script to submit several MT sequences to a computer cluster
Python Fortran Shell Makefile
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
MTgrids/a0.0_b0.25_wdMedd
executables
other_data
scripts
systems
README

README

Prepared by Mads Sørensen

The following document introduces the submission script relevant to investigate XRB systems using MESA.
The guide is intended to help with generating and submitting a job list on a computer cluster.
When a new simulation is needed, one simply redo this guide with the appropriate changes to a new set of simulations.

The overall doing of the scripts:
For a set of initial parameters of orbital period P, donor mass M2, and accretor Mbh, 
at the onset of Roche Lobe Overflow (RLO) the script evaluate if the given parameters lead 
to a potential solution in the orbital evolution of the system considered.

The evaluations considers the radius of M2 at RLO onset. It then evaluates mass transfer 
given some efficiency alpha and beta which are the fraction of mass lost from the donor 
and from the accretor respectively. The mass lost at the donor carries with it the specific
angular momentum of the donor. likewise for mass lost from the accretor this carries with it the specific
angular momentum of the accretor. 
The mass transfer is done under the fiducial test and consists of a simple Lagrangian point mass binary model.

******************************************************************************************
SHORT VERSION:
1) Adjust the relevant inlist files in the MESA project folder.
2) Adjust all paths in SetupScriptsFile.py
3) Generate job list:
$ python GenerateJobList_ncpu.py ncpus SetupScriptsPathsFile alpha beta system > ~/.../main_folder/other_data/a"alpha"_b"beta"_"other"_JobList_1.txt
4) Adjust SubmitJobList.py to the relevant queue management system
5) Submit job list
$ python SubmitJobList.py "number of jobs to submit" "path to SetupScriptPathsFile"

******************************************************************************************
LOOOOONG VERSION

Content:
1) Folder structure
2) How to submit a job

------------------------------------------------------------------------------------------
1] Folder structure
Keep the folder structure locally and on the cluster for easier management. 
Job lists are most convenient created locally, especially for large lists.
Be modest and create initially a small list.

1) Underneath a main folder you need the following folders:
I.E: ~/.../main_folder/
	a) /executables
	b) /MTgrids
	c) /other_data
	d) /scripts
	e) /SINGLEgrids
	f) /systems

a) executables:
Contains the MESA project folders required for each simulation. The examples shown reveal some MESA projects.
In other words, the executables contains the initial conditions of each simulation.

b) MTgrids:
Contains the raw data produced from your simulations on which you will perform your data analysis.

c) other_data:
Primarily used to save the different job lists generated each time a simulation is submitted.

d) scripts:
Contains all relevant functions and programs to setup a job list and submit this list to a cluster.

e) SINGLEgrids:
Contains data on single stars of specific initial conditions, primarily metallicity and mass. 
Before a job list  can be created you need to have the single grids ready.
For systematics make the subfolders of SINGLEgrids to be: /SINGLEgrids/Z"value"/"mass_of_star"/
e.g. 
SINGLEgrids/Z0.02/1.0/
would contain a single star of mass 1 solar mass with a solar neighbourhood metallicity Z=0.02. 
Stack stars of equal metallicity within the same folder.

f) systems:
This folder contains a simple .dat file with relevant observational parameters of the system. 
The file is divided into three columns which contain:
I) parameter name
II) parameter value
III) 1-sigma error on parameter value


------------------------------------------------------------------------------------------
2] How to submit a job--------------------------------------------------------------------
This part explains the steps needed to generate and submit a job list.

2.1 First things first...
2.2 preparing folder MTgrids
2.3 Preparing folder executables
2.4 Generating a job list
2.5 Queue Management script
2.6 The SetupScriptsPathsFile.py
2.7 Submitting a job list

2.1] First things first...----------------------------------------------------------------
Before continuing make sure the following things are handled:

a) A grid of single stars has been simulated and copied to a location as described 
under section 1]e) above.

b) A data file containing observational values which you have available.
For the generation of a job list in the current version the following parameters are needed:
P: Orbital period in days.
M2: Mass of donor star in solar masses
Mbh: Mass of accreting compact object, e.g. black hole, in solar masses
Aspin: Kerr spin of black hole in units of c.

