Search

Dynamic and adaptive mesh refinement is pivotal in high-resolution, multi-physics, multi-model simulations, necessitating precise physics resolution in localized areas across expansive domains. Today's supercomputers' extreme heterogeneity presents a significant challenge for dynamically adaptive codes, highlighting the importance of achieving performance portability at scale. Our research focuses on astrophysical simulations, particularly stellar mergers, to elucidate early universe dynamics. We present Octo-Tiger, leveraging Kokkos, HPX, and SIMD for portable performance at scale in complex, massively parallel adaptive multi-physics simulations. Octo-Tiger supports diverse processors, accelerators, and network backends. Experiments demonstrate exceptional scalability across several heterogeneous supercomputers including Perlmutter, Frontier, and Fugaku, encompassing major GPU architectures and x86, ARM, and RISC-V CPUs. Parallel efficiency of 47.59% (110,080 cores and 6880 hybrid A100 GPUs) on a full-system run on Perlmutter (26% HPCG peak performance) and 51.37% (using 32,768 cores and 2,048 MI250X) on Frontier are achieved.

Most of the baryonic matter of the Universe resides in a highly-ionized gaseous intergalactic medium. This gas flows along dark-matter filaments toward galaxy superclusters, clusters, and groups until it pools around the galaxies into a circumgalactic medium. Eventually, the gas settles into the interstellar medium of the galaxies, where it fuels the successive generations of star formation that ultimately produce the stars and heavy elements that make up galaxies today. The gas has been studied for decades using absorption lines produced by Hydrogen and various ions of heavy elements in the spectra of background quasi-stellar objects (QSOs). But directly imaging the extremely faint glow of this "cosmic web" of intergalactic and circumgalactic gas has remained an elusive goal of observational cosmology. Some recent progress has been made by using imaging spectrographs to record high-redshift Ly$\alpha$ emission, although over only very narrow fields of view. Here we report direct images of intergalactic and circumgalactic gas in the distant Universe obtained using the Condor Array Telescope that we purposely built to reach extremely low-surface-brightness sensitivities over very wide fields of view. We show that these images directly detect and characterize the imprint of Ly$\alpha$ emission from the cosmic web at an overwhelming statistical significance. By stacking portions of the images centered on tens of thousands of galaxies of known redshift, we show that they also reveal extremely faint emission from H$^0$, C$^{3+}$, and Mg$^+$ and absorption from cosmic dust in the tenuous outskirts of the galaxies. Our results demonstrate that sensitive imaging observations can now detect and characterize emission (and absorption) from the cosmic web of intergalactic and circumgalactic gas (and dust)., Comment: 35 pages, 5 figure, 6 extended data figures

A century or less separates the thermonuclear-powered eruptions of recurrent novae in the hydrogen-rich envelopes of massive white dwarfs. The colliding ejecta of successive recurrent nova events are predicted to always generate very large (tens of parsecs) super-remnants; only two examples are currently known. T CrB offers an excellent opportunity to test this prediction. As it will almost certainly undergo its next, once-in ~80-year recurrent nova event between 2024 and 2026, we carried out very deep narrowband and continuum imaging to search for the predicted, piled-up ejecta of the past millenia. While nothing is detected in continuum or narrowband [OIII] images, a ~30-parsec-diameter, faint nebulosity surrounding T CrB is clearly present in deep Halpha, [NII] and [SII] narrowband Condor Array Telescope imagery. We predict that these newly detected nebulosities, as well as the recent ejecta that have not yet reached the super-remnant, are far too optically-thin to capture all but a tiny fraction of the photons emitted by RN flashes. We thus predict that fluorescent light echoes will NOT be detectable following the imminent nova flash of T CrB. Dust may be released by the T CrB red giant wind in pre-eruption outbursts, but we have no reliable estimates of its quantity or geometrical distribution. While we cannot predict the morphology or intensity of dust-induced continuum light echoes following the coming flash, we encourage multi-epoch Hubble Space Telescope optical imaging as well as James Webb Space Telescope infrared imaging of T CrB during the year after it erupts., Comment: 14 pages, 6 Figures, in press in The Astrophysical Journal Letters

