Search

The conventional wisdom among educators and literacy gurus is that reading comprehension depends on the acquisition of isolatable, teachable, and generalizable skills. Consequently, many elementary and middle school English classrooms follow the "reading workshop" model, an approach to literacy instruction, with several variations that typically involve teachers spending a few minutes modeling a supposedly important skill before sending students off to practice by reading self-selected but appropriately "leveled" books. This policy brief challenges that orthodoxy. It asserts that, once students have learned to decode, reading books and other texts of any purported "level" with understanding depends more on knowledge than skills and that successful knowledge building requires explicit, carefully sequenced and paced, teacher-directed instruction across multiple subjects, including but not limited to social studies, science, and literature. Key questions asked in this report include: (1) Does reading comprehension depend on acquiring a set of teachable skills?; (2) Do students need practice with "just right" books?; (3) Does letting students choose the books they read foster the motivation necessary to improve reading comprehension?; and (4) Does extended literacy instruction enhance reading comprehension?

We present RUBIX, a fully tested, well-documented, and modular Open Source tool developed in JAX, designed to forward model IFU cubes of galaxies from cosmological hydrodynamical simulations. The code automatically parallelizes computations across multiple GPUs, demonstrating performance improvements over state-of-the-art codes by a factor of 600. This optimization reduces compute times from hours to only seconds. RUBIX leverages JAX's auto-differentiation capabilities to enable not only forward modeling but also gradient computations through the entire pipeline paving the way for new methodological approaches such as e.g. gradient-based optimization of astrophysics model parameters. RUBIX is open-source and available on GitHub: https://github.com/ufuk-cakir/rubix., Comment: accepted to the Machine Learning and the Physical Sciences Workshop, NeurIPS 2024

Radial metallicity gradients are fundamental to understanding galaxy formation and evolution. In our high-resolution simulation of a NIHAO-UHD Milky Way analogue, we analyze the linearity, scatter, spatial coherence, and age-related variations of metallicity gradients using young stars and gas. While a global linear model generally captures the gradient, it ever so slightly overestimates metallicity in the inner galaxy and underestimates it in the outer regions of our simulated galaxy. Both a quadratic model, showing an initially steeper gradient that smoothly flattens outward, and a piecewise linear model with a break radius at 10~kpc (2.5 effective radii) fit the data equally better. The spread of [Fe/H] of young stars in the simulation increases by tenfold from the innermost to the outer galaxy at a radius of 20~kpc. We find that stars born at similar times along radial spirals drive this spread in the outer galaxy, with a chemical under- and over-enhancement of up to 0.1 dex at leading and trailing regions of such spirals, respectively. This localised chemical variance highlights the need to examine radial and azimuthal selection effects for both Galactic and extragalactic observational studies. The arguably idealised but volume-complete simulations suggest that future studies should not only test linear and piecewise linear gradients, but also non-linear functions such as quadratic ones to test for a smooth gradient rather than one with a break radius. Either finding would help to determine the importance of different enrichment or mixing pathways and thus our understanding of galaxy formation and evolution scenarios., Comment: 18 pages, 20 figures, submitted to OJAp, comments welcome

As large language models (LLMs) become increasingly capable, it is prudent to assess whether safety measures remain effective even if LLMs intentionally try to bypass them. Previous work introduced control evaluations, an adversarial framework for testing deployment strategies of untrusted models (i.e., models which might be trying to bypass safety measures). While prior work treats a single failure as unacceptable, we perform control evaluations in a "distributed threat setting" -- a setting where no single action is catastrophic and no single action provides overwhelming evidence of misalignment. We approach this problem with a two-level deployment framework that uses an adaptive macro-protocol to choose between micro-protocols. Micro-protocols operate on a single task, using a less capable, but extensively tested (trusted) model to harness and monitor the untrusted model. Meanwhile, the macro-protocol maintains an adaptive credence on the untrusted model's alignment based on its past actions, using it to pick between safer and riskier micro-protocols. We evaluate our method in a code generation testbed where a red team attempts to generate subtly backdoored code with an LLM whose deployment is safeguarded by a blue team. We plot Pareto frontiers of safety (# of non-backdoored solutions) and usefulness (# of correct solutions). At a given level of usefulness, our adaptive deployment strategy reduces the number of backdoors by 80% compared to non-adaptive baselines.

