GALAXY N-body simulation code
© 2014 J. A. Sellwood, formerly Rutgers University, now Steward Observatory
This website lists the updates made in the various GALAXY
software releases
A detailed description of how to install and use the code is given
in the updated
on-line
manual
Known bugs in version 16.11
None - v16.11 fixes all known bugs in previous versions
Updates
- Aug 16, 2023. Release of v16.11. Shu DFs for disks now created
with an improved algorithm.
- The principal improvement is to the software to set up a DF
by Shu's method. This part of the code has been completely rewritten
- see the notes in section 15.7 on p82. The new method is both faster
and better than the old, but it still requires careful management as
described in "Advice to the user" (section 15.7.4).
- This release also makes it possible to specify the run
number on the command line for most of the executables in the package.
If the run number is omitted, the code still prompts for it. analys
can be invoked with multiple run numbers if it is desired to compare
the evolution of the first mentioned run with others.
- Other minor improvements are to fix little bugs and to
improve parts of the analysis software as detailed in
DOCS/revisions.v16.11
- Mar 3, 2023. Release of v16.10. This release features
improvements to the derivation of a Shu DF for disks.
- Program smooth should be run interactively only if
it is to create the DF for a disk by Shu's method using dfShuc,
non-interactive runs are OK for other DFs. It creates the .dft
file which is now used separately by dfShut.
- New functions QofR and taperR are supplied to
set default profiles that can be edited by the user, if desired.
- A number of minor improvements to fix little bugs and to
improve parts of the analysis software are also included.
- Sep 14, 2022. Release of v16.04. fixed 2 bugs that had been
introduced in v16.0.
- Eliminated partial solutions for the field on polar grids,
since accel dependent zones are not spatially separate. This was a
bug, introduced at v16, that would have caused large changes to the
total energy
- Another recently introduced bug that is now fixed caused
failure to restart simulations that used the hybrid code on
multiple processors
- Jun 1, 2022. Release of v16.0. Principal changes at version 16.0
- added acceleration-dependent criterion for time steps
- changed the .dat file to input the longest time step that
will used, which will be reduced by successive factors of two in
either spatial zones or as required by accelerations
- input analysis time interval and duration of the simulation
to be in units of time, not steps
- created a blkdat segment to initialize variables in
common
- made more use of allocatable arrays
- made ptcls a 2D array
- a number of minor bug fixes (see revisions.v16.0 in the
DOCS directory)
- updated most of the README files, which unfortunately
contained a lot of misinformation in version 15.
- May 25, 2020. Release of v15.4. Fixed a significant bug,
found by Vance Wheeler, that caused large force errors inside
the guard radius from the s3d when not all particles have the
same mass. The self-force on a particle that was subtracted
from s3d grid force was not scaled by the relative particle
mass.
- July 20, 2019. Release of v15.34. Fixes of a few very minor
bugs of little significance (see the changelog.txt file in the DOCS
directory for a full list), but there were somewhat extensive
improvements to the EXAMPLES directory in the form of updated README
files and standardized scripts.
- May 18, 2019. Release of v15.33. Fixed array subscripting
caused by different roundoff in gfortran from other compilers. This
will have very occasionally, and at one random step only, caused the
wrong accelerations to have been applied to particles on polar
grids.
- February 4, 2019. Release of v15.32
Very minor update. Bug in qsetup.f fixed, tiny changes to galaxy.f
and to dflook.f, and the gfortran options in the Makefile now trap
floating point errors
- January 3, 2019. Release of v15.31.
- The arrays islist and s3rad, which were in common blocks admin and
grids respectively are now allocatable, and therefore held in module
aarrays. The local iput buffer in psetup is now allocatable also
- The total energy (PE + KE) of active particles now includes
contributions from all active mass components - previously it was the
first two only - a book-keeping error only
- Other improvements and minor bug fixes in GALAXY
- Changes in setup code
- Some minor improvements to plotting software
- August 6, 2018. Release of v15.22.
