arXiv subjects

Andreas Koch

Publications and source records attributed to Andreas Koch.

At least 19 recordsLinked to original sources

arXiv2026

A hybrid sharp-diffuse interface approach to accurately model melt pool dynamics with rapid evaporation in laser-based processing of metals

Predictive simulation of melt pool dynamics in laser-based processing of metals, e.g., laser beam welding or laser powder bed fusion additive manufacturing, requires accurate resolution of thermo-hydrodynamic interactions at the melt-gas interface. Here, evaporation-induced recoil pressure and temperature-dependent surface tension govern the flow. Because these mechanisms depend sensitively, often exponentially, on the interface temperature, reliable predictions demand highly accurate heat transfer models. Popular diffuse-interface formulations smear the extreme thermal gradients as typical for laser-metal interactions, leading to interface temperature errors that critically degrade the accuracy of interface force predictions and melt pool dynamics. We present a hybrid sharp-diffuse interface approach for high-fidelity modelling of melt pool thermo-hydrodynamics with rapid evaporation. The heat transfer problem is represented using a sharp-interface unfitted finite element (CutFEM) formulation, enabling accurate prediction of the temperature field. The multi-phase flow problem, characterized by large density ratios and complex interface dynamics, is accurately captured using a robust level-set-based one-fluid diffuse-interface finite element formulation. Consistent coupling is achieved by extending the sharp-interface temperature into a narrow interface region to evaluate temperature-dependent interface forces within the diffuse-interface flow framework. In practically relevant benchmarks, the sharp-interface thermal model exhibits second-order spatial convergence, enabling finite element sizes two orders of magnitude larger than the diffuse-interface approach for 1 accuracy. In a novel coupled thermo-hydrodynamic benchmark representative of laser-metal interactions, the hybrid approach is one order of magnitude more accurate than a purely diffuse-interface model on the same mesh. Robu

cs.CE↗

arXiv2026

Bringing Data Transformations Near-Memory for Low-Latency Analytics in HTAP Environments

In this paper we propose an approach for executing data transformations near- or in-storage on intelligent storage systems. The currently prevailing approach of extracting the data and then transforming it to a target format suffers degraded performance during transformation and causes heavy data movement. Our results show robust performance of foreground workloads and lower resource contention. Our vision draws architectural opportunities in multi-engine and multi-system settings, as well as for reuse.

cs.DB↗

arXiv2025

PUL: Pre-load in Software for Caches Wouldn't Always Play Along

Memory latencies and bandwidth are major factors, limiting system performance and scalability. Modern CPUs aim at hiding latencies by employing large caches, out-of-order execution, or complex hardware prefetchers. However, software-based prefetching exhibits higher efficiency, improving with newer CPU generations. In this paper we investigate software-based, post-Moore systems that offload operations to intelligent memories. We show that software-based prefetching has even higher potential in near-data processing settings by maximizing compute utilization through compute/IO interleaving.

cs.DB↗

arXiv2024

A survey on FPGA-based accelerator for ML models

This paper thoroughly surveys machine learning (ML) algorithms acceleration in hardware accelerators, focusing on Field-Programmable Gate Arrays (FPGAs). It reviews 287 out of 1138 papers from the past six years, sourced from four top FPGA conferences. Such selection underscores the increasing integration of ML and FPGA technologies and their mutual importance in technological advancement. Research clearly emphasises inference acceleration (81\%) compared to training acceleration (13\%). Additionally, the findings reveals that CNN dominates current FPGA acceleration research while emerging models like GNN show obvious growth trends. The categorization of the FPGA research papers reveals a wide range of topics, demonstrating the growing relevance of ML in FPGA research. This comprehensive analysis provides valuable insights into the current trends and future directions of FPGA research in the context of ML applications.

cs.AR↗

arXiv2023

PQC-HA: A Framework for Prototyping and In-Hardware Evaluation of Post-Quantum Cryptography Hardware Accelerators

