arXiv subjects

Marwan Gebran

Publications and source records attributed to Marwan Gebran.

11 recordsLinked to original sources

arXiv2026

RHEED pattern classification by a convolutional neural network for the growth of chalcogenide thin films and nanostructures

The use of reflection high energy electron diffraction (RHEED) plays a critical role for in-situ characterization in molecular beam epitaxy, pulsed laser deposition and sputtering. While sensitive to crystal symmetries and morphology, it is used ubiquitously to determine the growth modes of thin films. However, analysis of RHEED patterns depends on skilled experts and is therefore difficult to incorporate into the growth strategy in real-time. The development of machine learning (ML) processes, specifically convolutional neural networks (CNNs), presents a unique opportunity towards real-time RHEED pattern recognition. In this study, we develop a CNN model that can accurately classify four common and distinct RHEED patterns present in chalcogenide thin film growth. Its reached accuracy reached 94.9% for single run and 91.2% when averaged over 20 seeds. Our network is able to distinguish the nucleation of three common growth modes encountered in epitaxy, namely Volmer Weber, Stransky-Krastanov and Frank-van der Merwe, potentially enabling future automation of substrate temperature and shutter control informed by RHEED data. The network is material-agnostic and distinguishes the VW process with greater than 98% accuracy but is somewhat more limited in its ability to properly classify roughening and the initiation of Stransky-Krastanov growth. Our findings show that ML techniques can be successfully implemented even in cases where there is no detailed knowledge of growth chemistry providing an avenue towards real-time incorporation of ML to control nanostructure nucleation and thin film morphology.

cond-mat.mtrl-sci↗

arXiv2025

TheUse of Conditional Variational Autoencoders in Generating Stellar Spectra

We present a conditional variational autoencoder (CVAE) that generates stellar spectra covering 4000 $\le$ $T_{\mathrm{eff}$ $\le$ 11,000 K, $2.0 \le \log g \le 5.0$ dex, $-1.5 \le [\mathrm{M}/\mathrm{H}] \le +1.5$ dex, $v\sin i \le 300$ km/s, $\xi_t$ between 0 and 4 km/s, and for any instrumental resolving powers less than 115,000. The spectra can be calculated in the wavelength range 4450-5400 \AA. Trained on a grid of \textsc{SYNSPEC} spectra, the network synthesizes a spectrum in around two orders of magnitude faster than line-by-line radiative transfer. We validate the CVAE on $10^4$ test spectra unseen during training. Pixel-wise statistics yield a median absolute residual of <$1.8\times10^{-3}$ flux units with no wavelength-dependent bias. A residual error map across the parameters plane shows $\langle|\Delta F|\rangle<2\times10^{-3}$ everywhere, and marginal diagnostics versus $T_{\mathrm{eff}}$, $\log g$, $v\sin i$, $\xi_t$, and $[Fe/H]$\ reveal no relevant trends. These results demonstrate that the CVAE can serve as a drop-in, physics-aware surrogate for radiative transfer codes, enabling real-time forward modeling in stellar parameter inference and offering promising tools for spectra synthesis for large astrophysical data analysis.

astro-ph.SR↗

arXiv2024

High Precision Binding Energies from Physics Informed Machine Learning

Twelve physics-informed machine learning models have been trained to model binding energy residuals. Our approach begins with determining the difference between measured experimental binding energies and three different mass models. Then four machine learning approaches are used to train on each energy difference. The most successful ML technique, both in interpolation and extrapolation, is the least squares boosted ensemble of trees. The best model resulting from that technique utilizes eight physical features to model the difference between experimental atomic binding energy values in AME 2012 and the Duflo Zuker mass model. This resulted in a model that fit the training data with a standard deviation of 17 keV and that has a standard deviation of 92 keV when compared all of the values in the AME 2020. The extrapolation capability of each model is discussed, and the accuracy of predicting new mass measurements has also been tested.

nucl-th↗

arXiv2024

Deep Learning application for stellar parameters determination: III- Denoising Procedure

In this third paper in a series, we investigate the need of spectra denoising for the derivation of stellar parameters. We have used two distinct datasets for this work. The first one contains spectra in the range of 4450-5400 {\AA} at a resolution of 42000 and the second in the range of 8400-8800 {\AA} at a resolution of 11500. We constructed two denoising techniques, an autoencoder, and a Principal Component Analysis. Using random Gaussian noise added to synthetic spectra, we have trained a Neural Network to derive the stellar parameters Teff, log g, ve sin i, {\xi}t, and [M/H] of the denoised spectra. We find that, independently of the denoising technique, the stellar parameters accuracy values do not improve once we denoise the synthetic spectra. This is true with and without applying data augmentation to the stellar parameters Neural Network.

