In this third article 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 4,450–5,400 Å at a resolution of 42,000, and the second in the range of 8,400–8,800 Å at a resolution of 11,500. 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{T}_{{\rm{eff}}}, logg\log g, vesini{v}_{{\rm{e}}}\sin i, ξt{\xi }_{{\rm{t}}}, and [M/H] of the denoised spectra. We find that, independently of the denoising technique, the accuracy values of stellar parameters do not improve once we denoise the synthetic spectra. This is true with and without applying data augmentation to the stellar parameters neural network.

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.

Abstract Machine learning is an efficient method for analysing and interpreting the increasing amount of astronomical data that are 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 parameter determination. Using 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 eff {T}_{{\rm{eff}}} , log g \log g , [M/H], and v e sin i {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.

We present a method for deriving stellar fundamental parameters. It is based on a regularized sliced inverse regression (RSIR). We first tested it on noisy synthetic spectra of A, F, G, and K-type stars, and inverted simultaneously their atmospheric fundamental parameters: Teff, log g, [M/H] and vsini. Different learning databases were calculated using a range of sampling in Teff, log g, vsini, and [M/H]. Combined with a principal component analysis (PCA) nearest neighbors (NN) search, the size of the learning database is reduced. A Tikhonov regularization is applied, given the ill-conditioning of SIR. For all spectral types, decreasing the size of the learning database allowed us to reach internal accuracies better than PCA-based NN-search using larger learning databases. For each analyzed parameter, we have reached internal errors that are smaller than the sampling step of the parameter. We have also applied the technique to a sample of observed FGK and A stars. For a selection of well studied stars, the inverted parameters are in agreement with the ones derived in previous studies. The RSIR inversion technique, complemented with PCA pre-processing proves to be efficient in estimating stellar parameters of A, F, G, and K stars.

This short review aims to clarify upon the origins of the so-called Eddington-Barbier relationships, which relate the emergent specific intensity and the flux to the photospheric source function at specific optical depths. Here we discuss about the assumptions behind the original derivation of Barbier (1943). We also point to the fact that Milne had already formulated these two relations in 1921.

Our aim is to evaluate fundamental parameters from the analysis of the electromagnetic spectra of stars. We may use 10 3 -10 5 spectra; each spectrum being a vector with 10 2 -10 4 coordinates. We thus face the so-called “curse of dimensionality”. We look for a method to reduce the size of this data-space, keeping only the most relevant information. As a reference method, we use principal component analysis (PCA) to reduce dimensionality. However, PCA is an unsupervised method, therefore its subspace was not consistent with the parameter. We thus tested a supervised method based on Sliced Inverse Regression (SIR), which provides a subspace consistent with the parameter. It also shares analogies with factorial discriminant analysis: the method slices the database along the parameter variation, and builds the subspace which maximizes the inter-slice variance, while standardizing the total projected variance of the data. Nevertheless the performances of SIR were not satisfying in standard usage, because of the non-monotonicity of the unknown function linking the data to the parameter and because of the noise propagation. We show that better performances can be achieved by selecting the most relevant directions for parameter inference. Preliminary tests are performed on synthetic pseudo-line profiles plus noise. Using one direction, we show that compared to PCA, the error associated with SIR is 50% smaller on a non-linear parameter, and 70% smaler on a linear parameter. Moreover, using a selected direction, the error is 80% smaller for a non-linear parameter, and 95% smaller for a linear parameter.

This article presents an on-line tool and its accompanying software resources for the numerical solution of basic radiation transfer out of local thermodynamic equilibrium (LTE). State-of-the-art stationary iterative methods such as Accelerated Λ-iteration and Gauss–Seidel schemes, using a short characteristics-based formal solver are used. We also comment on typical numerical experiments associated to the basic non-LTE radiation problem. These resources are intended for the largest use and benefit, in support to more classical radiation transfer lectures usually given at the Master level.