In the third round of the NIST Post-Quantum Cryptography standardization project, the focus is on optimizing software and hardware implementations of candidate schemes. The winning schemes are CRYSTALS Kyber and CRYSTALS Dilithium, which serve as a Key Encapsulation Mechanism (KEM) and Digital Signature Algorithm (DSA), respectively. This study utilizes the TaPaSCo open-source framework to create hardware building blocks for both schemes using High-level Synthesis (HLS) from minimally modified ANSI C software reference implementations across all security levels. Additionally, a generic TaPaSCo host runtime application is developed in Rust to verify their functionality through the standard NIST interface, utilizing the corresponding Known Answer Test mechanism on actual hardware. Building on this foundation, the communication overhead for TaPaSCo hardware accelerators on PCIe-connected FPGA devices is evaluated and compared with previous work and optimized AVX2 software reference implementations. The results demonstrate the feasibility of verifying and evaluating the performance of Post-Quantum Cryptography accelerators on real hardware using TaPaSCo. Furthermore, the off-chip accelerator communication overhead of the NIST standard interface is measured, which, on its own, outweighs the execution wall clock time of the optimized software reference implementation of Kyber at Security Level 1.

cs.CR↗

arXiv2021

End-to-End Annotator Bias Approximation on Crowdsourced Single-Label Sentiment Analysis

Sentiment analysis is often a crowdsourcing task prone to subjective labels given by many annotators. It is not yet fully understood how the annotation bias of each annotator can be modeled correctly with state-of-the-art methods. However, resolving annotator bias precisely and reliably is the key to understand annotators' labeling behavior and to successfully resolve corresponding individual misconceptions and wrongdoings regarding the annotation task. Our contribution is an explanation and improvement for precise neural end-to-end bias modeling and ground truth estimation, which reduces an undesired mismatch in that regard of the existing state-of-the-art. Classification experiments show that it has potential to improve accuracy in cases where each sample is annotated only by one single annotator. We provide the whole source code publicly and release an own domain-specific sentiment dataset containing 10,000 sentences discussing organic food products. These are crawled from social media and are singly labeled by 10 non-expert annotators.

cs.CL↗

arXiv2021

Explainable Online Validation of Machine Learning Models for Practical Applications

We present a reformulation of the regression and classification, which aims to validate the result of a machine learning algorithm. Our reformulation simplifies the original problem and validates the result of the machine learning algorithm using the training data. Since the validation of machine learning algorithms must always be explainable, we perform our experiments with the kNN algorithm as well as with an algorithm based on conditional probabilities, which is proposed in this work. For the evaluation of our approach, three publicly available data sets were used and three classification and two regression problems were evaluated. The presented algorithm based on conditional probabilities is also online capable and requires only a fraction of memory compared to the kNN algorithm.

cs.LG↗

arXiv2020

Difference in chemical composition between the bright and faint red clump stars in the Milky Way bulge

The double red clump (RC) observed in color-magnitude diagrams of the Milky Way bulge is at the heart of the current debate on the structure and formation origin of the bulge. This feature can be explained by the difference between the two RCs either in distance ("X-shaped scenario") or in chemical composition ("multiple-population scenario"). Here we report our high-resolution spectroscopy for the RC and red giant branch stars in a high-latitude field (b ~ -8.5$°$) of the bulge. We find a difference in [Fe/H] between the stars in the bright and faint RC regimes, in the sense that the bright stars are enhanced in [Fe/H] with respect to the faint stars by 0.149 $\pm$ 0.036 dex. The stars on the bright RC are also enhanced in [Na/Fe] but appear to be depleted in [Al/Fe] and [O/Fe], although more observations are required to confirm the significance of these differences. Interestingly, these chemical patterns are similar to those observed among multiple stellar populations in the metal-rich bulge globular cluster Terzan 5. In addition, we find a number of Na-rich stars, which would corroborate the presence of multiple populations in the bulge. Our results support an origin of the double RC from dissolved globular clusters that harbor multiple stellar populations. Thus, our study suggests that a substantial fraction of the outer bulge stars would have originated from the assembly of such stellar systems in the early phase of the Milky Way formation.

astro-ph.GA↗

arXiv2020

Mono-enriched stars and Galactic chemical evolution -- Possible biases in observations and theory