Galaxies evolve hierarchically through merging with lower-mass systems and the remnants of destroyed galaxies are a key indicator of the past assembly history of our Galaxy. However, accurately measuring the properties of the accreted galaxies and hence unraveling the Milky Way's (MW) formation history is a challenging task. Here we introduce CASBI (Chemical Abundance Simulation Based Inference), a novel inference pipeline for Galactic Archeology based on Simulation-based Inference methods. CASBI leverages on the fact that there is a well defined mass-metallicity relation for galaxies and performs inference of key galaxy properties based on multi-dimensional chemical abundances of stars in the stellar halo. Hence, we recast the problem of unraveling the merger history of the MW into a SBI problem to recover the properties of the building blocks (e.g. total stellar mass and infall time) using the multi-dimensional chemical abundances of stars in the stellar halo as observable. With CASBI we are able to recover the full posterior probability of properties of building blocks of Milky Way like galaxies. We highlight CASBI's potential by inferring posteriors for the stellar masses of completely phase mixed dwarf galaxies solely from the 2d-distributions of stellar abundance in the iron vs. oxygen plane and find accurate and precise inference results.

The UK Research and Innovation Digital Research Infrastructure (DRI) needs to operate sustainably in the future, encompassing its use of energy and resources, and embedded computer hardware carbon emissions. Transition concepts towards less unsustainable operations will inform the future design and operations of DRI. A problem remains that, while the skills and knowledge for solving net zero challenges already exist within the UK's DRI community, the mechanisms for sharing them and enabling behavior change are missing. Without adopting community-driven approaches, individual stakeholders may feel isolated and uncertain about how to play their role in the transition. A research programme was funded to give voice to the ground-level stakeholders of the DRI ecosystem for the co-creation of carbon downshift concepts. This article presents the results of the programme, with the goal to inform a fair and just transition from the ground-level, complementing the top-down interventions of energy efficiency policies and renewable energies integration. A workshop-based innovation method was developed for researching stakeholder recommendations and perspectives on the sustainable transition of the UK's DRI. We find that giving a purposeful voice to the stakeholders for shaping their own future sustainable DRI environment can be achieved by a guided, expert-integrated, interactive and problem-focused workshop series. The chosen workshop design is impactful on creating bottom-up agency for climate action by first defining the high-level problems of unsustainability in energy and fossil-fuel consumption, and then connecting them to the ground-level circumstances of DRI stakeholders. This approach to stakeholder management should initiate a sustainable transition that promises to kick-start impactful changes from within communities, adding to high-level efforts from economics, policy, and governance., Comment: 18 pages, 1 figure

We present $\texttt{LAMINAR}$, a novel unsupervised machine learning pipeline designed to enhance the representation of structure within data via producing a more-informative distance metric. Analysis methods in the physical sciences often rely on standard metrics to define geometric relationships in data, which may fail to capture the underlying structure of complex data sets. $\texttt{LAMINAR}$ addresses this by using a continuous-normalising-flow and inverse-transform-sampling to define a Riemannian manifold in the data space without the need for the user to specify a metric over the data a-priori. The result is a locally-adaptive-metric that produces structurally-informative density-based distances. We demonstrate the utility of $\texttt{LAMINAR}$ by comparing its output to the Euclidean metric for structured data sets., Comment: Accepted to the NeurIPS 2024 Machine Learning and the Physical Sciences workshop. 6 pages, 6 figures

