arXiv subjects

David Katz

Publications and source records attributed to David Katz.

At least 19 recordsLinked to original sources

arXiv2026

Radial velocity and atmospheric parameter calculations for the GaiaNIR spectrograph

Context. The upcoming GaiaNIR mission is currently planning to add a near-infrared spectrograph to its payload in order to enhance its scientific return, particularly for mapping the dust-obscured regions of the Milky Way. Aims. This study aims to identify the optimal wavelength region between 800 and 2300 nm for the proposed GaiaNIR spectrograph to maximize the precision of radial velocities and atmospheric parameters. Methods. To find its spectral range, we generated 10000 synthetic spectra from the BOSZ library across a wide range of stellar parameters, with resolutions varying from 5000 to 20000. By cross-correlating these mock observations with ideal templates, we assessed the statistical scatter of velocity residuals to isolate six candidate windows for further atmospheric parameter testing. Results. Our analysis finds that the 1926 - 1968 nm window at R = 16100 - 20100 in the K-band is the preferred strategic choice, because it has the potential to reach radial velocity precision for the brightest FGKM stars of about 160 - 260 m/s depending on resolution, while providing precision of the atmospheric parameters close to Gaia's Radial Velocity Spectrometer. We also identified a second region between 1158 and 1202 nm (R = 9300 - 11600), that has slightly lower radial velocity precision, but at a wider temperature range than the K-band. Both regions make it possible to derive abundances of ten species at these resolutions: O, Na, Mg, Si, Ca, Ti, V, Cr, Mn, Ni in the K-band, and Mg, Si, K, Ca, Ti, V, Cr, Mn, Co, Ni between 1158 - 1202 nm. Conclusions. The K-band delivers sufficiently precise measurements for the mission's primary cool star targets while taking advantage of significantly lower interstellar extinction, enabling the mapping of the dust-obscured regions of the Milky Way.

astro-ph.IM↗

arXiv2026

Automated structural testing of LLM-based agents: methods, framework, and case studies

LLM-based agents are rapidly being adopted across diverse domains. Since they interact with users without supervision, they must be tested extensively. Current testing approaches focus on acceptance-level evaluation from the user's perspective. While intuitive, these tests require manual evaluation, are difficult to automate, do not facilitate root cause analysis, and incur expensive test environments. In this paper, we present methods to enable structural testing of LLM-based agents. Our approach utilizes traces (based on OpenTelemetry) to capture agent trajectories, employs mocking to enforce reproducible LLM behavior, and adds assertions to automate test verification. This enables testing agent components and interactions at a deeper technical level within automated workflows. We demonstrate how structural testing enables the adaptation of software engineering best practices to agents, including the test automation pyramid, regression testing, test-driven development, and multi-language testing. In representative case studies, we demonstrate automated execution and faster root-cause analysis. Collectively, these methods reduce testing costs and improve agent quality through higher coverage, reusability, and earlier defect detection. We provide an open source reference implementation on GitHub.

cs.SE↗

arXiv2026

An MLIR Lowering Pipeline for Stencils at Wafer-Scale

The Cerebras Wafer-Scale Engine (WSE) delivers performance at an unprecedented scale of over 900,000 compute units, all connected via a single-wafer on-chip interconnect. Initially designed for AI, the WSE architecture is also well-suited for High Performance Computing (HPC). However, its distributed asynchronous programming model diverges significantly from the simple sequential or bulk-synchronous programs that one would typically derive for a given mathematical program description. Targeting the WSE requires a bespoke re-implementation when porting existing code. The absence of WSE support in compilers such as MLIR, meant that there was little hope for automating this process. Stencils are ubiquitous in HPC, and in this paper we explore the hypothesis that domain specific information about stencils can be leveraged by the compiler to automatically target the WSE without requiring application-level code changes. We present a compiler pipeline that transforms stencil-based kernels into highly optimized CSL code for the WSE, bridging the semantic gap between the mathematical representation of the problem and the WSE's asynchronous execution model. Based upon five benchmarks across three HPC programming technologies, running on both the Cerebras WSE2 and WSE3, our approach delivers comparable, if not slightly better, performance than manually optimized code. Furthermore, without requiring any application level code changes, performance on the WSE3 is around 14 times faster than 128 Nvidia A100 GPUs and 20 times faster than 128 nodes of a CPU-based Cray-EX supercomputer when using our approach.

