree of aggregation of spin-coated thin films can be easily tuned by thermal annealing. We demonstrate the relationship between charge carrier mobility and the degree of aggregation. The mobility is increased by approximately three orders of magnitude due to aggregation. We combine variable angle spectroscopic ellipsometry and polarization-resolved absorption spectroscopy with density functional theory to demonstrate that the aggregated molecules are oriented in an upright, standing configuration relative to the substrate surface. This arrangement involves a co-facial orientation of the molecular pi-systems which is advantageous for lateral charge transport. By utilizing highly oriented pyrolytic graphite as an ordered substrate, we are able to template the growth of the merocyanine layer in vapor phase deposition, and to improve the in-plane morphological order drastically. By correlating atomic force microscopy and photoluminescence microspectroscopy we observe oriented domains of 100s of {\mu}m^2 in size, emitting linearly polarized light, whereby maintaining the edge-on molecular arrangement. This promises a further significant enhancement of lateral charge carrier mobility.

Neutron scattering characterization of materials allows for the study of entanglement and microscopic structure, but is inefficient to simulate classically for comparison to theoretical models and predictions. However, quantum processors, notably analog quantum simulators, have the potential to offer an unprecedented, efficient method of Hamiltonian simulation by evolving a state in real time to compute phase transitions, dynamical properties, and entanglement witnesses. Here, we present a method for simulating neutron scattering on QuEra's Aquila processor by measuring the dynamic structure factor (DSF) for the prototypical example of the critical transverse field Ising chain, and propose a method for error mitigation. We provide numerical simulations and experimental results for the performance of the procedure on the hardware, up to a chain of length $L=25$. Additionally, the DSF result is used to compute the quantum Fisher information (QFI) density, where we confirm bipartite entanglement in the system experimentally., Comment: 13 pages, 10 figures

The exceptional magnetic, optical and phononic properties of Yttrium Iron Garnet (YIG) make it unique for spin-wave based and photonic applications. Yet, nanostructuring crystalline YIG and manipulating its magnetism in a non-destructive way is an outstanding challenge, and so far mostly limited to two-dimensional capabilities. Here, we show that irradiation of single-crystal YIG films with a focused UV laser drives a stable, giant enhancement of the perpendicular magnetic anisotropy, preserving the crystalline quality. This modulation is highly confined at the nanoscale in both the lateral and vertical directions, and its extension within the volume can be finely tuned with a continuous depth-control. By harnessing these three-dimensional anisotropy profiles, we demonstrate a large tuning of the spin-wave band structure, volume spatial localization, and non-reciprocity, realizing proof-of-principle 3D magnonic crystals. This straightforward, single-step, laser nanofabrication of three-dimensional magnetic systems based on crystalline YIG thin films opens the way to design novel functions in magnonic and magneto-optic devices., Comment: 17 pages, 5 figures

We explore the enhanced self-calibration of photometric galaxy redshift distributions, $n(z)$, through the combination of up to six two-point functions. Our $\rm 3\times2pt$ configuration is comprised of photometric shear, spectroscopic galaxy clustering, and spectroscopic-photometric galaxy-galaxy lensing (GGL). We further include spectroscopic-photometric cross-clustering; photometric GGL; and photometric auto-clustering, using the photometric shear sample as density tracer. We perform simulated likelihood forecasts of the cosmological and nuisance parameter constraints for Stage-III- and Stage-IV-like surveys. For the Stage-III-like case, we employ realistic but perturbed redshift distributions, and distinguish between "coherent" shifting in one direction, versus more internal scattering and full-shape errors. For perfectly known $n(z)$, a $\rm 6\times2pt$ analysis gains $\sim40\%$ in Figure of Merit (FoM) in the $S_8\equiv\sigma_8\sqrt{\Omega_{\rm m}/0.3}$ and $\Omega_{\rm m}$ plane relative to the $\rm 3\times2pt$ analysis. If untreated, coherent and incoherent redshift errors lead to inaccurate inferences of $S_8$ and $\Omega_{\rm m}$, respectively. Employing bin-wise scalar shifts $\delta{z}_i$ in the tomographic mean redshifts reduces cosmological parameter biases, with a $\rm 6x2pt$ analysis constraining the shift parameters with $2-4$ times the precision of a photometric $\rm 3^{ph}\times2pt$ analysis. For the Stage-IV-like survey, a $\rm 6\times2pt$ analysis doubles the FoM($\sigma_8{-}\Omega_{\rm m}$) compared to any $\rm 3\times2pt$ or $\rm 3^{ph}\times2pt$ analysis, and is only $8\%$ less constraining than if the $n(z)$ were perfectly known. A Gaussian mixture model for the $n(z)$ reduces mean-redshift errors and preserves the $n(z)$ shape. It also yields the most accurate and precise cosmological constraints for any $N\rm\times2pt$ configuration given $n(z)$ biases., Comment: 38 pages, 20 figures, to be submitted to A&A

