Nils Candebat, Giuseppe Germano Sacco, Laura Magrini, Francesco Belfiore, Mathieu Van-der-Swaelmen, Stefano Zibetti
Context: New spectroscopic surveys will increase the number of astronomical objects requiring characterization by over tenfold.. Machine learning tools are required to address this data deluge in a fast and accurate fashion. Most machine learning algorithms can not estimate error directly, making them unsuitable for reliable science. Aims: We aim to train a supervised deep-learning algorithm tailored for high-resolution observational stellar spectra. This algorithm accurately infer precise estimates while providing coherent estimates of uncertainties by leveraging information from both the neural network and the spectra. Methods: We train a conditional Invertible Neural Network (cINN) on observational spectroscopic data obtained from the GIRAFFE spectrograph (HR10 and HR21 setups) within the Gaia-ESO survey. A key features of cINN is its ability to produce the Bayesian posterior distribution of parameters for each spectrum. By analyzing this distribution, we inferred parameters and their uncertainties. Several tests have been applied to study how parameters and errors are estimated. Results: We achieved an accuracy of 28K in $T_{text{eff}}$, 0.06 dex in $log g$, 0.03 dex in $[text{Fe/H}]$, and between 0.05 dex and 0.17 dex for the other abundances for high quality spectra. Accuracy remains stable with low signal-to-noise ratio spectra. The uncertainties obtained are well within the same order of magnitude. The network accurately reproduces astrophysical relationships both on the scale of the Milky Way and within smaller star clusters. We created a table containing the new parameters generated by our cINN. Conclusion: This neural network represents a compelling proposition for future astronomical surveys. These coherent derived uncertainties make it possible to reuse these estimates in other works as Bayesian priors and thus present a solid basis for future work.
{"title":"Inferring stellar parameters and their uncertainties from high-resolution spectroscopy using invertible neural networks","authors":"Nils Candebat, Giuseppe Germano Sacco, Laura Magrini, Francesco Belfiore, Mathieu Van-der-Swaelmen, Stefano Zibetti","doi":"arxiv-2409.10621","DOIUrl":"https://doi.org/arxiv-2409.10621","url":null,"abstract":"Context: New spectroscopic surveys will increase the number of astronomical\u0000objects requiring characterization by over tenfold.. Machine learning tools are\u0000required to address this data deluge in a fast and accurate fashion. Most\u0000machine learning algorithms can not estimate error directly, making them\u0000unsuitable for reliable science. Aims: We aim to train a supervised deep-learning algorithm tailored for\u0000high-resolution observational stellar spectra. This algorithm accurately infer\u0000precise estimates while providing coherent estimates of uncertainties by\u0000leveraging information from both the neural network and the spectra. Methods: We train a conditional Invertible Neural Network (cINN) on\u0000observational spectroscopic data obtained from the GIRAFFE spectrograph (HR10\u0000and HR21 setups) within the Gaia-ESO survey. A key features of cINN is its\u0000ability to produce the Bayesian posterior distribution of parameters for each\u0000spectrum. By analyzing this distribution, we inferred parameters and their\u0000uncertainties. Several tests have been applied to study how parameters and\u0000errors are estimated. Results: We achieved an accuracy of 28K in $T_{text{eff}}$, 0.06 dex in\u0000$log g$, 0.03 dex in $[text{Fe/H}]$, and between 0.05 dex and 0.17 dex for\u0000the other abundances for high quality spectra. Accuracy remains stable with low\u0000signal-to-noise ratio spectra. The uncertainties obtained are well within the\u0000same order of magnitude. The network accurately reproduces astrophysical\u0000relationships both on the scale of the Milky Way and within smaller star\u0000clusters. We created a table containing the new parameters generated by our\u0000cINN. Conclusion: This neural network represents a compelling proposition for\u0000future astronomical surveys. These coherent derived uncertainties make it\u0000possible to reuse these estimates in other works as Bayesian priors and thus\u0000present a solid basis for future work.","PeriodicalId":501068,"journal":{"name":"arXiv - PHYS - Solar and Stellar Astrophysics","volume":"40 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142257982","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The Serkowski relation is the cornerstone of studies of starlight polarization as a function of wavelength. Although empirical, its extensive use since its inception to describe polarization induced by interstellar dust has elevated the relation to the status of an indisputable "law", serving as the benchmark for validating interstellar dust grain models. We revisit the effects of the 3D structure of the interstellar medium (ISM) on the wavelength dependence of interstellar polarization. We use analytical models to show how the wavelength dependence of both the polarization fraction and direction is affected by the presence of multiple clouds along the line of sight (LOS), accounting for recent developments in dust distribution modelling and utilizing an expanded archive of stellar polarization measurements. We highlight concrete examples of stars whose polarization profiles are severely affected by LOS variations of the dust grain and magnetic field properties, and we provide a recipe to accurately fit multiple cloud Serkowski models to such cases. We present, for the first time, compelling observational evidence that the 3D structure of the magnetized ISM often results to the violation of the Serkowski relation. We show that 3D effects impact interstellar cloud parameters derived from Serkowski fits. In particular, the dust size distribution in single - cloud sightlines may differ from analyses that ignore 3D effects, with important implications for dust modelling in the Galaxy. Our results suggest that multiband stellar polarization measurements offer an independent probe of the LOS variations of the magnetic field, constituting a valuable new tool for the 3D cartography of the ISM. We caution that, unless 3D effects are explicitly accounted for, a poor fit to the Serkowski relation does not, by itself, constitute conclusive evidence that a star is intrinsically polarized.
