aba4633 Files

sphNG calculation of the GW Ori system
--------------------------------------

This directory contains the initial conditions, input files, and Fortran
code required to perform the smoothed particle hydrodynamics (SPH)
calculation, the results of which were presented in the paper.

It also contains output files from the original SPH calculation and 
the SPLASH configuration files that that were used to generate 
Figure 3 of the paper.

The Fortran SPH code, called sphNG, is provided under the 

To run the SPH calculation and to produce images presented in Fig. 3, 
the following sequence should be followed:


1)  Compiling the SPH code.  The tarball should be expanded to provide
the Fortran source code.  

Inside the sphNG source code directory is a "LICENSE.txt" file that
contains the license information regarding the use of the sphNG code.

The file "Makefile" contains compilation flags for various machines.
The calculation used in the paper was performed on the Isca 
linux supercomputer at the University of Exeter, UK, on a single 
28-core Intel compute node.  The Intel Fortran compiler version:

ifort (IFORT) 16.0.3 20160415

was used to compile the code.  The Makefile contains the compilation
flags that were used to compile the Fortran code on Isca.  These are 
selected when a linux/unix environment variable is set to:

export SYSTEM=isca

NOTE that the code was compiled using BIG endian for memory addressing,
and the Fortran binary files are also written in BIG endian format.

The SPH code is compiled as:

make mpi=no openmp=yes gradhrk

This will produce an executable called

sph_tree_rk_gradh

that has OpenMP parallelisation.


2)  Running the SPH calculation.  Two input files are required:

GWO0000 : this is a Fortran binary file (big endian format)
ifile.gas : this is an ASCII text file that contains various parameters

A copy of the ifile.gas file should be made, e.g. like:

cp ifile.gas ifile

To run the code:

./sph_tree_rk_gradh evolution ifile

or equivalent.  The code will use OpenMP parallelisation if this is
available (the number of threads to use, etc must be set).  

The code should then run, producing a sequence of binary dump files.
Dump files GWO0100, GWO0200, GWO0300, GWO0400, GWO0500, and GWO0595
from the original calculation have been provided (gzipped).

The code will also produce an ASCII text output file called "gwori001".
The original "gwori001" file has also been provided.  The original
calculation was run with OpenMP parallisation on 28 threads and
took about 30 hours to run.

The 595-th dump file from the original calculation is the one that was 
used to produce Fig.3 from the paper (i.e. GWO0595).  Note, however, that
re-running the calculation will not produce an identical evolution
to that from the original calculation because the evolution is chaotic
and the sphNG code does not enforce thread ordering.  This means that 
the results from OpenMP parallel calculations differ at the level of 
round-off error.  For non-chaotic systems this is usually unimportant, 
but for chaotic systems this can lead to qualitatively different 
behaviour with time.


3)  To produce the images presented in Fig. 3, the SPLASH visualisation
software must be used to render the images.  SPLASH is open source,
publicly available software available from:

http://users.monash.edu.au/~dprice/splash/

Version v2.7.0 of SPLASH was used.

SPLASH uses the configuration files provided:
splash.defaults
splash.limits
splash.units

Once SPLASH is installed, to produce the images in Fig. 3, do the following:

ssplash GWO0595

Remember that big endian must be used.  This may require setting appropriate
environment variables, such as:

export ENDIAN=BIG
export GFORTRAN_CONVERT_UNIT=big_endian
export F_UFMTENDIAN=big

The images are produce by choosing option 19, and outputing the result
to /png.  This should produce a PNG image file:

splash.png

The original image file from SPLASH that was used to create Fig. 3 in the
paper has been provided.

---------------------------------------------------------------------------