arXiv subjects

Stephan Frank

Publications and source records attributed to Stephan Frank.

13 recordsLinked to original sources

arXiv2026

Evolutionary Warm-Starts for Reinforcement Learning in Industrial Continuous Control

Reinforcement learning (RL) is still rarely applied in industrial control, partly due to the difficulty of training reliable agents for real-world conditions. This work investigates how evolution strategies can support RL in such settings by introducing a continuous-control adaptation of an industrial sorting benchmark. The CMA-ES algorithm is used to generate high-quality demonstrations that warm-start RL agents. Results show that CMA-ES-guided initialization significantly improves stability and performance. Furthermore, the demonstration trajectories generated with the CMA-ES provide a strong oracle reference performance level, which is of interest in its own right. The study delivers a focused proof of concept for hybrid evolutionary-RL approaches and a basis for future, more complex industrial applications.

cs.NE↗

arXiv2018

A High Signal-to-Noise HST Spectrum Toward J1009+0713: Precise Absorption Measurements in the CGM of Two Galaxies

High signal-to-noise spectra toward background quasars are crucial for uncovering weak absorption in the circumgalactic medium (CGM) of intervening galaxies, such as the diagnostic lines of N V that provide insight to the ionization process of warm gas but typically have low equivalent widths. We present a new spectrum from the Hubble Space Telescope with a signal-to-noise ratio of $\sim20-35$ toward the quasar SDSS J1009+0713 and analyze absorption systems in the CGM of two $L^\star$ galaxies close to the line of sight. We identify additional absorption in the CGM of these galaxies that was not reported by the previous lower signal-to-noise spectrum, as well as Milky Way absorbers and quasar outflows from J1009+0713. We measure $\log (N_\mathrm{NV}/N_\mathrm{OVI})\sim-1.1$ for two CGM absorbers, inconsistent with gas in collisional ionization equilibrium and consistent with a radiatively cooling bulk flow of $\sim50-150$ km s$^{-1}$, which could be produced by galactic winds. These column density ratios are also consistent with those found for other $L^\star$ galaxies and for some gas in the Milky Way's halo. We place upper limits of $\log (N_\mathrm{NV}/N_\mathrm{OVI})<-1.8$ to $-1.2$ for other O VI absorbers in the same halos, which suggests that O VI is produced by different processes in different parts of the CGM, even within the same galactic halo. Together with the kinematically different structure of high- and low-ionization lines, these results indicate there are many components to a single galaxy's gaseous halo. We find the redshift number density of Ly-$\alpha$ forest absorbers and broad Ly-$\alpha$ absorbers are consistent with expectations at this redshift.

astro-ph.GA↗

arXiv2017

Agnostic Stacking of Intergalactic Doublet Absorption: Measuring the NeVIII Population

We present a blind search for doublet intergalactic metal absorption with a method dubbed `agnostic stacking'. Using a forward-modelling framework we combine this with direct detections in the literature to measure the overall metal population. We apply this novel approach to the search for NeVIII absorption in a set of 26 high-quality COS spectra. We probe to an unprecedented low limit of log N$>$12.3 at 0.47$\leq z \leq$1.34 over a pathlength $\Delta$z = 7.36. This method selects apparent absorption without requiring knowledge of its source. Stacking this mixed population dilutes doublet features in composite spectra in a deterministic manner, allowing us to measure the proportion corresponding to NeVIII absorption. We stack potential NeVIII absorption in two regimes: absorption too weak to be significant in direct line studies (12.3 $<$ log N $<$ 13.7), and strong absorbers (log N $>$ 13.7). We do not detect NeVIII absorption in either regime. Combining our measurements with direct detections, we find that the NeVIII population is reproduced with a power law column density distribution function with slope $\beta = -1.86 \substack{+0.18 \\ -0.26}$ and normalisation log $f_{13.7} = -13.99 \substack{+0.20 \\ -0.23}$, leading to an incidence rate of strong NeVIII absorbers $dn/dz =1.38 \substack{+0.97 \\ -0.82}$. We infer a cosmic mass density for NeVIII gas with 12.3 $<$ log N $<$ 15.0 of $\Omega _{NeVIII} = 2.2 \substack{+1.6 \\ _-1.2} \times 10^{-8}$, a value significantly lower that than predicted by recent simulations. We translate this density into an estimate of the baryon density $\Omega _{b} \approx 1.8 \times 10^{-3}$, constituting 4\% of the total baryonic mass.

astro-ph.CO↗

arXiv2016

C II radiative cooling of the Galactic diffuse interstellar medium: Insight about the star formation in Damped Lyman-alpha systems

