Context.
In June 2022,
Gaia
DR3 provided the astronomy community with about one million spectra from the Radial Velocity Spectrometer (RVS) covering the CaII triplet region. In the next
Gaia
data ...releases, we anticipate the number of RVS spectra to successively increase from several 10 million spectra to eventually more than 200 million spectra. Thus, stellar spectra are projected to be produced on an ‘industrial scale’, with numbers well above those for current and anticipated ground-based surveys. However, one-third of the published spectra have 15 ≤
S /N
≤ 25 per pixel such that they pose problems for classical spectral analysis pipelines, and therefore, alternative ways to tap into these large datasets need to be devised.
Aims.
We aim to leverage the versatility and capabilities of machine learning techniques for supercharged stellar parametrisation by combining
Gaia
-RVS spectra with the full set of
Gaia
products and high-resolution, high-quality ground-based spectroscopic reference datasets.
Methods.
We developed a hybrid convolutional neural network (CNN) that combines the
Gaia
DR3 RVS spectra, photometry (G, G_BP, G_RP), parallaxes, and XP coefficients to derive atmospheric parameters (
T
eff
, log(g) as well as overall M/H) and chemical abundances (Fe/H and
α
/M). We trained the CNN with a high-quality training sample based on APOGEE DR17 labels.
Results.
With this CNN, we derived homogeneous atmospheric parameters and abundances for 886 080 RVS stars that show remarkable precision and accuracy compared to external datasets (such as GALAH and asteroseismology). The CNN is robust against noise in the RVS data, and we derive very precise labels down to S/N =15. We managed to characterise the
α
/M - M/H bimodality from the inner regions to the outer parts of the Milky Way, which has never been done using RVS spectra or similar datasets.
Conclusions.
This work is the first to combine machine learning with such diverse datasets and paves the way for large-scale machine learning analysis of
Gaia
-RVS spectra from future data releases. Large, high-quality datasets can be optimally combined thanks to the CNN, thereby realising the full power of spectroscopy, astrometry, and photometry.
Accurate atmospheric parameters and chemical composition of stars play a vital role in characterizing physical parameters of exoplanetary systems and understanding of their formation. A full ...asteroseismic characterization of a star is also possible if its main atmospheric parameters are known. The NASA Transiting Exoplanet Survey Satellite (TESS) space telescope will play a very important role in searching of exoplanets around bright stars and stellar asteroseismic variability research. We have observed all 302 bright (V < 8 mag) and cooler than F5 spectral class stars in the northern TESS continuous viewing zone with a 1.65 m telescope at the Mol tai Astronomical Observatory of Vilnius University and the high-resolution Vilnius University Echelle Spectrograph. We uniformly determined the main atmospheric parameters, ages, orbital parameters, velocity components, and precise abundances of 24 chemical species (C(C2), N(CN), O i, Na i, Mg i, Al i, Si i, Si ii, Ca i, Ca ii, Sc i, Sc ii, Ti i, Ti ii, V i, Cr i, Cr ii, Mn i, Fe i, Fe ii, Co i, Ni i, Cu i, and Zn i) for 277 slowly rotating single stars in the field. About 83% of the sample stars exhibit the Mg/Si ratios greater than 1.0 and may potentially harbor rocky planets in their systems.
Abstract
In fulfilling the aims of the planetary and asteroseismic research missions, such as that of the NASA Transiting Exoplanet Survey Satellite (TESS) space telescope, accurate stellar ...atmospheric parameters and a detailed chemical composition are required as inputs. We have observed high-resolution spectra for all 848 bright (
V
< 8 mag) stars that are cooler than F5 spectral class in the area up to 12 deg surrounding the northern TESS continuous viewing zone and uniformly determined the main atmospheric parameters, ages, orbital parameters, velocity components, and precise abundances of up to 24 chemical species (C(C
2
), N(CN), O
i
, Na
i
, Mg
i
, Al
i
, Si
i
, Si
ii
, Ca
i
, Ca
ii
, Sc
i
, Sc
ii
, Ti
i
, Ti
ii
, V
i
, Cr
i
, Cr
ii
, Mn
i
, Fe
i
, Fe
ii
, Co
i
, Ni
i
, Cu
i
, and Zn
i
) for 740 slowly rotating stars. The analysis of 25 planet-hosting stars in our sample drove us to the following conclusions: the dwarf stars hosting high-mass planets are more metal rich than those with low-mass planets. We find slightly negative C/O and Mg/Si slopes toward the stars with high-mass planets. All the low-mass planet hosts in our sample show positive ΔEl/Fe versus condensation temperature slopes, in particular, the star with the largest number of various planets. The high-mass planet hosts have a diversity of slopes, but in more metal-rich, older, and cooler stars, the positive elemental abundance slopes are more common.