The Vera C. Rubin Observatory is due to commence the 10-year Legacy Survey of Space and Time (LSST) at the end of 2025. To detect transient/variable sources and identify solar system objects (SSOs), the processing pipelines require templates of the static sky to perform difference imaging. During the first year of the LSST, templates must be generated as the survey progresses, otherwise SSOs cannot be discovered nightly. The incremental template generation strategy has not been finalized; therefore, we use the Metric Analysis Framework (MAF) and a simulation of the survey cadence (one_snap_v4.0_10yrs}) to explore template generation in Year 1. We have assessed the effects of generating templates over timescales of days-weeks, when at least four images of sufficient quality are available for $\geq90\%$ of the visit. We predict that SSO discoveries will begin $\sim$2-3 months after the start of the survey. We find that the ability of the LSST to discover SSOs in real-time is reduced in Year 1. This is especially true for detections in areas of the sky that receive fewer visits, such as the North Ecliptic Spur (NES), and in less commonly used filters, such as the $u$ and $g$-bands. The lack of templates in the NES dominates the loss of real-time SSO discoveries; across the whole sky the MAF Main-Belt asteroid (MBA) discovery metric decreases by up to $63\%$ compared to the baseline observing strategy, whereas the metric decreases by up to $79\%$ for MBAs in the NES alone., Comment: Submitted for consideration in the ApJS Rubin LSST Survey Strategy Optimization Focus Issue

The purpose of this paper is to explore properties of the whisker topology, which is a topology endowed on the fundamental group and whose utility is to detect locally complicated phenomena in pathological topological spaces. We show that the whisker topology preserves products, resolve an open question regarding the existence of a space which makes $\pi_1^{wh}(X,x_0)$ a non-discrete, non-abelian, and Hausdorff topological group, and show the whisker topology is not separable on the earring group $\pi_1(\Er^1,x_0)$.

We used the Condor Array Telescope to obtain deep imaging observations through luminance broad-band and He II, [O III], He I, H$\alpha$, [N II], and [S II] narrow-band filters of an extended region of the M81 Group spanning $\approx 8 \times 8$ deg$^2$ on the sky centered near M81 and M82. Here we report aspects of these observations that are specifically related to (1) a remarkable filament known as the "Ursa Major Arc" that stretches $\approx 30$ deg on the sky roughly in the direction of Ursa Major, (2) a "Giant Shell of Ionized Gas" that stretches $\approx 0.8$ deg on the sky located $\approx 0.6$ deg NW of M82, and (3) a remarkable network of ionized gaseous filaments revealed by the new Condor observations that appear to connect the arc, the shell, and various of the galaxies of the M81 Group and, by extension, the group itself. We measure flux ratios between the various ions to help to distinguish photoionized from shock-ionized gas, and we find that the flux ratios of the arc and shell are not indicative of shock ionization. This provides strong evidence against a previous interpretation of the arc as an interstellar shock produced by an unrecognized supernova. We suggest that all of these objects, including the arc, are associated with the M81 Group and are located at roughly the distance $\approx 3.6$ Mpc of M81, that the arc is an intergalactic filament, and that the objects are associated with the low-redshift cosmic web., Comment: 8 pages, 4 figures

We used the Condor Array Telescope to obtain deep imaging observations through the luminance broad-band and He II 468.6 nm, [O III] 500.7 nm, He I 587.5 nm, H$\alpha$, [N II] 658.4 nm, and [S II] 671.6 nm narrow-band filters of an extended region comprising 13 "Condor fields" spanning $\approx 8 \times 8$ deg$^2$ on the sky centered near M81 and M82. Here we describe the acquisition and processing of these observations, which together constitute unique very deep imaging observations of a large portion of the M81 Group through a complement of broad- and narrow-band filters. The images are characterized by an intricate web of faint, diffuse, continuum produced by starlight scattered from Galactic cirrus, and all prominent cirrus features identified in the broad-band image can also be identified in the narrow-band images. We subtracted the luminance image from the narrow-band images to leave more or less only line emission in the difference images, and we masked regions of the resulting images around stars at an isophotal limit. The difference images exhibit extensive extended structures of ionized gas in the direction of the M81 Group, from known galaxies of the M81 Group, clouds of gas, filamentary structures, and apparent or possible bubbles or shells. Specifically, the difference images show a remarkable filament known as the "Ursa Major Arc;" a remarkable network of criss-crossed filaments between M81 and NGC 2976, some of which intersect and overlap the Ursa Major Arc; and details of a "giant shell of ionized gas.", Comment: 15 pages, 8 figures

