NeoNarval observations of Chi Cyg - Stokes V

This repository contains the data, and scripts to reproduce the magnetic field detection of Chi Cyg, obtained from Neo-Narval spectropolarimetric observations in 2025.

Author : Alexis Lavail, astro@lavail.net

Reduced files and pre-processing

Neo-Narval reduced fits files are in the root of this repository, these are files named with the convention NEO_YYYYMMDD_hhmmss_st3.fits. The structure of these files is explained by the documentation available at https://tbl.omp.eu/instruments/neo-narval/, and also in the PDF file NeoNarval_FITS.pdf.

For each epoch (july, august, october), the individual spectra are first spliced, and then co-added to produce a single I, V, and N spectrum. This is done with the nn_addsplice_spectra.py script. The command to run it for the three epochs is shown in run.sh (./run.sh).

The nn_addsplice_spectra.py script processes Neo-Narval spectropolarimetric data from the reduced FITS files. It splices together overlapping spectral orders and averages multiple observations of the same target.

Its splice_spectrum() function reads spectropolarimetric data (wavelength, intensity, polarization, error, null) from FITS files and extracts metadata including BERV (Barycentric Earth Radial Velocity), Stokes parameter, and target name. It processes orders, identifies overlapping regions between adjacent spectral orders, and blends these overlaps using weighted averaging based on both spatial position within the overlap region and uncertainty weights (inverse variance weighting). The function removes NaN values and applies BERV correction to wavelengths.

The main script processes all input FITS files provided as command-line arguments, interpolates all spectra onto a common wavelength grid (from the first file), and averages intensity, error, Stokes, and null data across all files. It creates an output directory named {star}_{daterange}-{stokes}/ and saves three ASCII files (intensity, polarization, and null spectra) plus a FITS file with all processed data in a binary table. The script includes sanity checks for consistent Stokes parameters and target names across all input files.

Computing LSD profiles

We used Least squares deconvolution to produce mean line profiles combined from many (10000+) spectral lines in order to achieve significant SNR. The line list, extracted from VALD3 using extract stellar is in lsd-repository/t3500/t3500g+0.5.sel (./lsd-repository/t3500/t3500g+0.5.sel). This file is in the VALD3 short format. The python script create-linemask-from-vald-2-drs.py (lsd-repository/t3500/create-linemask-from-vald-2-drs.py) parses the line list, operates the selection of spectral lines, and output line lists in .fits, .pickle, and .ascii format. The line weights are not the same for Stokes I (w = d) and Stokes VN (w = d \lambda g_{\textrm{eff}}). The linelist for intensity is lsd-repository/t3500/t3500g+0.5.sel_atmo_I.lin, for polarisation it is lsd-repository/t3500/t3500g+0.5.sel_atmo_Zeeman.lin.

The LSD profiles computed with these linelists are found, for e.g the August epoch at:

• ChiCyg_2025-08-08to2025-08-16-V/t3500/ChiCyg_2025-08-08to2025-08-16-stokesI_all.lsd
• ChiCyg_2025-08-08to2025-08-16-V/t3500/ChiCyg_2025-08-08to2025-08-16-stokesV_all.lsd
• ChiCyg_2025-08-08to2025-08-16-V/t3500/ChiCyg_2025-08-08to2025-08-16-stokesN_all.lsd

To produce the Figure 1. presented in the paper, run

python3 plot_IV.py chicyg t3500