This is: http://sail.zpf.fer.hr/fd3, last update July 2014.
A tool for spectral disentangling of double-lined spectroscopic binary stars
The spectral disentangling technique can be applied on a time series of observed spectra of a spectroscopic double-lined binary star (SB2) to determine the parameters of orbit and reconstruct the spectra of component stars, without the use of template spectra.
fd3 is a tool for disentangling of the spectra of SB2 stars, capable also of resolving the possible third companion. It uses a combination of ideas borrowed from Simon and Sturm (1994) and Hadrava (1995), as well as from Hensberge, Ilijić & Torres (2008). fd3 performs the separation of spectra in the Fourier space which is faster, but in several respects less versatile than the wavelength-space separation. (Wavelength-space separation is implemented in the twin code CRES.)
fd3 is written in C and is designed as a command-line utility for a Unix-like operating system. It is available as the executable file for Linux (x86), or as the complete archive file containing the source files and the files needed to reproduce the examples documented on this web page (if compiling fd3 from the source, do not forget to install and link the GNU Scientific Library).
fd3 uses a three component hierarchical orbital model. Components denoted A and B are assumed to form a close pair in an eccentric Keplerian orbit. The A--B pair is assumed to be in a wide eccentric Keplerian orbit AB--C with the third star denoted C. The light-time effect in the wide orbit is taken care of by fd3.
B
/
/
C -----------------------------+---------------------+ cm(AB)
cm(ABC) /
A
Please note that handling of orbital parameters related to apsidal motion of the A--B orbit is not fully implemented yet.
Any subset of components A, B and C can be used. This includes the possibility of applying the method on a single, double or triple-lined systems. One can use tight orbit with (not yet functional) apsidal motion, or wide orbit with light-time effect. Of course, the component C can also be used to model any static additive features in the composite spectra.
One data file containing the complete time sequence of continuum-normalised composite spectra uniformly (equidistantly) sampled in the logarythm of the wavelength, covering the whole sprctral region that was observed (reduced), can be used for working with fd3 in different spectral sub-regions. The first column this master file must contain the log-wavelength values, and the subsequent columns must contain the normalised flux values. The fd3 output files that contain model component spectra and residuals follow the same format.
All other input (times and other parameters of observations, orbital parameters and other instructions) is supplied either by answering the prompts that the program prints on the screen, or more conveniently by feeding in a "control file". All times should be given in days, all velocities in kilometers in second (internally, fd3 uses c = 299800 km/s), and all angles (phases) in degrees.
The file containing log-wavelengths and amplitudes of the artificial composite spectra is art_double.master.obs. Command line is:
$ ./fd3 < art_double.in > art_double.out &
The control file art_double.in has an uncomprehendible structure unless read in parallel with the output file art_double.out. Output files are:
Input spectra: art_single.master.obs
$ ./fd3 < art_single.in > art_single.out &
Output files: art_single.obs, art_single.mod, art_single.res, art_single.rvs, art_single.log.
Input spectra: art_triple.master.obs
$ ./fd3 < art_triple.in > art_triple.out &
Output files: art_triple.obs, art_triple.mod, art_triple.res, art_triple.rvs, art_triple.log.
Here I reproduce the run on real data that is documented in my Master thesis (2003), where v.1 of the code was used. The observed spectra are in V453_Cyg.master.obs. Command line is:
$ ./fd3 < V453_Cyg.in > V453_Cyg.out &
Output files: V453_Cyg.obs, V453_Cyg.mod, V453_Cyg.res, V453_Cyg.rvs, V453_Cyg.log.
Versions 1 and 2 of the code were called FDBinary. They were in some respects more sophisticated than fd3, but could handle only two components. They also had a different, more difficult to use, user interface. FDBinary V.1 is described in my Master thesis (2003). I recommend using the fd3 instead of these codes.
Techniques in general (biased selection):
FDBinary applications (possibly incomplete list):
Saša Ilijić,
sasa.ilijic@fer.hr
Department of Applied Physics,
FER,
University of Zagreb
Unska 3, HR-10000 Zgreb, Croatia