bagemass Wiki
Bayesian age and mass estimates for transiting planet host stars
Status: Beta
Brought to you by:
pmaxted
Menu
▾
▴
Home
This is the README file for version 1.0 of bagemass.
VERSION HISTORY
1.0 - First export version.
31 Jul 2014, p.maxted@keele.ac.uk
This software is distributed under the terms of the GNU General Public
License GPLv3 (https://www.gnu.org/licenses/gpl.html).
Please cite the following publication if you publish results based on this
software.
P. F. L. Maxted, A. M. Serenelli and J. Southworth
2014, MNRAS, submitted.
Index:
1. CONTENTS
2. INSTALLATION
3. RUNNING THE PROGRAM
4. OUTPUT FILES
- CONTENTS
README: This file.
Makefile: Used by GNU make to compile and install the software.
data/: Directory containing spline fits to grids of stellar models stored as
FITS files.
src/: Source code
- INSTALLATION
This software should compile without much trouble on any modern linux system,
the only package you may need to install is cfitsio:
http://heasarc.gsfc.nasa.gov/fitsio/fitsio.html
Step-by-step instructions
Commands entered on the command line are preceeded by "$".
-
Download the compressed tar file bagemass.tar.gz from sourceforge
-
Unpack the archive
$ tar xzvf bagemass-1.0.tar.gz
This will create a directory bagemass-1.0
-
Move to the directory created in the previous step.
$ cd bagemass-1.0
-
Use any text editor to edit the file Makefile, e.g.,
$ pico Makefile
Change the values of the variables INSTALL, LCFITSIO, F77 and FFLAGS as
needed. -
Compile and install
$ make
You may see some warnings when src/herm2ev.f and src/herm3ev.f are
compiled - these can be ignored.The following directories will be created in the installation directory
if they do not already exist.bin/
share/
share/bagemass/ -
[Optional] Remove object files
$ make clean
-
If the "bin" sub-directory in the installation directory is not already in
the PATH system variable of your shell, add it now (and "rehash" if needed). -
Test run bagemass
$ bagemass
You should see the following output on your terminal
This is bagemass version 1.0
(p.maxted@keele.ac.uk)Available alpha_MLT and He-enhancement values are ...
1 1.78 0.00
2 1.50 0.00
3 1.78 0.02
Enter model grid number [1] >
If you hit return at all the prompts you should see lots of numbers printed to
screen, finishing with the following output. (The actual values printed will be
slightly different to this example.)
Mean age = 1.788 +/- 1.238
Mean mass = 1.075 +/- 0.045
Mean [Fe/H]_init = 0.019 +/- 0.084--
Please cite the following paper if you publish the
results of this analysis.
P. F. L. Maxted, A. M. Serenelli and J. Southworth
2014, MNRAS, submitted.
The following output files should also be created every time bagemass is run.
zams.dat
track_mlo.dat
track_mhi.dat
track.dat
isochrone_alo.dat
isochrone_ahi.dat
isochrone.dat
chain.dat
- RUNNING THE PROGRAM
First read the description of the method and models in Maxted et al., 2014.
The program is run using the following command
$ bagemass
The version number of the program is printed and the user is offered a choice
of model grids, e.g.,
Available alpha_MLT and He-enhancement values are ...
1 1.78 0.00
2 1.50 0.00
3 1.78 0.02
Enter model grid number [1] >
The default grid number is shown in square brackets. Hit return to accept the
default or enter the grid number desired and then hit return.
The full path to the model grid is displayed and the models are loaded (this
can take a few seconds).
The next set of prompts enable the user to enter the observed data for the
star to be studied. See Maxted et al. 2014 for a description of these 4
observed quantities.
Enter T_eff [K] [6000., 100., -100.] >
Enter log L/Lsun [0.0, 5.0, -5.0] >
Enter observed metalicity, [Fe/H]_s [0.0, 0.1, -0.1] >
Enter observed density, rho/rho_Sun [1.00, 0.01, -0.01] >
Observations are entered in the format
value +error -error
This enables the user to use asymmetrical error bars if needed. If the
measurements has symmetrical error bars, simply enter this value twice, once
as a positive value and once as a negative value. All three numbers must be
entered on the same line, followed by return.