cs.DC↗

arXiv2025

A new Gaia census of OB associations within 1 kpc

OB associations are primordial tracers of star formation and Galactic structure. Originally defined about 80 years ago, their historical membership lists have been superseded thanks to the precise astrometry from ESA's \textit{Gaia}'s satellite. Recent studies have however been mostly focused on individual OB associations or limited by the coverage of spectroscopic surveys. In this paper, we exploit a complete census of $\sim$25,000 O- and B-type stars within 1 kpc of the Sun to produce a highly-reliable catalogue of 56 OB associations using the HDBSCAN clustering algorithm, increasing the number of known OB associations by a factor of two within this volume. We assess the validity of this catalogue by crossmatching our OB association members with other catalogues of OB associations, star clusters and young stellar groups, confirming the high-confidence of our census of OB associations. We characterize these OB associations physically (total initial stellar mass, number of OB stars, ...) and kinematically (velocity dispersion, linear expansion ages, ...). The majority of the OB associations (38 out of 56) exhibit a significant expansion pattern in at least one direction, including 12 in both plane-of-the-sky directions, though differences in expansion velocity suggest anisotropical expansion patterns. We compare the locations of these OB associations with superclouds and features in the local Milky Way such as the Radcliffe Wave and discuss the implications for star formation in the solar neighbourhood.

astro-ph.GA↗

arXiv2025

An MLIR pipeline for offloading Fortran to FPGAs via OpenMP

With the slowing of Moore's Law, heterogeneous computing platforms such as Field Programmable Gate Arrays (FPGAs) have gained increasing interest for accelerating HPC workloads. In this work we present, to the best of our knowledge, the first implementation of selective code offloading to FPGAs via the OpenMP target directive within MLIR. Our approach combines the MLIR OpenMP dialect with a High-Level Synthesis (HLS) dialect to provide a portable compilation flow targeting FPGAs. Unlike prior OpenMP FPGA efforts that rely on custom compilers, by contrast we integrate with MLIR and so support any MLIR-compatible front end, demonstrated here with Flang. Building upon a range of existing MLIR building blocks significantly reduces the effort required and demonstrates the composability benefits of the MLIR ecosystem. Our approach supports manual optimisation of offloaded kernels through standard OpenMP directives, and this work establishes a flexible and extensible path for directive-based FPGA acceleration integrated within the MLIR ecosystem.

cs.DC↗

arXiv2025

The Gaia RVS Benchmark Stars. III . Confirmed and New High Radial Velocity Stars From Gaia DR3

High-velocity stars are interesting targets to unveil the formation of the Milky Way. In fact they can be recently accreted from an infalling dwarf galaxies or they can be the result of a turbulent merging of galaxies. Gaia is providing the community a way to select stars for their kinematics and the radial velocity, one of the speed components, is derived from the Gaia RVS spectrum. High absolute radial velocity values are sensitive to be false positive. They are rare and as such they are more easily impacted than lower velocity stars, by contamination by very low SNR spurious measures. We here investigate a sample of 26 stars with Gaia absolute radial velocity in excess of 500___km/s with spectroscopic follow-up observations with UVES. For all but one star the extreme radial velocity is confirmed and these stars are all metal-poor and, as expected, enhanced in the elements. The complete chemical inventory confirm the large star-to-star scatter for the heavy elements and their on average larger ratio over iron with respect to the Sun, in agreement with the literature.

astro-ph.GA↗

arXiv2025

Young stars discovered in dwarf spheroidal galaxies confirm their recent infall into the Milky way

Recent observations from the ESA Gaia satellite and with the ESO VLT, have identified the presence of a population of young, 0.5 to 2 Gyr old, stars in the halo and in dwarf spheroidal galaxies surrounding the Milky Way. It suggests that MW dwarf galaxies, currently devoid of gas, had, until recent times, enough gas to sustain a burst of star formation. The recent loss of gas coincides with their arrival in the vicinity of the Milky Way, in agreement with orbital predictions from Gaia that indicate that most dwarf galaxies reached the Milky Way halo less than 3 Gyr years ago. This completely changes the interpretation of their dynamics, mass, and dark matter content.