We demonstrate how the composition of two unsupervised clustering algorithms, $\texttt{AstroLink}$ and $\texttt{FuzzyCat}$, makes for a powerful tool when studying galaxy formation and evolution. $\texttt{AstroLink}$ is a general-purpose astrophysical clustering algorithm built for extracting meaningful hierarchical structure from point-cloud data defined over any feature space, while $\texttt{FuzzyCat}$ is a generalised soft-clustering algorithm that propagates the dynamical effects of underlying data processes into a fuzzy hierarchy of stable fuzzy clusters. Their composition, $\texttt{FuzzyCat}$ $\circ$ $\texttt{AstroLink}$, can therefore identify a fuzzy hierarchy of astrophysically- and statistically-significant fuzzy clusters within any point-based data set whose representation is subject to changes caused by some underlying process. Furthermore, the pipeline achieves this without relying upon strong assumptions about the data, the change process, the number/importance of specific structure types, or much user input -- thereby making itself applicable to a wide range of fields in the physical sciences. We find that for the task of structurally decomposing simulated galaxies into their constituents, our context-agnostic approach has a substantial impact on the diversity and completeness of the structures extracted as well as on their relationship within the broader galactic structural hierarchy -- revealing dwarf galaxies, infalling groups, stellar streams (and their progenitors), stellar shells, galactic bulges, and star-forming regions., Comment: Accepted to the NeurIPS 2024 Machine Learning and the Physical Sciences workshop. 9 pages, 2 figures

We present jf1uids, a one-dimensional fluid solver that can, by virtue of a geometric formulation of the Euler equations, model radially symmetric fluid problems in a conservative manner, i.e., without losing mass or energy. For spherical problems, such as ideal supernova explosions or stellar wind-blown bubble expansions, simulating only along a radial dimension drastically reduces compute and memory demands compared to a full three-dimensional method. This simplification also alleviates constraints on backpropagation through the solver. Written in JAX, jf1uids is a GPU-compatible and fully differentiable simulator. We demonstrate the advantages of this differentiable physics simulator by retrieving the wind's parameters for an adiabatic stellar wind expansion from the final fluid state using gradient descent. As part of a larger "stellar winds, cosmic rays and machine learning" research track, jf1uids serves as a solid foundation to be extended with additional physics modules, foremost cosmic rays and a neural-net powered gas-cooling surrogate and improved by higher order and more accurate numerical schemes. All code is available under https://github.com/leo1200/jf1uids/., Comment: 8 pages, 4 figures, accepted for the Machine Learning and the Physical Sciences workshop at NeurIPS 2024

We sketch how developers of frontier AI systems could construct a structured rationale -- a 'safety case' -- that an AI system is unlikely to cause catastrophic outcomes through scheming. Scheming is a potential threat model where AI systems could pursue misaligned goals covertly, hiding their true capabilities and objectives. In this report, we propose three arguments that safety cases could use in relation to scheming. For each argument we sketch how evidence could be gathered from empirical evaluations, and what assumptions would need to be met to provide strong assurance. First, developers of frontier AI systems could argue that AI systems are not capable of scheming (Scheming Inability). Second, one could argue that AI systems are not capable of posing harm through scheming (Harm Inability). Third, one could argue that control measures around the AI systems would prevent unacceptable outcomes even if the AI systems intentionally attempted to subvert them (Harm Control). Additionally, we discuss how safety cases might be supported by evidence that an AI system is reasonably aligned with its developers (Alignment). Finally, we point out that many of the assumptions required to make these safety arguments have not been confidently satisfied to date and require making progress on multiple open research problems.

We introduce CODES, a benchmark for comprehensive evaluation of surrogate architectures for coupled ODE systems. Besides standard metrics like mean squared error (MSE) and inference time, CODES provides insights into surrogate behaviour across multiple dimensions like interpolation, extrapolation, sparse data, uncertainty quantification and gradient correlation. The benchmark emphasizes usability through features such as integrated parallel training, a web-based configuration generator, and pre-implemented baseline models and datasets. Extensive documentation ensures sustainability and provides the foundation for collaborative improvement. By offering a fair and multi-faceted comparison, CODES helps researchers select the most suitable surrogate for their specific dataset and application while deepening our understanding of surrogate learning behaviour., Comment: 13 pages, 10 figures, accepted for the Machine Learning and the Physical Sciences workshop at NeurIPS 2024, source code available on GitHub at https://github.com/robin-janssen/CODES-Benchmark