Context.
To take full advantage of upcoming large-scale spectroscopic surveys, it will be necessary to parameterize millions of stellar spectra in an efficient way. Machine learning methods, ...especially convolutional neural networks (CNNs), will be among the main tools geared at achieving this task.
Aims.
We aim to prepare the groundwork for machine learning techniques for the next generation of spectroscopic surveys, such as 4MOST and WEAVE. Our goal is to show that CNNs can predict accurate stellar labels from relevant spectral features in a physically meaningful way. The predicted labels can be used to investigate properties of the Milky Way galaxy.
Methods.
We built a neural network and trained it on GIRAFFE spectra with their associated stellar labels from the sixth internal
Gaia
-ESO data release. Our network architecture contains several convolutional layers that allow the network to identify absorption features in the input spectra. The internal uncertainty was estimated from multiple network models. We used the t-distributed stochastic neighbor embedding tool to remove bad spectra from our training sample.
Results.
Our neural network is able to predict the atmospheric parameters
T
eff
and log(
g
) as well as the chemical abundances Mg/Fe, Al/Fe, and Fe/H for 36 904 stellar spectra. The training precision is 37 K for
T
eff
, 0.06 dex for log(
g
), 0.05 dex for Mg/Fe, 0.08 dex for Al/Fe, and 0.04 dex for Fe/H. Network gradients reveal that the network is inferring the labels in a physically meaningful way from spectral features. We validated our methodology using benchmark stars and recovered the properties of different stellar populations in the Milky Way galaxy.
Conclusions.
Such a study provides very good insights into the application of machine learning for the analysis of large-scale spectroscopic surveys, such as WEAVE and 4MOST Milky Way disk and bulge low- and high-resolution (4MIDABLE-LR and -HR). The community will have to put substantial efforts into building proactive training sets for machine learning methods to minimize any possible systematics.
Context. Young open clusters (ages of less than 200 Myr) have been observed to exhibit several peculiarities in their chemical compositions. These anomalies include a slightly sub-solar iron content, ...super-solar abundances of some atomic species (e.g. ionised chromium), and atypical enhancements of Ba/Fe, with values up to ~0.7 dex. Regarding the behaviour of the other s-process elements like yttrium, zirconium, lanthanum, and cerium, there is general disagreement in the literature: some authors claim that they follow the same trend as barium, while others find solar abundances at all ages. Aims. In this work we expand upon our previous analysis of a sample of five young open clusters (IC 2391, IC 2602, IC 4665, NGC 2516, and NGC 2547) and one star-forming region (NGC 2264), with the aim of determining abundances of different neutron-capture elements, mainly Cu I, Sr I, Sr II, Y II, Zr II, Ba II, La II, and Ce II. For NGC 2264 and NGC 2547 we present the measurements of these elements for the first time. Methods. We analysed high-resolution, high signal-to-noise spectra of 23 solar-type stars observed within the Gaia-ESO survey. After a careful selection, we derived abundances of isolated and clean lines via spectral synthesis computations and in a strictly differential way with respect to the Sun. Results. We find that our clusters have solar Cu/Fe within the uncertainties, while we confirm that Ba/Fe is super-solar, with values ranging from +0.22 to +0.64 dex. Our analysis also points to a mild enhancement of Y, with Y/Fe ratios covering values between 0 and +0.3 dex. For the other s-process elements we find that X/Fe ratios are solar at all ages. Conclusions. It is not possible to reconcile the anomalous behaviour of Ba and Y at young ages with standard stellar yields and Galactic chemical evolution model predictions. We explore different possible scenarios related to the behaviour of spectral lines, from the dependence on the different ionisation stages and the sensitivity to the presence of magnetic fields (through the Landé factor) to the first ionisation potential effect. We also investigate the possibility that they may arise from alterations of the structure of the stellar photosphere due to the increased levels of stellar activity that affect the spectral line formation, and consequently the derived abundances. These effects seem to be stronger in stars at ages of less than ~ 100 Myr. However, we are still unable to explain these enhancements, and the Ba puzzle remains unsolved. With the present study we suggest that other elements, for example Sr, Zr, La, and Ce, might be more reliable tracer of the s-process at young ages, and we strongly encourage further critical observations.