astro-ph.GA↗

arXiv2025

Quadrupole signature as a kinematic diagnostic to constrain bar properties : implications for the Milky Way

The presence of a 'butterfly' or a quadrupole structure in the stellar mean radial velocity ($ $) field of the Milky Way is well known from the Gaia and the APOGEE surveys. Past studies indicated that a stellar bar can excite such a quadrupole feature in the $< V_R >$ distribution. However, a systematic study investigating the co-evolution of bar and quadrupole structure is largely missing. Furthermore, whether this quadrupole structure in $ $ can be used as a robust kinematic diagnostic to constrain bar properties, particularly for the Milky Way, is still beyond our grasp. Here, we investigate the bar-induced quadrupole feature using a suite of isolated $N$-body models forming prominent bars and a sample of Milky Way-like barred galaxies from the TNG50 cosmological simulation. We demonstrate that the properties of the quadrupole (strength, length, and orientation) are strongly correlated with the bar properties, regardless of the choice of the thin/thick disc stars; thereby making the quadrupole feature an excellent kinematic diagnostic for constraining the bar properties. In presence of spirals, the estimator which takes into account the phase-angle of $m = 4$ Fourier moment, serves as a more appropriate estimator for measuring the length of the quadrupole. Further, we constructed a novel Gaia-like mock dataset from a simulated bar model while incorporating the dust extinction and the broad trends of observational errors of the Gaia survey. The quadrupole properties (strength and length) estimated from those Gaia-like mock data are larger ($\sim 35-45$ percent) when compared with their true values. We showed that the majority of this effect is due to the uncertainty in parallax measurement. This demonstrates that the quadrupole structure in Gaia data is likely a result of dominant Gaia parallax errors/biases, almost masking the true inherent signature of the MW bar.

astro-ph.GA↗

arXiv2024

Generative AI Toolkit -- a framework for increasing the quality of LLM-based applications over their whole life cycle

As LLM-based applications reach millions of customers, ensuring their scalability and continuous quality improvement is critical for success. However, the current workflows for developing, maintaining, and operating (DevOps) these applications are predominantly manual, slow, and based on trial-and-error. With this paper we introduce the Generative AI Toolkit, which automates essential workflows over the whole life cycle of LLM-based applications. The toolkit helps to configure, test, continuously monitor and optimize Generative AI applications such as agents, thus significantly improving quality while shortening release cycles. We showcase the effectiveness of our toolkit on representative use cases, share best practices, and outline future enhancements. Since we are convinced that our Generative AI Toolkit is helpful for other teams, we are open sourcing it on and hope that others will use, forward, adapt and improve

cs.SE↗

arXiv2024

The disc origin of the Milky Way bulge: On the high velocity dispersion of metal-rich stars at low latitude

Previous studies of the chemo-kinematic properties of stars in the Galactic bulge have revealed a puzzling trend. Along the bulge minor axis, and close to the Galactic plane, metal-rich stars display a higher line-of-sight velocity dispersion compared to metal-poor stars, while at higher latitudes metal-rich stars have lower velocity dispersions than metal-poor stars, similar to what is found in the Galactic disc. In this work, we re-examine this issue, by studying the dependence of line-of-sight velocity dispersions on metallicity and latitude in the latest APOGEE Data Release 17, confirming the results of previous works. We then analyse an N-body simulation of a Milky Way-like galaxy, also taking into account observational biases introduced by the APOGEE selection function. We show that the inversion in the line-of-sight velocity dispersion-latitude relation observed in the Galactic bulge can be reproduced by our model. We show that this inversion is a natural consequence of a scenario in which the bulge is a boxy/peanut-shaped structure, whose metal-rich and metal-poor stars mainly originate from the thin and thick disc of the Milky Way, respectively. Due to their cold kinematics, metal-rich, thin disc stars, are efficiently trapped in the boxy/peanut bulge, and, at low latitudes, show a strong barred morphology, which results in high velocity dispersions which are larger than those attained by the metal-poor populations. Extremely metal-rich stars in the Galactic bulge, which have received renewed attention in the literature, do follow the same trends as those of the metal-rich populations. The line-of-sight velocity-latitude relation observed in the Galactic bulge for metal-poor and metal-rich stars are thus both an effect of the intrinsic nature of the Galactic bulge and of the angle at which we observe it from the Sun.