谢尔科夫斯基关系是研究星光极化与波长函数关系的基石。尽管它是经验性的,但自诞生以来,它被广泛用于描述星际尘埃引起的偏振,这使得该关系上升为无可争议的 "定律",成为验证星际尘埃粒子模型的基准。我们重新审视了星际介质(ISM)的三维结构对星际偏振波长依赖性的影响。我们使用分析模型来说明偏振分数和方向的波长依赖性如何受到沿视线(LOS)存在的多个云团的影响,同时考虑到尘粒分布模型的最新发展,并利用扩充的恒星偏振测量档案。我们强调了一些恒星的具体例子,这些恒星的偏振剖面受到视线中尘粒和磁场特性变化的严重影响,我们还提供了精确拟合多云 Serkowski 模型的方法。我们首次提出了令人信服的观测证据,表明磁化 ISM 的三维结构经常导致违反塞尔科夫斯基相关性。我们表明,三维效应会影响由塞尔科夫斯基拟合得出的星际云参数。特别是,单个云视线中的尘埃大小分布可能与忽略三维效应的分析不同,这对银河系的尘埃建模具有重要影响。我们的研究结果表明,多波段恒星偏振测量可以独立探测 LOS 磁场的变化,是绘制 ISM 三维地图的宝贵新工具。我们要提醒的是,除非三维效应被明确考虑在内,否则与塞尔科夫斯基关系的拟合效果不佳本身并不能构成恒星本质上偏振的确凿证据。
{"title":"3D ISM structure challenges the Serkowski relation","authors":"Nikolaos Mandarakas, Konstantinos Tassis, Raphael Skalidis","doi":"arxiv-2409.10317","DOIUrl":"https://doi.org/arxiv-2409.10317","url":null,"abstract":"The Serkowski relation is the cornerstone of studies of starlight\u0000polarization as a function of wavelength. Although empirical, its extensive use\u0000since its inception to describe polarization induced by interstellar dust has\u0000elevated the relation to the status of an indisputable \"law\", serving as the\u0000benchmark for validating interstellar dust grain models. We revisit the effects\u0000of the 3D structure of the interstellar medium (ISM) on the wavelength\u0000dependence of interstellar polarization. We use analytical models to show how\u0000the wavelength dependence of both the polarization fraction and direction is\u0000affected by the presence of multiple clouds along the line of sight (LOS),\u0000accounting for recent developments in dust distribution modelling and utilizing\u0000an expanded archive of stellar polarization measurements. We highlight concrete\u0000examples of stars whose polarization profiles are severely affected by LOS\u0000variations of the dust grain and magnetic field properties, and we provide a\u0000recipe to accurately fit multiple cloud Serkowski models to such cases. We\u0000present, for the first time, compelling observational evidence that the 3D\u0000structure of the magnetized ISM often results to the violation of the Serkowski\u0000relation. We show that 3D effects impact interstellar cloud parameters derived\u0000from Serkowski fits. In particular, the dust size distribution in single -\u0000cloud sightlines may differ from analyses that ignore 3D effects, with\u0000important implications for dust modelling in the Galaxy. Our results suggest\u0000that multiband stellar polarization measurements offer an independent probe of\u0000the LOS variations of the magnetic field, constituting a valuable new tool for\u0000the 3D cartography of the ISM. We caution that, unless 3D effects are\u0000explicitly accounted for, a poor fit to the Serkowski relation does not, by\u0000itself, constitute conclusive evidence that a star is intrinsically polarized.","PeriodicalId":501068,"journal":{"name":"arXiv - PHYS - Solar and Stellar Astrophysics","volume":"14 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142258046","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We present a study of the detached eclipsing binary TV~Mon using spectra from the LAMOST medium-resolution survey and ASAS-SN, CoRoT photometry. We applied multiple-epochs spectral fitting to derive RV and spectral parameters. The analysis of eclipses in CoRoT data told us relative sizes of the stellar components and almost edge-on circular orbit. Combining spectral and photometrical solution we estimated masses and radii of the components: $M_{A,B}=2.063pm0.033,~0.218pm0.004~M_odot$, $R_{A,B}=2.427pm0.014,~2.901pm0.016~R_odot$. SED analysis and Gaia parallax allowed us to get estimation of temperatures $T_{A,B}=7624^{+194}_{-174},~5184^{+130}_{-123}$ K and distance $d=907pm11$ pc. We identified three $delta$ Scuti type pulsation frequencies in primary component, while we also suspect TV~Mon having a long period variability with period $P_{rm long}sim128$ days and spot activity in secondary component. This system experienced intensive mass transfer and mass ratio reversal in the past, currently showing no signs of mass transfer in the spectra. The low mass component will lose its outer envelope and shrink to the helium white dwarf, which mass and orbital period are in good agreement with evolutionary models predictions.