astro-ph.SR↗

arXiv2024

Generating stellar spectra using Neural Networks

A new generative technique is presented in this paper that uses Deep Learning to reconstruct stellar spectra based on a set of stellar parameters. Two different Neural Networks were trained allowing the generation of new spectra. First, an autoencoder is trained on a set of BAFGK synthetic data calculated using ATLAS9 model atmospheres and SYNSPEC radiative transfer code. These spectra are calculated in the wavelength range of Gaia RVS between 8 400 and 8 800 {\AA}. Second, we trained a Fully Dense Neural Network to relate the stellar parameters to the Latent Space of the autoencoder. Finally, we linked the Fully Dense Neural Network to the decoder part of the autoencoder and we built a model that uses as input any combination of $T_{eff}$, $\log g$, $v_e \sin i$, [M/H], and $\xi_t$ and output a normalized spectrum. The generated spectra are shown to represent all the line profiles and flux values as the ones calculated using the classical radiative transfer code. The accuracy of our technique is tested using a stellar parameter determination procedure and the results show that the generated spectra have the same characteristics as the synthetic ones.

astro-ph.SR↗

arXiv2023

The surface composition of six newly discovered chemically peculiar stars. Comparison to the HgMn stars $\mu$ Lep and $\beta$ Scl and the superficially normal B star $\nu$ Cap

We report on a detailed abundance study of six bright, mostly southern, slowly rotating late B stars: HD~1279 (B8III), HD~99803 (B9V), HD~123445 (B9V), HD~147550 (B9V), HD~171961 (B8III) and HD~202671 (B5II/III), hitherto reported as normal stars. We compare them to the two classical HgMn stars $\mu$ Lep and $\beta$ Scl and to the superficially normal star, $\nu$ Cap. In the spectra of the six stars, the \ion{Hg}{2} line at 3984 \AA\ line is clearly seen and numerous lines of P, Ti, Mn, Fe, Ga, Sr, Y, and Zr appear to be strong absorbers. A comparison of newly acquired and archival spectra of these objects with a grid of synthetic spectra for selected unblended lines reveals large overabundances of P, Ti, Cr, Mn, Sr, Y, Zr, Ba, Pt and Hg and underabundances of He, Mg, Sc and Ni. The effective temperatures, surface gravities, low projected rotational velocities and the peculiar abundance patterns of the six investigated stars show that they are new chemically peculiar stars, mostly new HgMn stars, and are reclassified as such. The evolutionary status of these stars has been inferred and their ages and masses estimated. The two most massive objects, HD~1279 and HD~202671, might have evolved away from the main-sequence recently, the other stars are main-sequence objects. HD~99803A is a sharp lined HgMn star with grazing eclipses; from TESS and MASCARA photometry we determine an orbital period of $P_{\rm orb} = 26.12022 \pm 0.00004$\,d.

astro-ph.SR↗

arXiv2022

Deep Learning application for stellar parameters determination: II- Application to observed spectra of AFGK stars

In this follow-up paper, we investigate the use of Convolutional Neural Network for deriving stellar parameters from observed spectra. Using hyperparameters determined previously, we have constructed a Neural Network architecture suitable for the derivation of Teff, log g, [M/H], and vesini. The network was constrained by applying it to databases of AFGK synthetic spectra at different resolutions. Then, parameters of A stars from Polarbase, SOPHIE, and ELODIE databases are derived as well as FGK stars from the Spectroscopic Survey of Stars in the Solar Neighbourhood. The network model average accuracy on the stellar parameters are found to be as low as 80 K for Teff , 0.06 dex for log g, 0.08 dex for [M/H], and 3 km/s for vesini for AFGK stars.

astro-ph.SR↗

arXiv2022

Chemically Peculiar Stars in the Open Cluster Stock 2

The recently re-discovered open cluster Stock 2, located roughly 375 pc away and about 400 Myr old, has the potential to be an exciting new testbed for our understanding of stellar evolution. We present results from a spectroscopic campaign to characterize stars near the cluster's main-sequence turnoff; our goal is to identify candidate chemically peculiar stars among the cluster's A stars. We obtained echelle spectra for 64 cluster members with ESPaDOnS on the 3.6-m Canada-France-Hawaii Telescope, Mauna Kea Observatory, USA, and for six stars with SOPHIE on the 1.93-m telescope at the Observatoire de Haute-Provence, France. We complemented these new observations with those of 13 high-mass cluster members from the HARPS-N archive; our overall sample is of 71 stars. We derived the fundamental parameters (Teff, log g, [M/H]) as well as vsini for our sample using the Sliced Inverse Regression (SIR) technique, and then used iSpec to derive individual abundances of 12 chemical species. With these abundance determinations, we identified nine A stars with anomalous levels of Sc, Ca, and other metallic lines. Follow-up observations of these Am candidates with a known age can transform them into benchmarks for evolutionary models that include atomic diffusion and help build a better understanding of the complex interactions between macroscopic and microscopic processes in stellar interiors.