The next three prompts enable the user to enter priors on the age, initial
metallicity and mass of the star.
For the age and metallicity, the default values are a flat distribution over
the valid model range. Priors values are entered in the format
loerr lolim hilim hierr
For chain values between lolim and hilim, the prior has a value of 1. If
hierr is >0 and the chain value is >hilim, then the prior has a value
exp(-0.5*(value-hilim)^2/hierr^2), and similarly for lolim and loerr. For
example, you can add a prior constraint that the age of star must be less than
the age of the Galactic disc (10 +/- 1 Gyr) using the following input.
$ Enter prior on age [Gyr] [0.0, 0.0, 17.5, 0.0] > 0.0 0.0 10 1
If hilim=0 then all trial chain points greater than hilim are rejected, and
similarly for lolim.
The prior on the mass is given by
log(Likelihood) = alpha*Mass
The default is to use a flat prior prior for mass, i.e.,alpha=0. For a
Salpeter distribution use alpha=-2.35.
The following prompts control the behaviour of Markov Chain calculation.
First there are 2 scaling factors that can be adjusted.
Enter initial step size scale [0.500] >
Enter MCMC step size scaling factor [1.00] >
These can generally be left at their default values, but you can experiment
with these values if you are having trouble getting your Markov chain to
converge. The first value scales the step size calculated by perturbing each
of the parameters in-turn prior to the first Markov chain. The second value
sets the step size in the two Markov chains as a factor of the standard
deviation of each jump parameter.
The last two prompts set the lengths of the two Markov chains.
Number of steps for MCMC burn-in [1000] >
Number of MCMC steps [10000] >
The default values are ok for test runs, but for proper analysis you will
probably want to use values of 50-100,000 for both chains.
The calculation starts as soon as the last prompt is completed.
The output to screen summarizes the progress of the calculation. This is
self-explanatory once you have read and understood the description of the
method in Maxted et al. (2014) and the description of the various output
variables in the following section.
One value of interest is the mean acceptance rate for the 2nd MCMC chain.
This will generally have a value of 0.5+-0.1 for a well-mixed chain, perhaps
slightly lower for complex cases. Values of the acceptance rate lower than
about 0.2 may suggest that the MCMC chain has not converged.
- OUTPUT FILES
The following output files should also be created every time bagemass is run.
chain.dat - Second Markov chain.
zams.dat - zero-age main sequence at best-fit [Fe/H]
track.dat - evolution file for best-fit mass and [Fe/H]
track_mlo.dat - evolution track for best-fit mass - error, and
best-fit [Fe/H]
track_mhi.dat - evolution track for best-fit mass + error, and
best-fit [Fe/H]
isochrone.dat - isochrone for best-fit age and [Fe/H]
isochrone_alo.dat - isochrone for best-fit age - error, and
best-fit [Fe/H]
isochrone_ahi.dat - isochrone for best-fit age + error, and
best-fit [Fe/H]
These are ascii files with comments in the headers preceeded with '#' that
describe the content of the file and give the variable names for each column.
The variables in chain.dat are as follows.
Step - Markov chain step number
Age - Age in Gyr
Mass - Stellar mass in solar units
[Fe/H]_i - Initial metallicity
DeltaR - Undocumented/disabled feature (should always by 0.0)
Teff - Effective temperature in Kelvin
LogL - Logarithm of the luminosity in solar units
[Fe/H]_s - Surface metallicity
rho - Stellar density in solar units.
Xc - Central hydrogen abundance per unit mass
loglike - Logarithm of the likelihood
Chisq - chi-squared.
The content of the files zams.dat, etc. is similar, but it addition the
following variables are provided.
Radius - Radius in solar units
logg - Logarithm of the surface gravity in units of cm.s^-2
Some variables appear twice in these files under column headings Radius_0,
Teff_0, etc.
Discussion
Please read Maxted et al. (2014) "Bayesian mass and age estimates for transiting planet host stars" (MNRAS submitted) for a complete description of the algorithm and models used in this software.