We report laboratory measurements of the Zeeman response of lines in the 0-0 Wing-Ford band of the F-X system (lambda similar to 1 mu m) of FeH, measured in magnetic fields 0.3 - 0.5 Tesla. New Lando factors are used to deduce the magnetic field in sunspots from Stokes V profiles recorded at the solar telescope THEMIS. The magnetic field deduced from atomic lines (Ti, Fe) is slightly higher than that found from FeH.

We present an overview of the conceptual design of the data handling unit of the ECS, the Control System for the European Solar Telescope (EST). We will focus on describing the critical requirements for this unit resulting from the overall design of the telescope, together with its architecture and the results of the feasibility analysis carried out to date.

We gratefully acknowledge financial support from the French agency ANR, and from the CNRS (PNPS programme) for this work.

We introduce the concepts for the control and data handling systems of the European Solar Telescope (EST), the main functional and technical requirements for the definition of these systems, and the outcomes from the trade-off analysis to date. Concerning the telescope control, EST will have performance requirements similar to those of current medium-sized night-time telescopes. On the other hand, the science goals of EST require the simultaneous operation of three instruments and of a large number of detectors. This leads to a projected data flux that will be technologically challenging and exceeds that of most other astronomical projects. We give an overview of the reference design of the control and data handling systems for the EST to date, focusing on the more critical and innovative aspects resulting from the overall design of the telescope.

This study concerns the fast and accurate solution of the line radiation transfer problem, under non-LTE conditions. We propose and evaluate an alternative iterative scheme to the classical ALI-Jacobi method, and to the more recently proposed Gauss-Seidel and successive over-relaxation (GS/SOR) schemes. Our study is indeed based on applying a preconditioned bi-conjugate gradient method (BiCG-P). Standard tests, in 1D plane parallel geometry and in the frame of the two-level atom model with monochromatic scattering are discussed. Rates of convergence between the previously mentioned iterative schemes are compared, as are their respective timing properties. The smoothing capability of the BiCG-P method is also demonstrated.

The vast majority of recent advances in the field of numerical radiative transfer relies on approximate operator methods better known in astrophysics as Accelerated Lambda-Iteration (ALI). A superior class of iterative schemes, in term of rates of convergence, such as Gauss-Seidel and Successive Overrelaxation methods were therefore quite naturally introduced in the field of radiative transfer by Trujillo Bueno & Fabiani Bendicho (1995); it was thoroughly described for the non-LTE two-level atom case. We describe hereafter in details how such methods can be generalized when dealing with non-LTE unpolarised radiation transfer with multilevel atomic models, in monodimensional geometry.

The Concordia Station (Candidi 2003), on the Antarctica Plateau, will soon become one of the best observatories to perform infrared observations in the 2–20 μm atmospheric windows and beyond, thanks to its low sky background, low temperature and high atmospheric transparency. The possibility of passively cooling the telescope is a further advantage. We describe here the first permanent Antarctic infrared telescope, under development for the Dome C base. It is the International Robotic Antarctic Infrared Telescope (IRAIT). We briefly outline a few scientific motivations for it, then we review the technical characteristics and the status of its development. The infrared camera for IRAIT is described in another dedicated paper in this volume.

Models predict that magnetic fields play a crucial role in the physics of astrophysical accretion disks and their associated winds and jets. For example, the rotation of the disk twists around the rotation axis the initially vertical magnetic field, which responds by slowing down the plasma in the disk and by causing it to fall towards the central star. The magnetic energy flux produced in this process points away from the disk, pushing the surface plasma outwards, leading to a wind from the disk and sometimes a collimated jet. But these predictions have hitherto not been supported by observations. Here we report the direct detection of the magnetic field in the core of the protostellar accretion disk FU Orionis. The surface field reaches strengths of about 1 kG close to the centre of the disk, and it includes a significant azimuthal component, in good agreement with recent models. But we find that the field is very filamentary and slows down the disk plasma much more than models predict, which may explain why FU Ori fails to collimate its wind into a jet.