A long sought after goal using chemical abundance patterns derived from metal-poor stars is to understand the Galactic chemical evolution (GCE) and to pin down the nature of the first stars (Pop III). Here, we use a sample of 14 metal-poor stars observed with the high-resolution spectrograph PEPSI at the LBT to derive abundances of 32 elements (34 including limits). We present well-sampled abundance patterns for all stars obtained using local thermodynamic equilibrium (LTE) radiative transfer codes and 1D hydrostatic model atmospheres. It is currently well known that the assumptions of 1D and LTE may hide several issues, thereby introducing biases in our interpretation as to the nature of the first stars and the GCE. Hence, we use non-LTE (NLTE) and correct the abundances using 3D model atmospheres to present a physically more reliable pattern. In order to infer the nature of the first stars, we compare unevolved, cool stars, enriched by a single event (`mono-enriched'), with a set of yield predictions to pin down the mass and energy of the Pop III progenitor. To date, only few bona fide second generation mono-enriched stars are known. A simple x^2-fit may bias our inferred mass and energy just as much as the simple 1D LTE abundance pattern, and we thus carried out our study with an improved fitting technique considering dilution and mixing. Our sample presents Carbon Enhanced Metal-Poor (CEMP) stars, some of which are promising true second generation (mono-enriched) stars. The unevolved, dwarf BD+09_2190 shows a mono-enriched signature which, combined with kinematical data, indicates that it moves in the outer halo and likely has been accreted onto the Milky Way early on. The Pop III progenitor was likely of 25.5M and 0.6 10^51erg (foe)/19.2M and 1.5foe in LTE/NLTE. Finally, we explore the predominant donor and formation site of the rapid and slow neutron-capture elements. Abridged

astro-ph.SR↗

arXiv2020

The Blanco DECam Bulge Survey. I. The Survey Description and Early Results

The Blanco Dark Energy Camera (DECam) Bulge survey is a Vera Rubin Observatory (LSST) pathfinder imaging survey, spanning $\sim 200$ sq. deg. of the Southern Galactic bulge, $-2^\circ <$b$< -13^\circ$ and $-11^\circ <$l$ < +11^\circ$. We have employed the CTIO-4m telescope and the Dark Energy Camera (DECam) to image a contiguous $\sim 200$ sq. deg. region of the relatively less reddened Southern Galactic bulge, in SDSS $u$ + Pan-STARRS$grizy$. Optical photometry with its large colour baseline will be used to investigate the age and metallicity distributions of the major structures of the bulge. Included in the survey footprint are 26 globular clusters imaged in all passbands. Over much of the bulge, we have Gaia DR2 matching astrometry to $i\sim 18$, deep enough to reach the faint end of the red clump. This paper provides the background, scientific case, and description of the survey. We present an array of new reddening-corrected colour-magnitude diagrams that span the extent of Southern Galactic bulge. We argue that a population of massive stars in the blue loop evolutionary phase, proposed to lie in the bulge, are instead at $\sim 2$ kpc from the Sun and likely red clump giants in the old disk. A bright red clump near $(l,b)=(+8^\circ,-4^\circ)$ may be a feature in the foreground disk, or related to the long bar reported in earlier work. We also report the first map of the blue horizontal branch population spanning the BDBS field of regard, and our data does not confirm the reality of a number of proposed globular clusters in the bulge.

astro-ph.GA↗

arXiv2020

Blanco DECam Bulge Survey (BDBS) II: Project Performance, Data Analysis, and Early Science Results

The Blanco DECam Bulge Survey (BDBS) imaged more than 200 square degrees of the Southern Galactic bulge using the ugrizY filters of the Dark Energy Camera, and produced point spread function photometry of approximately 250 million unique sources. In this paper, we present details regarding the construction and collation of survey catalogs, and also discuss the adopted calibration and dereddening procedures. Early science results are presented with a particular emphasis on the bulge metallicity distribution function and globular clusters. A key result is the strong correlation (sigma ~ 0.2 dex) between (u-i)o and [Fe/H] for bulge red clump giants. We utilized this relation to find that interior bulge fields may be well described by simple closed box enrichment models, but fields exterior to b ~ -6 degrees seem to require a secondary metal-poor component. Applying scaled versions of the closed box model to the outer bulge fields is shown to significantly reduce the strengths of any additional metal-poor components when compared to Gaussian mixture models. Additional results include: a confirmation that the u-band splits the sub-giant branch in M 22 as a function of metallicity, the detection of possible extra-tidal stars along the orbits of M 22 and FSR 1758, and additional evidence that NGC 6569 may have a small but discrete He spread, as evidenced by red clump luminosity variations in the reddest bands. We do not confirm previous claims that FSR 1758 is part of a larger extended structure.

astro-ph.GA↗

arXiv2020

GPU-Accelerated Drug Discovery with Docking on the Summit Supercomputer: Porting, Optimization, and Application to COVID-19 Research

Protein-ligand docking is an in silico tool used to screen potential drug compounds for their ability to bind to a given protein receptor within a drug-discovery campaign. Experimental drug screening is expensive and time consuming, and it is desirable to carry out large scale docking calculations in a high-throughput manner to narrow the experimental search space. Few of the existing computational docking tools were designed with high performance computing in mind. Therefore, optimizations to maximize use of high-performance computational resources available at leadership-class computing facilities enables these facilities to be leveraged for drug discovery. Here we present the porting, optimization, and validation of the AutoDock-GPU program for the Summit supercomputer, and its application to initial compound screening efforts to target proteins of the SARS-CoV-2 virus responsible for the current COVID-19 pandemic.

q-bio.BM↗

arXiv2020

The Bulge Radial Velocity Assay for RR Lyrae stars (BRAVA-RR) DR2: a Bimodal Bulge?