The far-infrared [C II] 158 micrometer fine structure transition is considered to be a dominant coolant in the interstellar medium. For this reason, under the assumption of a thermal steady state, it may be used to infer the heating rate and, in turn, the star formation rate in local, as well as in high redshift systems. In this work, radio and ultraviolet observations of the Galactic interstellar medium are used to understand whether C II is indeed a good tracer of the star formation rate. For a sample of high Galactic latitude sightlines, direct measurements of the temperature indicate the presence of C II in both the cold and the warm phases of the diffuse interstellar gas. The cold gas fraction (~ 10 - 50% of the total neutral gas column density) is not negligible even at high Galactic latitude. It is shown that, to correctly estimate the star formation rate, C II cooling in both the phases should hence be considered. The simple assumption, that the [C II] line originates only from either the cold or the warm phase, significantly underpredicts or overpredicts the star formation rate, respectively. These results are particularly important in the context of the Damped Lyman-alpha systems for which a similar method is often used to estimate the star formation rate. The derived star formation rates in such cases may not be reliable if the temperature of the gas under consideration is not constrained independently.

astro-ph.GA↗

arXiv2013

Probing the Circumgalactic Medium at High-Redshift Using Composite BOSS Spectra of Strong Lyman-alpha Forest Absorbers

We present composite spectra constructed from a sample of 242,150 Lyman-alpha (Lya) forest absorbers at redshifts 2.4<z<3.1 identified in quasar spectra from the Baryon Oscillation Spectroscopic Survey (BOSS) as part of Data Release 9 of the Sloan Digital Sky Survey III. We select forest absorbers by their flux in bins 138 km/s wide (approximately the size of the BOSS resolution element). We split these absorbers into five samples spanning the range of flux -0.05 < F<0.45. Tests on a smaller sample of high-resolution spectra show that our three strongest absorption bins would probe circumgalactic regions (projected separation < 300 proper kpc and |Delta v| < 300km/s) in about 60% of cases for very high signal-to-noise ratio. Within this subset, weakening Lya absorption is associated with decreasing purity of circumgalactic selection once BOSS noise is included. Our weaker two Lya absorption samples are dominated by the intergalactic medium. We present composite spectra of these samples and a catalogue of measured absorption features from HI and 13 metal ionization species, all of which we make available to the community. We compare measurements of seven Lyman series transitions in our composite spectra to single line models and obtain further constraints from their associated excess Lyman limit opacity. This analysis provides results consistent with column densities over the range 14.4 <~ Log (N_HI) <~ 16.45. We compare our measurements of metal absorption to a variety of simple single-line, single-phase models for a preliminary interpretation. Our results imply clumping on scales down to ~30 pc and near-solar metallicities in the circumgalactic samples, while high-ionization metal absorption consistent with typical IGM densities and metallicities is visible in all samples.

astro-ph.CO↗

arXiv2013

The ESO UVES Advanced Data Products Quasar Sample - II. Cosmological Evolution of the Neutral Gas Mass Density

Quasar foreground damped absorbers, associated with HI-rich galaxies allow to estimate the neutral gas mass over cosmic time, which is a possible indicator of gas consumption as star formation proceeds. The DLAs and sub-DLAs are believed to contain a large fraction of neutral gas mass in the Universe. In Paper I of the series, we present the results of a search for DLAs and sub-DLAs in the ESO-UVES Advanced Data Products dataset of 250 quasars. Here we use an unbiased sub-sample of sub-DLAs from this dataset. We build a subset of 122 quasars ranging from 1.5 <z_em < 5.0, suitable for statistical analysis. The statistical sample is analyzed in conjunction with other sub-DLA samples from the literature. This makes up a combined sample of 89 sub-DLAs over a redshift path of $\Delta z=193$. Redshift evolution of the number density and the line density are derived for sub-DLAs and compared with the LLSs and DLAs measurements from the literature. The results indicate that these three classes of absorbers are evolving in the redshift interval 1 < z < 5. The column density distribution, f(N,z), down to the sub-DLA limit is determined. The flattening of f_(N,z) in the sub-DLA regime is present in the observations. The redshift evolution of f_(N,z) down to sub-DLA regime is also presented, indicating the presence of more sub-DLAs at high-redshift as compared to low-redshift. f_(N,z) is further used to determine the neutral gas mass density, Omega_g, at 1.5 < z < 5.0. The complete sample shows that sub-DLAs contribute 8-20% to the total Omega_g from 1.5 < z < 5.0. In agreement with previous studies, no evolution of Omega_g is seen from low-redshift to high-redshift, suggesting that star formation solely cannot explain this non-evolution and replenishment of gas and/or recombination of ionized gas is needed. (Abridged)

