This document describes in an essential manner the use of GRASIL, and how to set the parameters affecting the model geometry, SF history, dust content. For details on our model, see this paper (Silva, L., Granato, G.L., Bressan, A., Danese, L., 1998, ApJ, 509, 103).
We distribute 2 versions
of GRASIL: The "light" and the full versions. The light version is almost the same we use,
but for the fact that it is more restrictive in the allowed
ranges of parameters. This is only to help the general user to
avoid unreasonable models, while any reasonable model should be
run by GRASIL. However we think that a quite a lot of care is
still necessary to use it. Also a few features which
are not useful for the general user are disabled. But for a more general and
flexible use we suggest to run the full version.
...is to generate star formation and chemical enrichment histories of the model by running CHE_EVO.
The program is invoked by the following command:
grasil {mod_nam} [par1=value1] [par2=value2]...
where the parameter mod_nam is a string (maximum 10
characters) used by the program to assign file names for
input and output, as described below. The grasil executable and
the libraries it uses may well reside in a different directory
than mod_nam.* files. In this case simply give the full path
before grasil in the above command. The command line parameter
within {} is compulsory, while those
within [] are optional. The strings
par1, par2 etc stand for the names of the parameters which are
set in the file mod_nam.par (see below). Whenever new
values are given in the command line (value1, value2 etc.), they
supersede those written in the .par file.
mod_nam.par: file with all the input parameters regulating the star formation history, galactic geometry, dust composition, directions of required spectra, library of SSPs etc. Note that the values of the parameters given in the mod_nam.par file may be overridden by specifying different values in the command line, with the syntax par=value, where par is the name of the parameter you want to alter, and value its new desired value. You can modify as many parameters as you want in this way.
mod_nam.sf: file with the star formation and chemical enrichment histories of the model. It is generated by our own program che_evo which is also distributed here.
Only for igeo=-1: mod_name.bst: in the case of composite (bulge + disk) geometry (set with igeo=-1 in the .par file), in addition to the .sf and .par files as input for grasil, grasil looks also for a .bst file that contains a tabulation of the bulge over total star formation rate of the model galaxy as a function of its age.
mod_nam.out: file in which useful checking quantities are written, such as the optical thickness, the luminosity with and without dust effects included (which 'ideally' should be equal…) and so on;
mod_nam.number and mod_name.ssp.number: SED in the direction specified by the extension number, which is an integer giving the angle in degrees between the direction and the polar axis; the output directions are specified by the user in the file mod_nam.par; The files mod_nam.number give the spectrum sampled at about 150 wavelengths from the UV to the radio, which are the internal working wavelengths. The files mod_name.ssp.number are much larger because they give the spectrum sampled at the same resolution of the SSP libraries (more than 1000 wavelenghts). Note that the latter files are not the trivial result of the interpolation of the former. By converse they are obtained by applying to the stellar SEDs of the various components (i.e. stars inside MC, stars in the disk, stars in the bulge), and computed at the resolution of the SSPs, the interpolated attenuation laws estimated by the model for the same component. In this way the detailed features of the stellar integrated specta are not lost.
mod_nam.spe and mod_name.ssp.spe: angle average of all the SEDs above; not produced if SED in only one direction is required, a good idea for spherical symmetric models...
The file consists of several sections, which can appear in any order and are started by a line consisting only in one of the following keywords.
Some of these sections (MAIN, MIXDIF and MIXMOL) are rather free form, in the sense that their parameters can appear in any order, each one on a different line of the section. The line contains the name of the parameter in the first ten columns, its value from column 11 to column 20, and the rest of the line is left for comments. In addition, each line starting with %, # or a blank space is ignored, and can be used for longer comments. The other sections are instead fixed form and their structure (order of lines, order of parameters in line, etc..) can't be changed. For an almost self explanatory example of .par file, have a look at this examplel.par.
The CPU time required by GRASIL depends a lot on the model. On fast PCs (say PII 300 or more) and Alphas typical CPU times may range from a few seconds to half an hour. To fully understand the following indicative sentences about this you should have seen the sample .par file examplel.par and possibly read this paper (Silva, L., Granato, G.L., Bressan, A., Danese, L., 1998, ApJ, 509, 103).
If you are not interested on dust effects but you want standard spectral synthesis, then you will set the parameter mmolfraz to a negative number, either from the command line or in the .par file, and get the answer in a few seconds. If in your model the cirrus is negligible, then you could proably simply set mmolfraz to 1 in which case the result requires 10-30 seconds. The most time consuming part is the computation of cirrus emission. Usually the effects of small thermally fluctuating grains is negligible in the global SED because their hot emission is by far dominated by MCs emission. Since the computations of the T distribution of these grains takes much longer than the computation of an equilibrium temperature, it is in general a good idea to run GRASIL with parameter flutflag set to 0. Models with spherical simmetry (igeo=1, which means King profiles) are faster than those without (igeo=2 or -1, in both cases the model has a disk). The former may take a few minutes, the latter of the order of 10 minutes, even much more with flutflag set to 1.0. For models with disk both the speed and the accuracy of the result are decresing functions of the flattening, that is of the ratio between the vertical and the radial scalelengths (safe ratios are between 0.2 and 0.1 or even somewhat less).