astro-ph.GA↗

arXiv2024

Timing the Milky Way bar formation and the accompanying radial migration episode

We derive the metallicity profile of the Milky Way low-$\alpha$ disc population from 2 to 20 kpc from the Galactic centre in 1 Gyr age bins using the astroNN catalogue, and show that it is highly structured, with a plateau between 4 and 7 kpc and a break at 10-12 kpc. We argue that these features result from the two main bar resonances, the corotation and the Outer Lindblad Resonance (OLR), respectively. We show that the break in the metallicity profile is most visible in stars having 7-8 Gyr, reaching an amplitude of about 0.4 dex, and is the signpost of the position of the bar OLR. The bar formation was accompanied by an episode of radial migration triggered by its slowing down and is responsible for spreading old metal-rich stars up to the OLR. The data show that the slowdown of the bar ended 6-7 Gyr ago. Based on numerical simulations that reproduce well the break observed in the metallicity profile, we argue that this implies that the bar formed in our Galaxy 8-10 Gyr ago. Analysis of the metallicity distribution as a function of radius shows no evidence of significant systematic outward radial migration after this first episode. We argue that the variation of the metallicity dispersion as a function of the guiding radius is dominated by the migration triggered by the bar, but also that the libration of orbits around the bar resonances induces a mixing that may have a significant impact on the observed metallicity dispersion. In contrast, the absence of a break in the metallicity profile of populations younger than about $\sim$6 Gyr and the flattening of the gradient at younger ages is interpreted as evidence that the strength of the bar has decreased, loosening its barrier effect and allowing the gas and metals on both sides of the OLR to mix, erasing the break. Beyond the OLR, stars younger than 7 Gyr show very small metallicity dispersion, suggesting no or limited migration.

astro-ph.GA↗

arXiv2023

Stragglers of the thick disc

Young alpha-rich (YAR) stars have been detected in the past as outliers to the local age $\rm-$ [$\alpha$/Fe] relation. These objects are enhanced in $\alpha$-elements but apparently younger than typical thick disc stars. We study the global kinematics and chemical properties of YAR giant stars in APOGEE DR17 survey and show that they have properties similar to those of the standard thick disc stellar population. This leads us to conclude that YAR are rejuvenated thick disc objects, most probably evolved blue stragglers. This is confirmed by their position in the Hertzsprung-Russel diagram (HRD). Extending our selection to dwarfs allows us to obtain the first general straggler distribution in an HRD of field stars. We also compare the elemental abundances of our sample with those of standard thick disc stars, and find that our YAR stars are shifted in oxygen, magnesium, sodium, and the slow neutron-capture element cerium. Although we detect no sign of binarity for most objects, the enhancement in cerium may be the signature of a mass transfer from an asymptotic giant branch companion. The most massive YAR stars suggest that mass transfer from an evolved star may not be the only formation pathway, and that other scenarios, such as collision or coalescence should be considered.

astro-ph.GA↗

arXiv2021

Rapid early gas accretion for the inner Galactic disc

Recent observations of the Milky Way and galaxies at high redshifts suggest that galaxy discs were already in place soon after the Big Bang. While the gas infall history of the Milky Way in the inner disc has long been assumed to be characterised by a short accretion time scale, this has not been directly constrained using observations. Using the unprecedented amount and quality of data of the inner regions of the Milky Way that has recently been produced by APOGEE and Gaia, we aim to derive strong constraints on the infall history of the inner (less than 6 kpc) Galaxy (with a focus on stars between 4-6 kpc, which we show is an appropriate proxy for the entire inner disc). We have implemented gas infall into a chemical evolution model of the Galaxy disc, and used a Schmidt-Kennicutt law to connect the infall to the star formation. We explore a number of models, and two different formulations of the infall law. In one formulation, the infall is non-parametric, and in the other the infall has an explicitly exponential form. We fit the model parameters to the time-[Si/Fe] distribution of solar vicinity stars, and the metallicity and [Si/Fe] distribution function of stars with a galactocentric radius between 4-6 kpc from APOGEE. Our results point to a fast early gas accretion, with an upper limit of accretion timescale of around 2 Gyr in the inner disc of the Milky Way. This suggests that at least half the baryons were in place within 2-3 Gyr of the Big Bang, and that half the stars of the inner disc formed within the first 5 Gyr, during the thick disc formation phase. This implies that the stellar mass of the inner disc is dominated by the thick disc, supporting our previous work, and that the gas accretion onto the inner disc was rapid and early.