Classical nova eruptions result from thermonuclear-powered runaways in, and ejection of, the hydrogen-rich envelopes of white dwarf stars accreted from their close binary companions. Novae brighten to up to 1,000,000 solar luminosities, and recur thousands of times over their lifetimes spanning several billion years. Between eruptions, mass transfer from the donor star to the white dwarf proceeds via an accretion disk unless the white dwarf possesses a strong magnetic field which can partially or totally disrupt the disk. In that case, accretion is focussed by the white dwarf's magnetic field towards its magnetic poles. Optical spectroscopy and interferometric radio maps demonstrate the presence of bipolar jets, typically arcsec in angular size, and orders of magnitude smaller than one parsec in linear size, during the days to months after nova eruptions. These jets expel collimated matter from the white dwarfs in nova binary stars, but well-resolved images of them are lacking. Here we report the Condor telescope's detection of a hitherto unknown, highly resolved and braided jet, three degrees (at least 25 parsecs) in length. The jet originates at the white dwarf of the old nova GK Persei (nova Per 1901 CE). It precesses on a ~ 3600 yr timescale, and must be at least 7200 years old. Detected across four decades of wavelength, the jet's ultimate energy source is likely the strong accretion shocks near the white dwarf's magnetic poles., Comment: 18 pages, 4 figures

Detecting or placing upper limits on spacetime variations of fundamental constants requires quantifying every potential source of uncertainty. We continue our previous study into the impact of continuum variations on measurements of the fine structure constant, here in the context of quasar absorption systems. An automated (hence objective and reproducible) continuum modelling method is reported in an accompanying paper. We apply the method to the $z_{abs}=1.7975$ absorption system towards the quasar PHL957. Multiple continuum fits are generated, and for each, we derive independent models of the system, each giving its own measurement of the fine structure constant $\alpha$. This process isolates and quantifies the error contribution associated with continuum placement uncertainty. This source of uncertainty, ignored in many previous measurements, arises in two ways: (i) slight local continuum tilt uncertainty generates small line shifts, and (ii) different continuum estimates produce slightly different kinematic structures in the absorption system model. Taking continuum placement uncertainty into account, the new PHL957 measurement we obtain is $\Delta\alpha/\alpha= -0.53^{+5.45}_{-5.51} \times 10^{-6}$. This measurement assumes terrestrial magnesium isotopic abundances. Recommendations are provided for future $\alpha$ measurements. Finally, we also note the potential importance of the effects identified here for future redshift drift experiments., Comment: 10 pages, 3 figures. Submitted to Open Journal of Astrophysics

Searches for variations of fundamental constants require a comprehensive understanding of measurement errors. This paper examines a source of error that is usually overlooked: the impact of continuum placement error. We investigate the problem using a high resolution, high signal to noise spectrum of the white dwarf G191$-$B2B. Narrow photospheric absorption lines allow us to search for new physics in the presence of a gravitational field approximately $10^4$ times that on Earth. Modelling photospheric lines requires knowing the underlying spectral continuum level. We describe the development of a fully automated, objective, and reproducible continuum estimation method. Measurements of the fine structure constant are produced using several continuum models. The results show that continuum placement variations result in small systematic shifts in the centroids of narrow photospheric absorption lines which impact significantly on fine structure constant measurements. This effect should therefore be included in the error budgets of future measurements. Our results suggest that continuum placement variations should be investigated in other contexts, including fine structure constant measurements in stars other than white dwarfs. The analysis presented here is based on NiV absorption lines in the photosphere of G191$-$B2B. Curiously, the inferred measurement of the fine structure constant obtained in this paper using NiV (the least negative of our measurements is $\Delta\alpha/\alpha = -1.462 \pm 1.121 \times 10^{-5}$) is inconsistent with the most recent previous G191$-$B2B photospheric measurement using FeV ($\Delta\alpha/\alpha = 6.36 \pm 0.35_{stat} \pm 1.84_{sys} \times 10^{-5}$). Given both measurements are derived from the same spectrum, we presume (but in this work are unable to check) that this 3.2$\sigma$ difference results from unknown laboratory wavelength systematics., Comment: Main paper comprises 11 pages, 4 figures, 1 table. 41 pages in total including appendices. Submitted to Open Journal of Astrophysics. arXiv admin note: substantial text overlap with arXiv:2212.00434

Deep neural networks have achieved impressive performance on many computer vision benchmarks in recent years. However, can we be confident that impressive performance on benchmarks will translate to strong performance in real-world environments? Many environments in the real world are safety critical, and even slight model failures can be catastrophic. Therefore, it is crucial to test models rigorously before deployment. We argue, through both statistical theory and empirical evidence, that selecting representative image datasets for testing a model is likely implausible in many domains. Furthermore, performance statistics calculated with non-representative image datasets are highly unreliable. As a consequence, we cannot guarantee that models which perform well on withheld test images will also perform well in the real world. Creating larger and larger datasets will not help, and bias aware datasets cannot solve this problem either. Ultimately, there is little statistical foundation for evaluating models using withheld test sets. We recommend that future evaluation methodologies focus on assessing a model's decision-making process, rather than metrics such as accuracy., Comment: 13 pages