astro-ph.CO↗

arXiv2012

Optical Monitoring of the Broad-Line Radio Galaxy 3C390.3

We have undertaken a new ground-based monitoring campaign on the BLRG 3C390.3 to improve the measurement of the size of the BLR and to estimate the black hole mass. Optical spectra and g-band images were observed in 2005 using the 2.4m telescope at MDM Observatory. Integrated emission-line flux variations were measured for Ha, Hb, Hg, and for HeII4686, as well as g-band fluxes and the optical AGN continuum at 5100A. The g-band fluxes and the AGN continuum vary simultaneously within the uncertainties, tau=(0.2+-1.1)days. We find that the emission-line variations are delayed with respect to the variable g-band continuum by tau(Ha)=56.3(+2.4-6.6)days, tau(Hb)=44.3(+3.0_-3.3)days, tau(Hg)=58.1(+4.3-6.1)days, and tau(HeII4686)=22.3(+6.5-3.8)days. The blue and red peak in the double peaked line profiles, as well as the blue and red outer profile wings, vary simultaneously within +-3 days. This provides strong support for gravitationally bound orbital motion of the dominant part of the line emitting gas. Combining the time delay of Ha and Hb and the separation of the blue and red peak in the broad double-peaked profiles in their rms spectra, we determine Mbh_vir=1.77(+0.29-0.31)x10^8Msol and using sigma_line of the rms spectra Mbh_vir=2.60(+0.23-0.31)x10^8Msol for the central black hole of 3C390.3, respectively. Using the inclination angle of the line emitting region the mass of the black hole amounts to Mbh=0.86(+0.19-0.18)x10^9 Msol (peak-separation) and Mbh=1.26(+0.21-0.16)x10^9 Msol (sigma_line), respectively. This result is consistent with the black hole masses indicated by simple accretion disk models to describe the observed double-peaked profiles, derived from the stellar dynamics of 3C390.3, and with the AGN radius-luminosity relation. Thus, 3C390.3 as a radio-loud AGN with a low Eddington ratio, Ledd/Lbol=0.02, follows the same AGN radius-luminosity relation as radio-quiet AGN.

astro-ph.CO↗

arXiv2010

The Composite Spectrum of Strong Lyman-alpha Forest Absorbers

We present a new method for probing the physical conditions and metal enrichment of the Intergalactic Medium: the composite spectrum of Ly-alpha forest absorbers. We apply this technique to a sample of 9480 Ly-alpha absorbers with redshift 2 < z < 3.5 identified in the spectra of 13,279 high-redshift quasars from the Sloan Digital Sky Survey (SDSS) Fifth Data Release (DR5). Absorbers are selected as local minima in the spectra with 2.4 < tau_Ly-alpha < 4.0; at SDSS resolution (~ 150km/s FWHM), these absorbers are blends of systems that are individually weaker. In the stacked spectra we detect seven Lyman-series lines and metal lines of O VI, N V, C IV, C III, Si IV, C II, Al II, Si II, Fe II, Mg II, and O I. Many of these lines have peak optical depths of < 0.02, but they are nonetheless detected at high statistical significance. Modeling the Lyman-series measurements implies that our selected systems have total H I column densities N_HI ~ 10^15.4cm-2. Assuming typical physical conditions rho / = 10, T = 10^4 - 10^4.5 K, and [Fe/H]= -2 yields reasonable agreement with the line strengths of high-ionization species, but it underpredicts the low-ionization species by two orders of magnitude or more. This discrepancy suggests that the low ionization lines arise in dense, cool, metal-rich clumps, present in some absorption systems.

astro-ph.CO↗

arXiv2009

A Search for Oxygen in the Low-Density Lyman-alpha Forest Using the Sloan Digital Sky Survey

We use 2167 Sloan Digital Sky Survey (SDSS) quasar spectra to search for low-density oxygen in the Intergalactic Medium (IGM). Oxygen absorption is detected on a pixel-by-pixel basis by its correlation with Lyman-alpha forest absorption. We have developed a novel Locally Calibrated Pixel (LCP) search method that uses adjacent regions of the spectrum to calibrate interlopers and spectral artifacts, which would otherwise limit the measurement of OVI absorption. Despite the challenges presented by searching for weak OVI within the Lyman-alpha forest in spectra of moderate resolution and signal-to-noise, we find a highly significant detection of absorption by oxygen at 2.7 < z < 3.2 (the null hypothesis has a chi^2=80 for 9 data points). We interpret our results using synthetic spectra generated from a lognormal density field assuming a mixed quasar-galaxy photoionizing background (Haardt & Madau 2001) and that it dominates the ionization fraction of detected OVI. The LCP search data can be fit by a constant metallicity model with [O/H] = -2.15_(-0.09)^(+0.07), but also by models in which low-density regions are unenriched and higher density regions have a higher metallicity. The density-dependent enrichment model by Aguirre et al. (2008) is also an acceptable fit. All our successful models have similar mass-weighted oxygen abundance, corresponding to [ _MW] = -2.45+-0.06. This result can be used to find the cosmic oxygen density in the Lyman-alpha forest, Omega_(Oxy, IGM) = 1.4(+-0.2)x10^(-6) = 3x10^(-4) Omega_b. This is the tightest constraint on the mass-weighted mean oxygen abundance and the cosmic oxygen density in the Lyman-alpha forest to date and indicates that it contains approximately 16% of metals produced by star formation (Bouch\'e et al. 2008) up to z = 3.