ABSTRACT
We present an empirical model of age-dependent photospheric lithium depletion, calibrated using a large homogeneously analysed sample of 6200 stars in 52 open clusters, with ages from 2 to ...6000 Myr and −0.3 < Fe/H < 0.2, observed in the Gaia-ESO spectroscopic survey. The model is used to obtain age estimates and posterior age probability distributions from measurements of the Li i 6708 Å equivalent width for individual (pre) main-sequence stars with 3000 < Teff/K < 6500, a domain where age determination from the HR diagram is either insensitive or highly model-dependent. In the best cases, precisions of 0.1 dex in log age are achievable; even higher precision can be obtained for coeval groups and associations where the individual age probabilities of their members can be combined. The method is validated on a sample of exoplanet-hosting young stars, finding agreement with claimed young ages for some, but not others. We obtain better than 10 per cent precision in age, and excellent agreement with published ages, for seven well-studied young moving groups. The derived ages for young clusters (<1 Gyr) in our sample are also in good agreement with their training ages, and consistent with several published model-insensitive lithium depletion boundary ages. For older clusters, there remain systematic age errors that could be as large as a factor of 2. There is no evidence to link these errors to any strong systematic metallicity dependence of (pre) main-sequence lithium depletion, at least in the range −0.29 < Fe/H < 0.18. Our methods and model are provided as software – ‘Empirical AGes from Lithium Equivalent widthS’ (eagles).
We present the detection and characterization of the full-orbit phase curve and secondary eclipse of the ultra-hot Jupiter WASP-33b at optical wavelengths, along with the pulsation spectrum of the ...host star. We analyzed data collected by the Transiting Exoplanet Survey Satellite (TESS) in sector 18. WASP-33b belongs to a very short list of highly irradiated exoplanets that were discovered from the ground and were later visited by TESS. The host star of WASP-33b is of δ Scuti-type and shows nonradial pulsations in the millimagnitude regime, with periods comparable to the period of the primary transit. These completely deform the photometric light curve, which hinders our interpretations. By carrying out a detailed determination of the pulsation spectrum of the host star, we find 29 pulsation frequencies with a signal-to-noise ratio higher than 4. After cleaning the light curve from the stellar pulsations, we confidently report a secondary eclipse depth of 305.8 ± 35.5 parts-per-million (ppm), along with an amplitude of the phase curve of 100.4 ± 13.1 ppm and a corresponding westward offset between the region of maximum brightness and the substellar point of 28.7 ± 7.1 degrees, making WASP-33b one of the few planets with such an offset found so far. Our derived Bond albedo, AB = 0.369 ± 0.050, and heat recirculation efficiency, ɛ = 0.189 ± 0.014, confirm again that he behavior of WASP-33b is similar to that of other hot Jupiters, despite the high irradiation received from its host star. By connecting the amplitude of the phase curve to the primary transit and depths of the secondary eclipse, we determine that the day- and nightside brightness temperatures of WASP-33b are 3014 ± 60 K and 1605 ± 45 K, respectively. From the detection of photometric variations due to gravitational interactions, we estimate a planet mass of MP = 2.81 ± 0.53 MJ. Based on analyzing the stellar pulsations in the frame of the planetary orbit, we find no signals of star-planet interactions.