{"title":"TV Mon - post mass transfer Algol type binary with $δ$ Scuti pulsations in primary component","authors":"Mikhail Kovalev, Zhenwei Li, Jianping Xiong, Azizbek Matekov, Zhang Bo, Xuefei Chen, Zhanwen Han","doi":"arxiv-2409.09902","DOIUrl":"https://doi.org/arxiv-2409.09902","url":null,"abstract":"We present a study of the detached eclipsing binary TV~Mon using spectra from\u0000the LAMOST medium-resolution survey and ASAS-SN, CoRoT photometry. We applied\u0000multiple-epochs spectral fitting to derive RV and spectral parameters. The\u0000analysis of eclipses in CoRoT data told us relative sizes of the stellar\u0000components and almost edge-on circular orbit. Combining spectral and\u0000photometrical solution we estimated masses and radii of the components:\u0000$M_{A,B}=2.063pm0.033,~0.218pm0.004~M_odot$,\u0000$R_{A,B}=2.427pm0.014,~2.901pm0.016~R_odot$. SED analysis and Gaia parallax\u0000allowed us to get estimation of temperatures\u0000$T_{A,B}=7624^{+194}_{-174},~5184^{+130}_{-123}$ K and distance $d=907pm11$\u0000pc. We identified three $delta$ Scuti type pulsation frequencies in primary\u0000component, while we also suspect TV~Mon having a long period variability with\u0000period $P_{rm long}sim128$ days and spot activity in secondary component.\u0000This system experienced intensive mass transfer and mass ratio reversal in the\u0000past, currently showing no signs of mass transfer in the spectra. The low mass\u0000component will lose its outer envelope and shrink to the helium white dwarf,\u0000which mass and orbital period are in good agreement with evolutionary models\u0000predictions.","PeriodicalId":501068,"journal":{"name":"arXiv - PHYS - Solar and Stellar Astrophysics","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142257986","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
S. C. Susarla, A. Chalumeau, C. Tiburzi, E. F. Keane, J. P. W. Verbiest, J. S. Hazboun, M. A. Krishnakumar, F. Iraci, G. M. Shaifullah, A. Golden, A. S. Bak Nielsen, J. Donner, J. M. Grießmeier, M. J. Keith, S. Osłowski, N. K. Porayko, M. Serylak, J. M. Anderson, M. Brüggen, B. Ciardi, R. J. Dettmar, M. Hoeft, J. Künsemöller, D. Schwarz, C. Vocks
High-precision pulsar timing is highly dependent on precise and accurate modeling of any effects that impact the data. It was shown that commonly used Solar Wind models do not accurately account for variability in the amplitude of the Solar wind on both short and long time scales. In this study, we test and validate a new, cutting-edge Solar wind modeling method included in the texttt{enterprise} software suite through extended simulations, and we apply it to investigate temporal variability in LOFAR data. Our model testing scheme in itself provides an invaluable asset for pulsar timing array (PTA) experiments. As improperly accounting for the solar wind signature in pulsar data can induce false-positive signals, it is of fundamental importance to include in any such investigations. We employ a Bayesian approach utilizing a continuously varying Gaussian process to model the solar wind referred to as Solar Wind Gaussian Process (SWGP). We conduct noise analysis on eight pulsars from the LOFAR dataset with most pulsars having a timespan of $sim 11$ years encompassing one full solar activity cycle. Our analysis reveals a strong correlation between the electron density at 1 AU and the ecliptic latitude (ELAT) of the pulsar. Pulsars with $|ELAT|< 3^{circ}$ exhibit significantly higher average electron densities. We observe distinct temporal patterns in electron densities in different pulsars. In particular, pulsars within $|ELAT|<