astro-ph.GA↗

arXiv2019

Chronos --- Taking the pulse of our Galactic neighbourhood (ESA Voyage 2050 White Paper)

The period 2035-50 considered in the ESA Voyage long-term plan will coincide with a series of foreseeable advances in the characterization of the stellar content of the Milky Way. The Gaia mission, combined with large-scale spectroscopic surveys, is helping to build an unprecedented census in terms of the astrometric, kinematic and chemical properties of Galactic stellar populations. Within a decade, precise measurements of such properties will be available for hundreds of millions of stars. Meanwhile, time-domain surveys initiated with CoRoT and Kepler/K2 and carried on by space missions such as TESS and PLATO or ground-based projects like the LSST, will have brought asteroseismology to a high level of maturity. The combination of precise ages from asteroseismology with astrometric and spectroscopic data, on large stellar samples, is allowing Galactic archaeologists to gain new insight into the assembly history of the Milky Way. Recent breakthroughs --- based on the detection of solar-like oscillations in tens of thousands of red-giant stars --- demonstrate the potential of such approach. Therefore, we are convinced that an all-sky, high-cadence, long-duration stellar variability survey will become a scientific priority in the 2035-50 period. The Chronos concept presented here consists in a time-domain extension to Gaia. It will allow for mass and age estimates for half a million red giants within 1.7 kpc from the Sun and hence shed a new light on our understanding of the Galactic dynamics and archaeology. In terms of the targeted pulsators, Chronos will bridge the gap between PLATO and the LSST by surveying stars all the way from the subgiant branch to the early AGB. Finally, it will surpass all previous surveys capable of conducting asteroseismology in terms of the combined sky coverage and duration of the observations (2 x 3.75 months over the whole sky and >5 years in the CVZ).

astro-ph.SR↗

arXiv2019

New light on the Gaia DR2 parallax zero-point: influence of the asteroseismic approach, in and beyond the Kepler field

The importance of studying the Gaia DR2 parallax zero-point by external means was underlined by Lindegren et al. (2018), and initiated by several works making use of Cepheids, eclipsing binaries, and asteroseismology. Despite a very efficient elimination of basic-angle variations, a small fluctuation remains and shows up as a small offset in the Gaia DR2 parallaxes. By combining astrometric, asteroseismic, spectroscopic, and photometric constraints, we undertake a new analysis of the Gaia parallax offset for nearly 3000 red-giant branch (RGB) and 2200 red clump (RC) stars observed by Kepler, as well as about 500 and 700 red giants (both RGB and RC) selected by the K2 Galactic Archaeology Program in campaigns 3 and 6. Engaging into a thorough comparison of the astrometric and asteroseismic parallaxes, we are able to highlight the influence of the asteroseismic method, and measure parallax offsets in the Kepler field that are compatible with independent estimates from literature and open clusters. Moreover, adding the K2 fields to our investigation allows us to retrieve a clear illustration of the positional dependence of the zero-point, in general agreement with the information provided by quasars. Lastly, we initiate a two-step methodology to make progress in the simultaneous calibration of the asteroseismic scaling relations and of the Gaia DR2 parallax offset, which will greatly benefit from the gain in precision with the third Data Release of Gaia.