Sufficiently capable models could subvert human oversight and decision-making in important contexts. For example, in the context of AI development, models could covertly sabotage efforts to evaluate their own dangerous capabilities, to monitor their behavior, or to make decisions about their deployment. We refer to this family of abilities as sabotage capabilities. We develop a set of related threat models and evaluations. These evaluations are designed to provide evidence that a given model, operating under a given set of mitigations, could not successfully sabotage a frontier model developer or other large organization's activities in any of these ways. We demonstrate these evaluations on Anthropic's Claude 3 Opus and Claude 3.5 Sonnet models. Our results suggest that for these models, minimal mitigations are currently sufficient to address sabotage risks, but that more realistic evaluations and stronger mitigations seem likely to be necessary soon as capabilities improve. We also survey related evaluations we tried and abandoned. Finally, we discuss the advantages of mitigation-aware capability evaluations, and of simulating large-scale deployments using small-scale statistics.

Recent works identified a way to recover the time evolution of a galaxy's disk metallicity gradient from the shape of its age--metallicity relation. However, the success of the method is dependent on how the width of the star-forming region evolves over time, which in turn is dependent on a galaxy's present-day bar strength. In this paper, we account for the time variation in the width of the star-forming region when deriving the interstellar medium (ISM) metallicity gradient evolution over time ($\rm \nabla [Fe/H](\tau)$), which provides more realistic birth radii estimates of Milky Way (MW) disk stars. Using MW/Andromeda analogues from the TNG50 simulation, we quantified the disk growth of newly born stars as a function of present-day bar strength to provide a correction that improves recovery of $\rm \nabla [Fe/H](\tau)$. In TNG50, we find that our correction reduces the median absolute error in recovering $\rm \nabla [Fe/H] (\tau)$ by over 30%. To confirm its universality, we test our correction on two galaxies from NIHAO-UHD and find the median absolute error is over 3 times smaller even in the presence of observational uncertainties for the barred, MW-like galaxy. Applying our correction to APOGEE DR17 red giant MW disk stars suggests the effects of merger events on $\rm \nabla [Fe/H](\tau)$ are less significant than originally found, and the corresponding estimated birth radii expose epochs when different migration mechanisms dominated. Our correction to account for the growth of the star-forming region in the disk allows for better recovery of the evolution of the MW disk's ISM metallicity gradient and, thus, more meaningful stellar birth radii estimates. With our results, we are able to suggest the evolution of not only the ISM gradient, but also the total stellar disk radial metallicity gradient, providing key constraints to select MW analogues across redshift., Comment: submitted to A&A

Recently, contrastive learning (CL), a technique most prominently used in natural language and computer vision, has been used to train informative representation spaces for galaxy spectra and images in a self-supervised manner. Following this idea, we implement CL for stars in the Milky Way, for which recent astronomical surveys have produced a huge amount of heterogeneous data. Specifically, we investigate Gaia XP coefficients and RVS spectra. Thus, the methods presented in this work lay the foundation for aggregating the knowledge implicitly contained in the multimodal data to enable downstream tasks like cross-modal generation or fused stellar parameter estimation. We find that CL results in a highly structured representation space that exhibits explicit physical meaning. Using this representation space to perform cross-modal generation and stellar label regression results in excellent performance with high-quality generated samples as well as accurate and precise label predictions., Comment: accepted to the Machine Learning and the Physical Sciences Workshop, NeurIPS 2024

