PYSHELLSPEC is an advanced astrophysical tool for modeling of binary systems with circumstellar matter (e.g. accretion disk, jet), computation of interferometric observables |V2|, arg T3, |T3|, comparison of light curves, spectro-interferometry with observations, and both global and local optimisation of system parameters. It is based on Shellspec, a long-characteristic LTE radiation transfer code by Budaj & Richards (2004).
If you use this code, please cite the original reference:
D. Mourard, M. Brož, J. Nemravová, P. Harmanec, J. Budaj, F. Baron, J.D. Monnier et al., Physical properties of β Lyr A and its opaque accretion disk, A&A 618, A112, 2018.
The author is Jana Nemravová, with numerous modifications by M. Brož as described in the Changelog. The latest version of Pyshellspec, together with the preprint of the respective paper, can be downloaded right here:
pyshellspec_20171207.tar.gz ... the actual version used in the paper pyterpol_20170728.tar.gz ... synthetic spectra generated by Pyterpol data_20171126.tar.gz ... observational data of β Lyr A fitting_nebula_LINES__103233.tar.gz ... example of a 'nebula' model fitting_spot_LIMCOF__102005.tar.gz ... example of a 'slab+spot' model Mourard_etal_2018_1807.04789.pdf ... Mourard etal. (2018) paper data_20190417.tar.gz ... new observational data pyshellspec_20190426_V47.tar.gz ... new Pyshellspec version pyterpol_20190430.tar.gz ... new Pyterpol data fitting_nebula_SPECTRA_V47.tar.gz ... new model (no fitting yet)
Best-fit images for 5 geometrical models:
Best-fit χ2 plots (cf. remaining systematics):
Minimum-maximum range searched for by optimisation (differential-evolution, simplex):
|slab power-law min.|
|slab power-law max.|
χ2 evolution during optimisation:
|A typical evolution of χ2 vs the number of iteration for a differential evolution (left), and simplex (right). Individual contributions to the χ2 are also indicated, where LC stands for the light curve, VIS the squared visibility, CLO the closure phase, and T3 the triple product amplitude. Pyshellspec enables to solve the inverse task by means of global and local optimisation of system parameters. For the differential evolution, there is no apparent convergence because the algorithm scans the allowed range of parameters. On the other hand, the simplex always shows a clear convergence to a local minimum, and one can check trade-off's between individual contributions (cf. LC and CLO which are sligthly decreased/increased).|
Miroslav Brož (firstname.lastname@example.org), Mar 5th 2018