Chemical evolution models predict a gradual build-up of $^{13}$C in the universe, based on empirical nuclear reaction rates and assumptions on the properties of stellar populations. However, old metal-poor stars within the Galaxy contain more $^{13}$C than is predicted, suggesting that further refinements to the models are necessary. Gas at high redshift provides important supplementary information at metallicities $-2\lesssim$ [Fe/H] $\lesssim-1$, for which there are only a few measurements in the Galaxy. We obtained new, high-quality, VLT/ESPRESSO observations of the QSO B1331+170 and used them to measure $^{12}$C/$^{13}$C in the damped Lyman-$\alpha$ system (DLA) at $z_{abs}=1.776$, with [Fe/H]=-1.27. AI-VPFIT, an Artificial Intelligence tool based on genetic algorithms and guided by a spectroscopic information criterion, was used to explore different possible kinematic structures of the carbon gas. Three hundred independent AI-VPFIT models of the absorption system were produced using pre-set $^{12}$C/$^{13}$C values, ranging from 4 to 500. Our results show that $^{12}$C/$^{13}$C$=28.5^{+51.5}_{-10.4}$, suggesting a possibility of $^{13}$C production at low metallicity., Comment: 11 pages, 6 figures. Three appendices. To appear in MNRAS

Several recent high-impact studies leverage large hospital-owned electrocardiographic (ECG) databases to model and predict patient mortality. MIMIC-IV, released September 2023, is the first comparable public dataset and includes 800,000 ECGs from a U.S. hospital system. Previously, the largest public ECG dataset was Code-15, containing 345,000 ECGs collected during routine care in Brazil. These datasets now provide an excellent resource for a broader audience to explore ECG survival modeling. Here, we benchmark survival model performance on Code-15 and MIMIC-IV with two neural network architectures, compare four deep survival modeling approaches to Cox regressions trained on classifier outputs, and evaluate performance at one to ten years. Our results yield AUROC and concordance scores comparable to past work (circa 0.8) and reasonable AUPRC scores (MIMIC-IV: 0.4-0.5, Code-15: 0.05-0.13) considering the fraction of ECG samples linked to a mortality (MIMIC-IV: 27\%, Code-15: 4\%). When evaluating models on the opposite dataset, AUROC and concordance values drop by 0.1-0.15, which may be due to cohort differences. All code and results are made public., Comment: 9 pages plus appendix, 2 figures

This note explains the equivalence between approximate rationalizability and approximate cost-rationalizability within the context of consumer demand. In connection with these results, we interpret Afriat's (1973) critical cost efficiency index (CCEI) as a measure of approximate rationalizability through cost inefficiency, in the sense that an agent is spending more money than is required to achieve her utility targets.

A goodness-of-fit index measures the consistency of consumption data with a given model of utility-maximization. We show that for the class of well-behaved (i.e., continuous and increasing) utility functions there is no goodness-of-fit index that is continuous and accurate, where the latter means that a perfect score is obtained if and only if a dataset can be rationalized by a well-behaved utility function. While many standard goodness-of-fit indices are inaccurate we show that these indices are (in a sense we make precise) essentially accurate. Goodness-of-fit indices are typically generated by loss functions and we find that standard loss functions usually do not yield a best-fitting utility function when they are minimized. Nonetheless, welfare comparisons can be made by working out a robust preference relation from the data.

Afriat's Theorem (1967) states that a dataset can be thought of as being generated by a consumer maximizing a continuous and increasing utility function if and only if it is free of revealed preference cycles containing a strict relation. The latter property is often known by its acronym, GARP (for generalized axiom of revealed preference). This paper surveys extensions and applications of Afriat's seminal result. We focus on those results where the consistency of a dataset with the maximization of a utility function satisfying some property can be characterized by a suitably modified version of GARP.

Accurate prediction of drivers' gaze is an important component of vision-based driver monitoring and assistive systems. Of particular interest are safety-critical episodes, such as performing maneuvers or crossing intersections. In such scenarios, drivers' gaze distribution changes significantly and becomes difficult to predict, especially if the task and context information is represented implicitly, as is common in many state-of-the-art models. However, explicit modeling of top-down factors affecting drivers' attention often requires additional information and annotations that may not be readily available. In this paper, we address the challenge of effective modeling of task and context with common sources of data for use in practical systems. To this end, we introduce SCOUT+, a task- and context-aware model for drivers' gaze prediction, which leverages route and map information inferred from commonly available GPS data. We evaluate our model on two datasets, DR(eye)VE and BDD-A, and demonstrate that using maps improves results compared to bottom-up models and reaches performance comparable to the top-down model SCOUT which relies on privileged ground truth information. Code is available at https://github.com/ykotseruba/SCOUT., Comment: Accepted at IEEE Intelligent Vehicles Symposium (IV), 2024