We identify stable regions in the residual stream of Transformers, where the model's output remains insensitive to small activation changes, but exhibits high sensitivity at region boundaries. These regions emerge during training and become more defined as training progresses or model size increases. The regions appear to be much larger than previously studied polytopes. Our analysis suggests that these stable regions align with semantic distinctions, where similar prompts cluster within regions, and activations from the same region lead to similar next token predictions. This work provides a promising research direction for understanding the complexity of neural networks, shedding light on training dynamics, and advancing interpretability., Comment: Presented at the Scientific Methods for Understanding Deep Learning (SciForDL) workshop at NeurIPS 2024

We study the properties of the Luminous Red Galaxies (LRGs) selected from the 4th data release of the Kilo Degree Survey (KiDS-1000) via galaxy-galaxy lensing of the background galaxies from KiDS-1000. We use a halo model formalism to interpret our measurements and obtain estimates of the halo masses and the satellite fractions of the LRGs, resulting in halo masses $2.7 \times 10^{12} h^{-1} {\rm M}_{\odot}

Sparse Auto-Encoders (SAEs) are commonly employed in mechanistic interpretability to decompose the residual stream into monosemantic SAE latents. Recent work demonstrates that perturbing a model's activations at an early layer results in a step-function-like change in the model's final layer activations. Furthermore, the model's sensitivity to this perturbation differs between model-generated (real) activations and random activations. In our study, we assess model sensitivity in order to compare real activations to synthetic activations composed of SAE latents. Our findings indicate that synthetic activations closely resemble real activations when we control for the sparsity and cosine similarity of the constituent SAE latents. This suggests that real activations cannot be explained by a simple "bag of SAE latents" lacking internal structure, and instead suggests that SAE latents possess significant geometric and statistical properties. Notably, we observe that our synthetic activations exhibit less pronounced activation plateaus compared to those typically surrounding real activations., Comment: Presented at the Attributing Model Behavior at Scale (ATTRIB) workshop at NeurIPS 2024

In this work, we study how the abundance and dynamics of populations of disrupting satellite galaxies change systematically as a function of host galaxy properties. We apply a theoretical model of the phase-mixing process to classify intact satellite galaxies, stellar stream-like and shell-like debris in ~1500 Milky Way-mass systems generated by a semi-analytic galaxy formation code, SatGen. In particular, we test the effect of host galaxy halo mass, disk mass, ratio of disk scale height to length, and stellar feedback model on disrupting satellite populations. We find that the counts of tidal debris are consistent across all host galaxy models, within a given host mass range, and that all models can have stream-like debris on low-energy orbits, consistent with those observed around the Milky Way. However, we find a preference for stream-like debris on lower-energy orbits in models with a thicker (lower-density) host disk or on higher-energy orbits in models with a more-massive host disk. Importantly, we observe significant halo-to-halo variance across all models. These results highlight the importance of simulating and observing large samples of Milky Way-mass galaxies and accounting for variations in host properties when using disrupting satellites in studies of near-field cosmology., Comment: 18+6 pages, 10+2 figures, code publicly available on GitHub

Tests for racial bias commonly assess whether two people of different races are treated differently. A fundamental challenge is that, because two people may differ in many ways, factors besides race might explain differences in treatment. Here, we propose a test for bias which circumvents the difficulty of comparing two people by instead assessing whether the $\textit{same person}$ is treated differently when their race is perceived differently. We apply our method to test for bias in police traffic stops, finding that the same driver is likelier to be searched or arrested by police when they are perceived as Hispanic than when they are perceived as white. Our test is broadly applicable to other datasets where race, gender, or other identity data are perceived rather than self-reported, and the same person is observed multiple times.

We present a novel relaxation time approximation for kinetic theory with spin which takes into account the nonlocality of particle collisions. In particular, it models the property of the microscopic nonlocal collision term to vanish in global, but not in local equilibrium. We study the asymptotic distribution function obtained as the solution of the Boltzmann equation within the nonlocal relaxation time approximation in the limit of small gradients and short relaxation time. We show that the resulting polarization agrees with the one obtained from the Zubarev formalism for a certain value of a coefficient that determines the time scale on which orbital angular momentum is converted into spin. This coefficient can be identified with a parameter related to the pseudo gauge choice in the Zubarev formalism. Finally, we demonstrate how the nonlocal collision term generates polarization from vorticity by studying a nonrelativistic rotating cylinder both from kinetic and hydrodynamic approaches, which are shown to be equivalent in this example., Comment: 16 pages