astro-ph.CO↗

arXiv2008

First-Year Spectroscopy for the SDSS-II Supernova Survey

This paper presents spectroscopy of supernovae discovered in the first season of the Sloan Digital Sky Survey-II Supernova Survey. This program searches for and measures multi-band light curves of supernovae in the redshift range z = 0.05 - 0.4, complementing existing surveys at lower and higher redshifts. Our goal is to better characterize the supernova population, with a particular focus on SNe Ia, improving their utility as cosmological distance indicators and as probes of dark energy. Our supernova spectroscopy program features rapid-response observations using telescopes of a range of apertures, and provides confirmation of the supernova and host-galaxy types as well as precise redshifts. We describe here the target identification and prioritization, data reduction, redshift measurement, and classification of 129 SNe Ia, 16 spectroscopically probable SNe Ia, 7 SNe Ib/c, and 11 SNe II from the first season. We also describe our efforts to measure and remove the substantial host galaxy contamination existing in the majority of our SN spectra.

astro-ph↗

arXiv2007

Disparate MgII Absorption Statistics towards Quasars and Gamma-Ray Bursts : A Possible Explanation

We examine the recent report by Prochter et al. (2006) that gamma-ray burst (GRB) sight lines have a much higher incidence of strong MgII absorption than quasar sight lines. We propose that the discrepancy is due to the different beam sizes of GRBs and quasars, and that the intervening MgII systems are clumpy with the dense part of each cloudlet of a similar size as the quasars, i.e. < 10^16 cm, but bigger than GRBs. We also discuss observational predictions of our proposed model. Most notably, in some cases the intervening MgII absorbers in GRB spectra should be seen varying, and quasars with smaller sizes should show an increased rate of strong MgII absorbers. In fact, our prediction of variable MgII lines in the GRB spectra has been now confirmed by Hao et al. (2007), who observed intervening FeII and MgII lines at z=1.48 to be strongly variable in the multi-epoch spectra of z=4.05 GRB060206.

astro-ph↗

arXiv2005

Studying the Variation of the Fine Structure Constant Using Emission Line Multiplets

As an extension of the method by Bahcall et al. (2004) to investigate the time dependence of the fine structure constant, we describe an approach based on new observations of forbidden line multiplets from different ionic species. We obtain optical spectra of fine structure transitions in [Ne III], [Ne V], [O III], [OI], and [SII] multiplets from a sample of 14 Seyfert 1.5 galaxies in the low-z range 0.035 < z < 0.281. Each source and each multiplet is independently analyzed to ascertain possible errors. Averaging over our sample, we obtain a conservative value alpha^2(t)/α^2(0) = 1.0030+-0.0014. However, our sample is limited in size and our fitting technique simplistic as we primarily intend to illustrate the scope and strengths of emission line studies of the time variation of the fine structure constant. The approach can be further extended and generalized to a "many-multiplet emission line method" analogous in principle to the corresponding method using absorption lines. With that aim, we note that the theoretical limits on emission line ratios of selected ions are precisely known, and provide well constrained selection criteria. We also discuss several other forbidden and allowed lines that may constitute the basis for a more rigorous study using high-resolution instruments on the next generation of 8 m class telescopes.

astro-ph↗

arXiv2003

The metal absorption systems of the QSO 0103-260 and the galaxy redshift distribution in the FORS Deep Field

Using the Uves echelle spectrograph at the ESO VLT, we observed the absorption line spectrum of the QSO 0103-260 in the Fors Deep Field. In addition to the expected Ly-alpha forest lines, we detected at least 16 metal absorption systems with highly different ionization levels in the observed spectral range. The redshifts of the metal absorption systems are strongly correlated with the redshift distribution of the high-z galaxies in the Fors Deep Field and with the strength (but not the number density) of the Ly-alpha forest lines. Both the metal systems and the galaxies show clustering at least up to the QSO emission line redshift of 3.365, but only few of these galaxy accumulations seem to form bound systems.

astro-ph↗