It is possible to remove one or more of these dependencies when generating a job list, 
but it also removes the need for this feature.


2.2] Preparing folder MTgrids-------------------------------------------------------------
-- Relevant for cluster --
Under MTgrids create a folder named like a"value_of_alpha"_b"value_of_beta", e.g.
a0.25_b0.50/ This folder contains mass transfer sequences all with alpha = 0.25 and beta = 0.50.
Under this folder the file "SetupScriptPaths.py" shall be found.
An example of this file is found under the folder  scripts/
It is also good practice to add to the folder name other extensions to indicate if there 
are other special things included into the simulations, e.g. Eddington Luminosity or no overshooting etc.

2.3] Preparing folder executables---------------------------------------------------------
-- Relevant for cluster --
A copy of a MESA project shall be put in here. For the job list we are interested in the alpha and beta mass transfer efficiencies.
Within the relevant MESA inlist file change alpha and beta to match the folder made under MTgrids/. 
Remember to also adjust for any other relevant inlist parameters needed.

2.4] Generating a job list----------------------------------------------------------------
Go to the folder scripts/ and open GenerateJobList_ncpu.py.
GenerateJobList_ncpu.py is written in python 2.7 using the anaconda distribution. 
Generate the job list locally and not on the cluster. 
For a long list it can be relative time consuming wrt the work environment of the cluster.

In lines 43 to 45, one has the initial values at RLOF onset of orbital period, donor mass and black hole mass.

These can be changed to fit the desired needs. Make sure that M2 has a match within the SINGLEgrid folder.

Lines 35-39 is the command line input when executing the program.
	ncpus 			= int(sys.argv[1])   	: Is the number of cpus/cores to use.
	SetupScriptPathsFile 	= sys.argv[2] 		: Path to file SetupScriptPathsFile, see MTgrids.
	alpha 			= float(sys.argv[3]) 	: Fraction of mass lost from donor.
	beta 			= float(sys.argv[4])	: Fraction of mass lost from accretor
	system 			= sys.argv[5] 		: name of system.dat file, i.e. "LMC_X-3.dat"

To execute the generate job list file the command is:
***
$ python GenerateJobList_ncpu.py ncpus SetupScriptsPathsFile alpha beta system
***
The program is set to simply write out to the screen.
To print to a file add to the execution commmand:
***
$ python GenerateJobList_ncpu.py ncpus SetupScriptsPathsFile alpha beta system > ~/.../main_folder/other_data/a0.0_b0.25_wdMedd_JobList_1.txt
***

Feel free to change the name of the text file but output it under other_data/. but remember to adjust the SetupSciptsPathsFile.py.

2.5] Queue management script—-------------------------------------------------------------
-- Relevant for cluster -- 
A script file adopted to the queue management system on the cluster shall be created and stored under other_data/
See the example baobab_shared.slurm which is for the SLURM system. Northwestern does not use SLURM. So this file needs to be edited significantly.

2.6] The SetupScriptsPathsFile.py---------------------------------------------------------
Adjust all relevant lines in the file so the paths point to the correct files.
It shall be done whenever a new batch of simulations is prepared for submission.
Make sure all file names etc. are changed when necessary.

2.7] Submitting a job list----------------------------------------------------------------
-- Relevant for cluster --
Unfortunately, when submitting a job list, the lists lines are deleted once they are submitted. Make sure to keep a copy of the job list locally.

Under scripts/ open SubmitJoblist.py.
Adjust relevant parameters to fit the choosen computer cluster and its queue management system. The present version is set up with SLURM.

To execute SubmitJobList.py do the following:
Logon to the cluster where the same folder structure as described in this document is available.
Copy the generated job list to the other_data/ folder.
Execute the command below.
***
$ python SubmitJobList.py "number of jobs to submit" "path to SetupScriptPathsFile"
***