Cosmological simulations are a powerful tool to advance our understanding of galaxy formation and many simulations model key properties of real galaxies. A question that naturally arises for such simulations in light of high-quality observational data is: How close are the models to reality? Due to the high-dimensionality of the problem, many previous studies evaluate galaxy simulations using simplified summary statistics of physical properties. In this work, we combine simulation-based Bayesian model comparison with a novel misspecification detection technique to compare simulated galaxy images of 6 hydrodynamical models observations. Since cosmological simulations are computationally costly, we address the problem of low simulation budgets by first training a $k$-sparse variational autoencoder (VAE) on the abundant dataset of SDSS images. The VAE learns to extract informative latent embeddings and delineates the typical set of real images. To reveal simulation gaps, we then perform out-of-distribution detection (OOD) based on the logits of classifiers trained on the embeddings of simulated images. Finally, we perform amortized Bayesian model comparison using probabilistic classification, identifying the relatively best-performing model along with partial explanations through SHAP values., Comment: accepted to the Machine Learning and the Physical Sciences workshop at Neurips 2024

Osteoarthritis (OA) is a prevalent degenerative joint disease, with the knee being the most commonly affected joint. Modern studies of knee joint injury and OA often measure biomechanical variables, particularly forces exerted during walking. However, the relationship among gait patterns, clinical profiles, and OA disease remains poorly understood. These biomechanical forces are typically represented as curves over time, but until recently, studies have relied on discrete values (or landmarks) to summarize these curves. This work aims to demonstrate the added value of analyzing full movement curves over conventional discrete summaries. Using data from the Intensive Diet and Exercise for Arthritis (IDEA) study (Messier et al., 2009, 2013), we developed a shape-based representation of variation in the full biomechanical curves. Compared to conventional discrete summaries, our approach yields more powerful predictors of disease progression and relevant clinical traits, as demonstrated by a nested model comparison. Notably, our work is among the first to use movement curves to predict disease progression and to quantitatively evaluate the added value of analyzing full movement curves over conventional discrete summaries

We analyse high-resolution ESPaDOnS/CFHT spectra of 20 very metal-poor stars ([Fe/H]~$<-2.0$) in the solar neighbourhood (within $\sim2$ kpc) selected to be on planar orbits (with a maximum height of $\lesssim4$ kpc). Targets include 11 prograde and 9 retrograde stars, spanning a wide range of eccentricities ($0.20-0.95$). Their chemical abundances are consistent with those observed in the Galactic halo but show a smaller spread, with no notable difference between progrades and retrogrades. This suggests a common chemical evolution and likely a shared formation site (except for one star). In this case, chemical evolution models indicate that the formation site would have had a baryonic mass of $\sim1.4\times10^9\msun$, similar to classical dwarf galaxies. High-energy supernovae and hypernovae are needed to reproduce the [X/Fe] up to the Fe-peak, while fast-rotating massive stars and neutron star merger events explain the [X/Fe] of the neutron-capture elements. The absence of Type Ia supernova signatures suggests a star formation duration of $\lesssim1$~Gyr. Cosmological zoom-in simulations support the scenario that an in-plane infall of a single system could disperse stars over a wide range of angular momenta during the early Galactic assembly. We propose that these stars originated in a proto-Galactic building block, which we name Loki. Less likely, if progrades and retrogrades formed in two different systems, their chemical evolution must have been very similar, with a combined baryonic mass twice that of a single system. Forthcoming surveys will provide a large and homogeneous dataset to investigate whether Loki is associated with any of the known detected structures. A comparison (primarily [$\alpha$/Fe]) with other VMPs moving in planar orbits suggests multiple systems contributed to the Galactic planar population, presenting some differences in their kinematical parameters., Comment: REference list updated; discussion on GSE updated