Radial velocities of 2768 fundamental mode RR Lyrae stars (RRLs) toward the Southern Galactic bulge are presented, spanning the southern bulge from -8 < l < +8 and -3 < b <-6. Distances derived from the pulsation properties of the RRLs are combined with Gaia proper motions to give constraints on the orbital motions of 1389 RRLs. The majority (~75%) of the bulge RRLs have orbits consistent with these stars being permanently bound to <3.5 kpc from the Galactic Center, similar to the bar. However, unlike the bulge giants, the RRLs exhibit slower rotation and a higher velocity dispersion. The higher velocity dispersion arises almost exclusively from halo interlopers passing through the inner Galaxy. We present 82 stars with space velocities > 500 km/s and find that the majority of these high-velocity stars are halo interlopers; it is unclear if a sub-sample of these stars with similar space velocities have a common origin. Once the 25% of the sample represented by halo interlopers is cleaned, we can clearly discern two populations of bulge RRLs in the inner Galaxy. One population of RRLs is not as tightly bound to the Galaxy (but is still confined to the inner ~3.5 kpc), and is both spatially and kinematically consistent with the barred bulge. The second population is more centrally concentrated and does not trace the bar. One possible interpretation is that this population was born prior to bar formation, as its spatial location, kinematics and pulsation properties suggest, possibly from an accretion event at high redshift.

astro-ph.SR↗

arXiv2020

Purveyors of fine halos. II. Chemodynamical association of halo stars with Milky Way globular clusters

We present chemodynamical links between the present-day Milky Way halo field star population and Galactic globular clusters (GCs) using a dataset that combines information from the $\rm{\it Gaia}$ space mission and the Sloan Digital Sky Survey (SDSS-IV, DR14). Moreover, we incorporated a sample of halo giant stars with a distinct chemical signature (strong CN bandheads) that resembles the light-elements anomaly otherwise only seen in the second generation of GC stellar populations. Using three different tagging techniques, we could successfully establish unique associations between 151 extratidal stars in the neighborhood of eight GCs. In addition, we traced the possible origin of about $62\%$ of the sample of CN-strong giants to their potential host clusters. Several of the involved GCs have been brought into connection with the Gaia-Enceladus and Sequoia merger events. By establishing kinematic and chemical connections between 17 CN-strong stars and their surrounding fields, we could identify co-moving groups of stars at the same [Fe/H] with a possible cluster origin. We found strong evidence that four CN-strong stars and their associates are connected to the Sagittarius stream whilst their tightly confined [Fe/H] may hint to a birth site in M 54. Finally, we provide tentative estimates for the fraction of first-generation cluster stars among all stars lost to the halo. In the immediate cluster vicinity, this value amounts to $50.0\pm16.7\%$ while the associations in the halo field rather imply $80.2_{-5.2}^{+4.9}\%$. We speculate that -- if proven real by spectroscopic follow-up -- the disparity between these numbers could indicate a major contribution of low-mass clusters to the overall number of stars escaped to the halo or, alternatively, point toward a strong mass loss from the first generation during early cluster dissolution. [abridged]

astro-ph.GA↗

arXiv2020

The Halos and Environments of Nearby Galaxies (HERON) II: The outer structure of edge-on galaxies

The HERON project is aimed at studying halos and low surface brightness details near galaxies. In this second HERON paper we consider in detail deep imaging (down to surface brightness of ~28 mag/arcsec$^2$ in the r band) for 35 galaxies, viewed edge-on. We confirm a range of low surface brightness features previously described in the literature but also report new ones. We classify the observed outer shapes of the galaxies into three main types (and their prototypes): disc/diamond-like (NGC891), oval (NGC4302), and boxy (NGC3628). We show that the shape of the outer disc in galaxies does not often follow the general 3D model of an exponential disc: 17 galaxies in our sample exhibit oval or even boxy isophotes at the periphery. Also, we show that the less flattened the outer disc is, the more oval or boxy its structure. Many galaxies in our sample have an asymmetric outer structure. We propose that the observed diversity of the galaxy outer shapes is defined by the merger history and its intensity: if no recent multiple minor or single major merging took place, the outer shape is diamond-like or discy. On the contrary, interacting galaxies show oval outer shapes, whereas recent merging appears to transform the outer shape to boxy.