astro-ph.SR↗

arXiv2018

The echo of the bar buckling: phase-space spirals in Gaia DR2

Using a single N-body simulation ($N=0.14\times 10^9$) we explore the formation, evolution and spatial variation of the phase-space spirals similar to those recently discovered by Antoja et al. in the Milky Way disk, with Gaia DR2. For the first time in the literature, we use a self-consistent N-body simulation of an isolated Milky Way-type galaxy to show that the phase-space spirals develop naturally from vertical oscillations driven by the buckling of the stellar bar. We claim that the physical mechanism standing behind the observed incomplete phase-space mixing process can be internal and not necessarily due to the perturbation induced by a massive satellite. In our model, the bending oscillations propagate outwards and produce axisymmetric variations of the mean vertical coordinate and of the vertical velocity component. As a consequence, the phase-space wrapping results in the formation of patterns with various morphology across the disk, depending on the bar orientation, distance to the galactic center and time elapsed since the bar buckling. Once bending waves appear, they are supported for a long time via disk self-gravity. The underlying physical mechanism implies the link between in-plane and vertical motion that leads directly to phase-space structures whose amplitude and shape are in remarkable agreement with those of the phase-space spirals observed in the Milky Way disk. In our isolated galaxy simulation, phase-space spirals are still distinguishable, at the solar neighbourhood, 3 Gyr after the buckling phase. The long-lived character of the phase-space spirals generated by the bar buckling instability cast doubts on the timing argument used so far to get back at the time of the onset of the perturbation: phase-space spirals may have been caused by perturbations originated several Gyrs ago, and not as recent as suggested so far.

astro-ph.GA↗

arXiv2016

The Milky Way's halo in 6D: Gaia's Radial Velocity Spectrometer performance

Gaia's Radial Velocity Spectrometer (RVS) has been operating in routine phase for over one year since initial commissioning. RVS continues to work well but the higher than expected levels of straylight reduce the limiting magnitude. The end-of-mission radial-velocity (RV) performance requirement for G2V stars was 15 km/s at V = 16.5 mag. Instead, 15 km/s precision is achieved at 15 < V < 16 mag, consistent with simulations that predict a loss of 1.4 mag. Simulations also suggest that changes to Gaia's onboard software could recover ~0.14 mag of this loss. Consequently Gaia's onboard software was upgraded in April 2015. The status of this new commissioning period is presented, as well as the latest scientific performance of the on-ground processing of RVS spectra. We illustrate the implications of the RVS limiting magnitude on Gaia's view of the Milky Way's halo in 6D using the Gaia Universe Model Snapshot (GUMS).

astro-ph.IM↗

arXiv2014

FORS2/VLT survey of Milky Way globular clusters I. Description of the method for derivation of metal abundances in the optical and application to NGC 6528, NGC 6553, M 71, NGC 6558, NGC 6426 and Terzan 8

(abridged) We have observed almost 1/3 of the globular clusters in the Milky Way, targeting distant and/or highly reddened objects, besides a few reference clusters. A large sample of red giant stars was observed with FORS2@VLT/ESO at R ~ 2,000. The method for derivation of stellar parameters is presented with application to six reference clusters. We aim at deriving the stellar parameters effective temperature, gravity, metallicity and alpha-element enhancement, as well as radial velocity, for membership confirmation of individual stars in each cluster. We analyse the spectra collected for the reference globular clusters NGC 6528, NGC 6553, M 71, NGC 6558, NGC 6426 and Terzan 8. They cover the full range of globular cluster metallicities, and are located in the bulge, disc and halo. Full spectrum fitting techniques are applied, by comparing each target spectrum with a stellar library in the optical region at 4560-5860 A. We employed the library of observed spectra MILES, and the synthetic library by Coelho et al. (2005). Validation of the method is achieved through recovery of the known atmospheric parameters for 49 well-studied stars that cover a wide range in the parameter space. We adopted as final stellar parameters (effective temperatures, gravities, metallicities) the average of results using MILES and Coelho et al. libraries. We identified 4 member stars in NGC 6528, 13 in NGC 6553, 10 in M 71, 5 in NGC 6558, 5 in NGC 6426 and 12 in Terzan 8. Radial velocities, Teff, log(g), [Fe/H] and alpha-element enhancements were derived. We derived abundances for NGC 6426 from spectroscopy for the first time. The method proved to be reliable for red giant stars observed with resolution R ~ 2,000, yielding results compatible with high-resolution spectroscopy. The derived alpha-element abundances show [A/Fe] vs. [Fe/H] consistent with that of field stars at the same metallicities.

astro-ph.GA↗