Understanding the morphology of galaxies is a critical aspect of astrophysics research, providing insight into the formation, evolution, and physical properties of these vast cosmic structures. Various observational and computational methods have been developed to quantify galaxy morphology, and with the advent of large galaxy simulations, the need for automated and effective classification methods has become increasingly important. This paper investigates the use of Principal Component Analysis (PCA) as an interpretable dimensionality reduction algorithm for galaxy morphology using the IllustrisTNG cosmological simulation dataset with the aim of developing a generative model for galaxies. We first generate a dataset of 2D images and 3D cubes of galaxies from the IllustrisTNG simulation, focusing on the mass, metallicity, and stellar age distribution of each galaxy. PCA is then applied to this data, transforming it into a lower-dimensional image space, where closeness of data points corresponds to morphological similarity. We find that PCA can effectively capture the key morphological features of galaxies, with a significant proportion of the variance in the data being explained by a small number of components. With our method we achieve a dimensionality reduction by a factor of $\sim200$ for 2D images and $\sim3650$ for 3D cubes at a reconstruction accuracy below five percent. Our results illustrate the potential of PCA in compressing large cosmological simulations into an interpretable generative model for galaxies that can easily be used in various downstream tasks such as galaxy classification and analysis., Comment: 13 pages, 9 without appendix, source code on GitHub: https://github.com/ufuk-cakir/MEGS , data on Zenodo: https://zenodo.org/record/8375344 , see also GAMMA: https://github.com/ufuk-cakir/GAMMA

To evaluate the safety and usefulness of deployment protocols for untrusted AIs, AI Control uses a red-teaming exercise played between a protocol designer and an adversary. This paper introduces AI-Control Games, a formal decision-making model of the red-teaming exercise as a multi-objective, partially observable, stochastic game. We also introduce methods for finding optimal protocols in AI-Control Games, by reducing them to a set of zero-sum partially observable stochastic games. We apply our formalism to model, evaluate and synthesise protocols for deploying untrusted language models as programming assistants, focusing on Trusted Monitoring protocols, which use weaker language models and limited human assistance. Finally, we demonstrate the utility of our formalism by showcasing improvements over empirical studies in existing settings, evaluating protocols in new settings, and analysing how modelling assumptions affect the safety and usefulness of protocols., Comment: 7 pages, with appendices

According to the well established hierarchical framework for galaxy evolution, galaxies grow through mergers with other galaxies and the LambdaCDM cosmological model predicts that the stellar halos of massive galaxies are rich in remnants from minor mergers. The Stellar Streams Legacy Survey (SSLS) has provided a first release of a catalogue with a statistically significant sample of stellar streams in the Local Universe, detected in deep images from DESI Legacy Surveys and the Dark Energy Survey (DES). The main objective is to compare the results of the observations of stellar tidal streams with predictions from state-of-the-art cosmological simulations regarding their abundance, up to a redshift z < 0.02, according to the LambdaCDM model. We use the predictions of the cosmological simulations Copernicus Complexio, TNG50 of the IllustrisTNG project, and Auriga to generate mock-images of nearby halos and search for stellar streams. We compare the stream frequency, characteristics and photometry in these images with DES observations. We find generally good agreement between the real images and the simulated ones regarding frequency, characteristics and photometry of the streams, while the stream morphology is somewhat different between observations and simulations, and between simulations themselves. By varying the sky background of the synthetic images to emulate different surface brightness limit levels, we also obtain predictions for the detection rate of stellar tidal streams up to a surface brightness limit of 35 mag arcsec^-2. The cosmological simulations predict that with an instrument such as the one used in the DES, it would be necessary to reach a surface brightness limit of 32 mag arcsec^-2 in the r-band to achieve a frequency of up to around 70% in the detection of stellar tidal streams around galaxies in the redshift range considered here., Comment: 19 pages, 23 figures

The transition energies of the two $1s$-core-excited soft X-ray lines (dubbed q and r) from $1s^2 2s ^1S_{1/2}$ to the respective upper levels $1s(^{2}S)2s2p(^{3}P) ^{2}P_{3/2}$ and $^{2}P_{1/2}$ of Li-like oxygen, fluorine and neon were measured and calibrated using several nearby transitions of He-like ions. The major remaining source of energy uncertainties in monochromators, the periodic fluctuations produced by imperfect angular encoder calibration, is addressed by a simultaneously running photoelectron spectroscopy measurement. This leads to an improved energy determination of 5 parts per million, showing fair agreement with previous theories as well as with our own, involving a complete treatment of the autoionizing states studied here. Our experimental results translate to an uncertainty of only 1.6\,km/s for the oxygen line qr-blend used to determine the outflow velocities of active galactic nuclei, ten times smaller than previously possible., Comment: 7 pages, 4 figures