astro-ph.GA↗

arXiv2019

Polarimetric Radar Cross-Sections of Pedestrians at Automotive Radar Frequencies

Simulation of radar cross-sections (RCS) of pedestrians at automotive radar frequencies forms a key tool for software verification test beds for advanced driver assistance systems. Two commonly used simulation methods are: the computationally simple scattering center model of dynamic humans; and the shooting and bouncing ray technique based on geometric optics. The latter technique is more accurate but due to its computational complexity, it is usually used only for modeling scattered returns of still human poses. In this work, we combine the two methods in a linear regression framework to accurately estimate the scattering coefficients or reflectivies of point scatterers in a realistic automotive radar signal model which we subsequently use to simulate range-time, Doppler-time and range-Doppler radar signatures. The simulated signatures show a normalized mean square error below 10% and a structural similarity above $81\%$ with respect to measurement results generated with an automotive radar at 77 GHz.

eess.SP↗

arXiv2019

An outer shade of Pal: Abundance analysis of the outer halo globular cluster Palomar 13

At a Galactocentric distance of 27 kpc, Pal 13 is an old globular cluster (GC) in the outer halo. We present a chemical abundance analysis of this remote system from high-resolution spectra obtained with Keck/HIRES. Owing to the low signal-to-noise ratio of the data, our analysis is based on a coaddition of the spectra of 18 member stars. We are able to determine integrated abundance ratios for 16 species of 14 elements, of $α$-elements (Mg,Si,Ca,Ti), Fe-peak (Sc,Mn,Cr,Ni,Cu,and Zn), and n-capture elements (Y,Ba). While the mean Na abundance is found to be slightly enhanced and halo-like, our method does not allow us to probe an abundance spread that would be expected in this light element if multiple populations are present in Pal 13. We find a metal-poor mean metallicity of $-1.91\pm0.05$ (statistical) $\pm$ 0.22 (systematic), confirming that Pal 13 is a typical metal-poor representative of the outer halo. While there are some differences between individual $α$-elements, such as halo-like Mg and Si versus the mildly lower Ca and Ti abundances, the mean [$α$/Fe] of 0.34$\pm$0.06 is consistent with the marginally lower $α$ component of the halo field and GC stars. We discuss our results in the context of other objects in the outer halo and consider which of these objects were likely accreted. We also discuss the properties of their progenitors. While chemically, Pal 13 is similar to Gaia-Enceladus and some of its GCs, this is not supported by its kinematic properties. Moreover, its chemodynamical similarity with NGC 5466, a progeny of the Sequoia accretion event, might indicate a common origin in this progenitor. However, the ambiguities in the full abundance space of this comparison emphasize the difficulties in unequivocally labeling a single GC as an accreted object, let alone assigning it to a single progenitor. (Abridged)

astro-ph.GA↗

arXiv2019

Identification of strontium in the merger of two neutron stars

Half of all the elements in the universe heavier than iron were created by rapid neutron capture. The theory for this astrophysical `$r$-process' was worked out six decades ago and requires an enormous neutron flux to make the bulk of these elements. Where this happens is still debated. A key piece of missing evidence is the identification of freshly-synthesised $r$-process elements in an astrophysical site. Current models and circumstantial evidence point to neutron star mergers as a probable $r$-process site, with the optical/infrared `kilonova' emerging in the days after the merger a likely place to detect the spectral signatures of newly-created neutron-capture elements. The kilonova, AT2017gfo, emerging from the gravitational-wave--discovered neutron star merger, GW170817, was the first kilonova where detailed spectra were recorded. When these spectra were first reported it was argued that they were broadly consonant with an outflow of radioactive heavy elements, however, there was no robust identification of any element. Here we report the identification of the neutron-capture element strontium in a re-analysis of these spectra. The detection of a neutron-capture element associated with the collision of two extreme-density stars establishes the origin of $r$-process elements in neutron star mergers, and demonstrates that neutron stars contain neutron-rich matter.

astro-ph.HE↗