astro-ph.SR↗

arXiv2022

Deep Learning application for stellar parameters determination: I- Constraining the hyperparameters

Machine Learning is an efficient method for analyzing and interpreting the increasing amount of astronomical data that is available. In this study, we show, a pedagogical approach that should benefit anyone willing to experiment with Deep Learning techniques in the context of stellar parameters determination. Utilizing the Convolutional Neural Network architecture, we give a step by step overview of how to select the optimal parameters for deriving the most accurate values for the stellar parameters of stars: T$_{\rm{eff}}$, $\log g$, [X/H], and $v_e \sin i$. Synthetic spectra with random noise were used to constrain this method and to mimic the observations. We found that each stellar parameter requires a different combination of network hyperparameters and the maximum accuracy reached depends on this combination, as well as, the Signal to Noise ratio of the observations, and the architecture of the network. We also show that this technique can be applied to other spectral types in different wavelength ranges after the technique has been optimized.

astro-ph.IM↗

arXiv2018

Pushing Automated Abundance Derivations Into the Cool Star Regime: A Test Using Three G and Two K Stars in Praesepe

We present the results of an abundance analysis of three G and two K dwarfs in the Praesepe open cluster based on high-resolution, moderate signal-to-noise-ratio spectra obtained with the ARC 3.5-m Telescope at Apache Point Observatory. Using a Principle Component Analysis and the BACCHUS automated spectral analysis code, we determined stellar parameters and abundances of up to 24 elements for each of our targets, which range in temperature from 6000 to 4600 K. The average derived iron abundance for the three G stars is 0.17+/-0.07 dex, consistent with the 0.12+/-0.04 dex derived by Boesgaard et al. (2013) for their sample of 11 solar-type Praesepe members, which included these G stars. To investigate the efficacy of using automated routines to derive the abundances of cooler main-sequence stars, we compared the abundances of the K dwarfs to those of the G dwarfs. Our abundances are consistent to <=0.1~dex for 13 of the 18 elements we report for all five of the stars, providing more evidence that G and K stars in a given open cluster are chemically homogeneous. The median difference between the mean G and K stars abundances is 0.08+/-0.05 dex, despite serious challenges with the noisier data for the fainter K dwarfs. Our results are encouraging for chemical tagging, as they indicate that it may be possible to use automated abundance determination techniques to identify chemically related main-sequence stars across larger temperature ranges than are usually considered in these experiments.

astro-ph.SR↗

arXiv2018

An empirical recipe for inelastic hydrogen-atom collisions in non-LTE calculations

We investigate the role of hydrogen collisions in NLTE spectral line synthesis, and introduce a new general empirical recipe to determine inelastic charge transfer (CT) and bound-bound hydrogen collisional rates. This recipe is based on fitting the energy functional dependence of published quantum collisional rate coefficients of several neutral elements (BeI, NaI, MgI, AlI, SiI and CaI) using simple polynomial equations. We perform thorough NLTE abundance calculation tests using our method for four different atoms, Na, Mg, Al and Si, for a broad range of stellar parameters. We then compare the results to calculations computed using the published quantum rates for all the corresponding elements. We also compare to results computed using excitation collisional rates via the commonly used Drawin equation for different fudge factors, SH, applied. We demonstrate that our proposed method is able to reproduce the NLTE abundance corrections performed with the quantum rates for different spectral types and metallicities for representative NaI and AlI lines to within $\le$0.05 dex and %\le%0.03 dex, respectively. For MgI and SiI lines, the method performs better for the cool giants and dwarfs, while larger discrepancies up to 0.2 dex could be obtained for some lines for the subgiants and warm dwarfs. We obtained larger NLTE correction differences between models incorporating Drawin rates relative to the quantum models by up to 0.4 dex. These discrepancies are potentially due to ignoring either or both CT and ionization collisional processes by hydrogen in our Drawin models. Our empirical fitting method performs well in its ability to reproduce, within narrow uncertainties, the abundance corrections computed with models incorporating quantum collisional rates. It could possibly be extended to other transitions or in the absence of published quantum calculations, to other elements as well.

astro-ph.SR↗