spinOS calculates binary orbital elements. Given a set of radial velocity measurements of a spectroscopic binary and/or relative position measurement of an astrometric binary, spinOS fits an orbital model by minimizing a chi squared metric. You need to supply an initial guess for the parameters that define the binary orbit:
- p: the period of the binary
- e: the eccentricity of the orbit
- i: the inclination of the orbit (with respect to the plane of the sky)
- omega: the argument of periastron of the secondary, with respect to its ascending node
- Omega: the longitude of the ascending node of the secondary, measured East from North
- t0: the time of periastron passage
- d: the distance to the system
- k1: the semiamplitude of the RV curve of the primary
- k2: the semiamplitude of the RV curve of the secondary
- gamma1: the peculiar velocity of the primary
- gamma2: the peculiar velocity of the secondary
- mt: the total mass of the system
This package provides a GUI to easily visualize your data. The GUI allows for easy plotting of data and models, as well as minimization of the model to your supplied data. The program then gives a best fit value for the parameters itemized above.
To use the GUI, simply run:
python spinOS.py [dir]
In the GUI data tab, put your working directory, where all your data files are relative to where you launched spinOS from. You can select files to be loaded in, or add Datasets and data entries manually in the respective tabs of the Data Manager. When loading in data, see formatting below.
In the System/parameters tab, you can play with the systemic parameters, and see the changes of the orbit on the plots by pressing 'refresh'. With the checkbuttons, indicate which parameters should be minimized. Below, some inferred parameters of the model are presented. Use the load guesses button to load all guesses from your guessfile indicated in the data tab. The save buttons save either the guesses to files you specify in the output names tab (warning: may overwrite!)
In the minimize tab, you can apply a custum weighting of the astrometric data to the chi-squared value (typically, you would want to increase this if you trust astrometry better than spectroscopy, and have only few astrometric measurements). You can minimize the model to the selected data with the minimize button, selecting a method first. Levenberg-Marquardt does local non-linear least squares minimization, while basinhopping tries to find a global minimum by iterating local Nelder-Mead simplex minimizations. Note that basinhopping can take considerably longer with lots of free parameters. Alternatively, you can select MCMC which first does a local LM minimization followed by an mcmc error estimation. When MCMC sampling, the first burn (default = 100) samples are discarded, and then only 1 every thin (default = 1) samples are retained in the final results. The philosophy behind this is that the underlying sampler does not draw independent samples from the posterior distribution, it first needs to 'settle' to the maximum likelihood region (hence the burning), and then a random walk will only yield independent results twice every time the characteristic autocorrelation time has passed (hence the thinning). These parameters are difficult to estimate beforehand. When minimizing, error are estimated as the diagonal elements of the correlation matrix, or as half of the difference between the 15.87 and 84.13 percentiles found in an Markov Chain Monte Carlo sampling if you selected MCMC.
In the plot controls tab, various checkbuttons are provided to plot certain elements on the plot windows on the right. The phase slider allows for overplotting a dot at the phase indicated (for illustrative purposes, eg, for visually connecting the apparent orbit with the RV plot).
This application expects the data to be in the following format: All data files should be plain text files, formatted as: for RV data:
JD(days) RV(km/s) error_on_RV(km/s)
e.g.:
45000 25.1 2.1
45860 -4.2 1.1
etc...
for AS data: either (set button E/N):
JD(days) E_separation(mas) N_separation(mas) semimajor_ax_errorellipse(mas) semiminor_ax_errorellipse(mas) angle_E_of_N_of_major_ax(deg)
e.g.:
48000 -2.5 2.4 0.1 0.8 60
48050 2.1 8.4 0.4 0.5 90
etc...
or (set button Sep/PA):
JD(days) separation(mas) PA(deg) semimajor_ax_errorellipse(mas) semiminor_ax_errorellipse(mas) angle_E_of_N_of_major_ax(deg)
e.g.:
48000 3.5 316 0.1 0.8 60
48050 8.7 76 0.4 0.5 90
etc...
for the guess file, format should be e.g.:
e 0.648 True
i 86.53 True
omega 211.0 True
Omega 67.3 True
t0 56547.1 True
k1 31.0 False
k2 52.0 True
p 3252.0 True
gamma1 15.8 False
gamma2 5.6 False
mt 30.0 True
All eleven parameters should be guessed if you load in guesses. (for their meaning see above)
python 3.9.10
tk 8.6.12
numpy 1.22.2
scipy 1.8.0
lmfit 1.0.3
matplotlib 3.5.1
emcee 3.1.1 (if MCMC error calculation is performed)
corner 2.2.1 (if MCMC corner diagram is plotted, you need pandas for this too)
A LaTeX distribution that allows matplotlib.rc(usetex=True)
the supplied yaml file can be used to configure your python evironment
Matthias Fabry
Instituut voor Sterrekunde, KU Leuven, Belgium
17 January 2022
Copyright 2020, 2021, 2022 Matthias Fabry. This software is released under the GNU GPL-3.0-or-later License.
We thank the authors of lmfit for the development of their package.