The region of $N\approx Z\approx 40$ has long been associated with strongly deformed nuclear configurations. The presence of this strong deformation was recently confirmed through lifetime measurements in $N\approx Z$ Sr and Zr nuclei. Theoretically, however, these nuclei present a challenge due to the vast valence space required to incorporate all deformation driving interactions. Recent state-of-the-art predictions indicate a near axial prolate deformation for $N=Z$ and $N=Z+2$ nuclei between $N=Z=36$ and $N=Z=40$. In this work we investigate the shores of this island of deformation through a sub-barrier Coulomb excitation study of the $N=Z+4$ nucleus, \textsuperscript{80}Sr. Extracting a spectroscopic quadrupole moment of $Q_s(2^+_1) = 0.45^{+0.83}_{-0.88}$~eb, we find that \textsuperscript{80}Sr is inconsistent with significant axial prolate deformation. This indicates that the predicted region of strong prolate deformation around $N=Z=40$ is tightly constrained to the quartet of nuclei: \textsuperscript{76,78}Sr and \textsuperscript{78,80}Zr., Comment: Submitted to Physics Letters B

The CONUS+ experiment aims to detect coherent elastic neutrino-nucleus scattering (CEvNS) of reactor antineutrinos on germanium nuclei in the fully coherent regime, continuing on this way the CONUS physics program started at the Brokdorf nuclear power plant, Germany. The CONUS+ setup is installed in the nuclear power plant in Leibstadt, Switzerland, at a distance of 20.7 m from the 3.6 GW thermal power reactor core. The CEvNS signature will be measured with the same four point-contact high-purity germanium (HPGe) detectors produced for the former experiment, however refurbished and with optimized low energy thresholds. To suppress the background in the CONUS+ detectors, the passive and active layers of the original CONUS shield were modified such to fit better to the significantly changed background conditions at the new experimental location. New data acquisition and monitoring systems were developed. A direct network connection between the experiment and the Max-Planck-Institut fur Kernphysik (MPIK) makes it possible to control and monitor data acquisition in real time. The impact of all these modifications is discussed with particular emphasis on the resulting CEvNS signal prediction for the first data collection phase of CONUS+. Prospects of the planned upgrade in a second phase integrating new larger HPGe detectors are also discussed., Comment: 14 pages. 12 figures

In reinforcement learning, specification gaming occurs when AI systems learn undesired behaviors that are highly rewarded due to misspecified training goals. Specification gaming can range from simple behaviors like sycophancy to sophisticated and pernicious behaviors like reward-tampering, where a model directly modifies its own reward mechanism. However, these more pernicious behaviors may be too complex to be discovered via exploration. In this paper, we study whether Large Language Model (LLM) assistants which find easily discovered forms of specification gaming will generalize to perform rarer and more blatant forms, up to and including reward-tampering. We construct a curriculum of increasingly sophisticated gameable environments and find that training on early-curriculum environments leads to more specification gaming on remaining environments. Strikingly, a small but non-negligible proportion of the time, LLM assistants trained on the full curriculum generalize zero-shot to directly rewriting their own reward function. Retraining an LLM not to game early-curriculum environments mitigates, but does not eliminate, reward-tampering in later environments. Moreover, adding harmlessness training to our gameable environments does not prevent reward-tampering. These results demonstrate that LLMs can generalize from common forms of specification gaming to more pernicious reward tampering and that such behavior may be nontrivial to remove., Comment: Make it easier to find samples from the model, and highlight that our operational definition of reward tampering has false positives where the model attempts to complete the task honestly but edits the reward. Add paragraph to conclusion to this effect, and add sentence to figure 1 to this effect