We present an atmospheric transmission spectrum of the ultra-hot Jupiter WASP-76 b by analyzing archival data obtained with the Space Telescope Imaging Spectrograph (STIS) on board the
Hubble
Space ...Telescope (HST). The dataset spans three transits, two with a wavelength coverage between 2900 and 5700 Å, and the third one between 5250 and 10 300 Å. From the one-dimensional, time dependent spectra we constructed white and chromatic light curves, the latter with typical integration band widths of ~200 Å. We computed the wavelength dependent planet-to-star radii ratios taking into consideration WASP-76’s companion. The resulting transmission spectrum of WASP-76 b is dominated by a spectral slope of increasing opacity towards shorter wavelengths of amplitude of about three scale heights under the assumption of planetary equilibrium temperature. If the slope is caused by Rayleigh scattering, we derive a lower limit to the temperature of ~870 K. Following-up on previous detection of atomic sodium derived from high resolution spectra, we re-analyzed HST data using narrower bands centered around sodium. From an atmospheric retrieval of this transmission spectrum, we report evidence of sodium at 2.9
σ
significance. In this case, the retrieved temperature at the top of the atmosphere (10
−5
bar) is 2300
−392
+412
K. We also find marginal evidence for titanium hydride. However, additional high resolution ground-based data are required to confirm this discovery.
Context. New space missions, such as NASA TESS or ESA PLATO, will focus on bright stars, which have been largely ignored by modern large surveys, especially in the northern sky. Spectroscopic ...information is of paramount importance in characterising the stars and analysing planets possibly orbiting them, and in studying the Galactic disc evolution. Aims. The aim of this work was to analyse all bright (V < 8 mag) F, G, and K dwarf stars using high-resolution spectra in the selected sky fields near the northern celestial pole. Methods. The observations were carried out with the 1.65 m diameter telescope at the Molėtai Astronomical Observatory and a fibre-fed high-resolution spectrograph covering a full visible wavelength range (4000–8500 Å). The atmospheric parameters were derived using the classical equivalent width approach while the individual chemical element abundances were determined from spectral synthesis. For both tasks the one-dimensional plane-parallel LTE MARCS stellar model atmospheres were applied. The NLTE effects for the majority of elemental abundances in our sample were negligible; however, we did calculate the NLTE corrections for the potassium abundances, as they were determined from the large 7698.9 Å line. For manganese and copper we have accounted for a hyperfine splitting. Results. We determined the main atmospheric parameters, kinematic properties, orbital parameters, and stellar ages for 109 newly observed stars and chemical abundances of Na I, Mg I, Al I, Si I, Si II, S I, K I, Ca I, Ca II, Sc I, Sc II, Ti I, Ti II, V I, Cr I, Cr II, Mn I, Fe I, Fe II, Co I, Ni I, Cu I, and Zn I for 249 F, G, and K dwarf stars observed in the present study and in our previous study. The Mg I/Fe I ratio was adopted to define the thin-disc (α-poor) and thick-disc (α-rich) stars in our sample. We explored the behaviour of 21 chemical species in the El/Fe I versus Fe I/H and El/Fe I versus age planes, and compared the results with the latest Galactic chemical evolution models. We also explored El/Fe I gradients according to the mean Galactocentric distances and maximum height above the Galactic plane. Conclusions. We found that in the Galactic thin-disc El/Fe I ratios of α-elements and aluminium have a positive trend with respect to age while the trend of Mn is clearly negative. Abundances of other species do not display significant trends. While the current theoretical models are able to reproduce the generic trends of the elements, they often seem to overestimate or underestimate the observational abundances. We found that the α-element and zinc abundances have slightly positive or flat radial and vertical gradients, while gradients for the odd-Z element Na, K, V, and Mn abundances are negative.
The discovery of lithium-rich giants contradicts expectations from canonical stellar evolution. Here we report on the serendipitous discovery of 20 Li-rich giants observed during the Gaia-ESO Survey, ...which includes the first nine Li-rich giant stars known towards the CoRoT fields. Most of our Li-rich giants have near-solar metallicities and stellar parameters consistent with being before the luminosity bump. This is difficult to reconcile with deep mixing models proposed to explain lithium enrichment, because these models can only operate at later evolutionary stages: at or past the luminosity bump. In an effort to shed light on the Li-rich phenomenon, we highlight recent evidence of the tidal destruction of close-in hot Jupiters at the sub-giant phase. We note that when coupled with models of planet accretion, the observed destruction of hot Jupiters actually predicts the existence of Li-rich giant stars, and suggests that Li-rich stars should be found early on the giant branch and occur more frequently with increasing metallicity. A comprehensive review of all known Li-rich giant stars reveals that this scenario is consistent with the data. However, more evolved or metal-poor stars are less likely to host close-in giant planets, implying that their Li-rich origin requires an alternative explanation, likely related to mixing scenarios rather than external phenomena.