In a previous work (arXiv:2306.05948), we constructed by convex integration examples of energy dissipating solutions to the 2D Euler equations on $\mathbb{R}^2$ with vorticity in the real Hardy space $H^p(\mathbb{R}^2)$. In the present paper, we develop tools that significantly improve that result in two ways: Firstly, we achieve vorticities in $H^p(\mathbb{R}^2)$ in the optimal range $p\in (0,1)$ compared to $(2/3,1)$ in our previous work. Secondly, the solutions constructed here possess compact support and in particular preserve linear and angular momenta., Comment: 37 pages. arXiv admin note: text overlap with arXiv:2306.05948

Energy is a primary constraint in processor design, and much of that energy is consumed in on-chip communication. Communication can be intra-core (e.g., from a register file to an ALU) or inter-core (e.g., over the on-chip network). In this paper, we use the on-chip network (OCN) as a case study for saving on-chip communication energy. We have identified a new way to reduce the OCN's link energy consumption by using line coding, a longstanding technique in information theory. Our line codes, called Low-Energy Line Codes (LELCs), reduce energy by reducing the frequency of voltage transitions of the links, and they achieve a range of energy/performance trade-offs.

We used the monochromatic soft-x-ray beamline P04 at the synchrotron-radiation facility PETRA III to resonantly excite the strongest $2p-3d$ transitions in neon-like Ni XIX ions, $[2p^6]_{J=0} \rightarrow [(2p^5)_{1/2}\,3d_{3/2}]_{J=1}$ and $[2p^6]_{J=0} \rightarrow [(2p^5)_{3/2}\,3d_{5/2}]_{J=1}$, respectively dubbed 3C and 3D, achieving a resolving power of 15\,000 and signal-to-background ratio of 30. We obtain their natural linewidths, with an accuracy of better than 10\%, as well as the oscillator-strength ratio $f(3C)/f(3D)$ = 2.51(11) from analysis of the resonant fluorescence spectra. These results agree with those of previous experiments, earlier predictions, and our own advanced calculations., Comment: 10 pages, 3 figures, 3 tables, published

Exploring the marks left by galactic accretion in the Milky Way helps us understand how our Galaxy was formed. However, finding and studying accreted stars and the galaxies they came from has been challenging. This study uses a simulation from the NIHAO project, which now includes a wider range of chemical compositions, to find better ways to spot these accreted stars. By comparing our findings with data from the GALAH spectroscopic survey, we confirm that the observationally established diagnostics of [Al/Fe] vs. [Mg/Mn] also show a separation of in-situ and accreted stars in the simulation, but stars from different accretion events tend to overlap in this plane even without observational uncertainties. Looking at the relationship between stellar age and linear or logarithmic abundances, such as [Fe/H], we can clearly separate different groups of these stars if the uncertainties in their chemical makeup are less than 0.15 dex and less than 20% for their ages. This method shows promise for studying the history of the Milky Way and other galaxies. Our work highlights how important it is to have accurate measurements of stellar ages and chemical content. It also shows how simulations can help us understand the complex process of galaxies merging and suggest how these events might relate to the differences we see between our Galaxy's thin and thick disk stars. This study provides a way to compare theoretical models with real observations, opening new paths for research in both our own Galaxy and beyond., Comment: 15 pages, 14 figures, submitted to MNRAS. Comments welcome. All code and data to reproduce the analysis and figures can be accessed via https://github.com/svenbuder/Accretion_ Clues_ObsSim

Chiral crystals and molecules were recently predicted to form an intriguing platform for unconventional orbital physics. Here, we report the observation of chirality-driven orbital textures in the bulk electronic structure of CoSi, a prototype member of the cubic B20 family of chiral crystals. Using circular dichroism in soft X-ray angle-resolved photoemission, we demonstrate the formation of a bulk orbital-angular-momentum texture and monopole-like orbital-momentum locking that depends on crystal handedness. We introduce the intrinsic chiral circular dichroism, icCD, as a differential photoemission observable and a natural probe of chiral electron states. Our findings render chiral crystals promising for spin-orbitronics applications., Comment: To be published in Physical Review Letters