- The main improvement in this minor update is to the Shu DF, which now
yields much better equilibria for heavy disks. In addition, it reports
either success or aborts with a helpful message on failure.
- The user now has the option to skip creating mpi versions of the main
programs. The single processor versions are still found in /progs
and the Makefile in that directory no longer attempts to create
mpi versions. The mpi versions have been moved to a separate directory
/progs_mpi, which has its own Makefile
- The program smooth now asks whether to create equal mass particles
before entering the number of particles to generate
- Subroutine msetup now reports more helpful diagnostics when there
are errors in the input data
- Tabs in the .dat file should no longer cause read errors
- Two minor changes to the mpi version of galaxy are to fix little
bugs, and a couple of analysis options have been improved
- February 6, 2018. Release of v15.20. The changes from
versions 15.0 and 15.1 are again minor. The generation of real*4
random numbers caused very occasional problems and has been
changed to real*8. Also anyone wishing to use guard radii
on the s3d grid will need this version, since v14.2 introduced
a subscripting error that has now been corrected.
- April 12, 2017. Release of v15.10 which elimitates the
restriction that prevented creating an equilibrium disk by Shu's
method in a compressed halo. The new capability can be tested by
running the script EXAMPLES/hybrid/run300.s
If you have already been working with v15.01, you can patch the code
you have with the minor differences that are described in the Appendix
(p84) of the revised on-line manual. First download
the tar
file, then cd to your GALAXY15 directory and move the downloaded
file there. Then issue the command "tar xzf update15.10.tgz" which
will replace the files that have been updated since v15.01. You will
then need to cd SRC/lib15 and execute "rebuild", then cd ../lib15_mpi
and execute "rebuild". Then cd ../progs and type "make" and then
"make clean" in the usual way.
- March 10, 2017. Release of v15, in which all calls to NAG
library routines have been eliminated. There is no update path
from older versions, and this version should be installed afresh from
the usual GALAXY
website.
- v14.52, replacing v14.50, released Dec 20, 2016. If you have
already been working with v14.50, or v14.511, you can patch the code
you have with the minor differences that are described in the Appendix
(p87) of the revised on-line manual. First download
the tar
file, then cd to your GALAXY directory and move the downloaded
file there. Then issue the command "tar xzf update14.52.tgz" which
will replace the 23 files that have been updated since v14.50. You
will then need to cd CODE/lib14 and execute "rebuild", then cd
../lib14_mpi and execute "rebuild". Then cd ../progs and type "make"
in the usual way.
- v14.511, replacing v14.50, released Nov 16, 2016. If you have
already been working with v14.50, you can patch the code with the very
minor differences that are described in the Appendix (p87) of the
revised on-line manual. First download
the tar
file, then cd to your GALAXY/CODE directory and move the
downloaded file there. Then issue the command "tar xzf
update14.511.tgz" which will replace the 12 files that have been
updated since v14.5. You will then need to cd lib14 and execute
"rebuild", then cd ../lib14_mpi and execute "rebuild". Then cd
../progs and type "make" in the usual way.
- New release of v14.5 on May 13, 2016. It is available at
the GALAXY
website. The additional capabilities and improvements of this version
are listed in the first Appendix (section 17) of the
on-line
manual.
- Jun 16, 2015. NAG changes The GALAXY package uses
Mark 21 of the NAG library and is consistent with Marks 18
through 22. The second appendix (section 18) of the
on-line
manual gives instructions on very minor changes that are needed
for linking with NAG Mark 17 and for Mark 23.
- Nov 26, 2014. The GALAXY package was updated to version
14.1, which contained a few significant improvements and also fixes
one major, and a few minor, bugs. These changes are described in
section 1.8 of
the on-line
manual.
An existing user wishing to update to version 14.1
can either start over by downloading the entire
revised GALAXY
package and re-install it, or
extract the files from this tar ball, which contains the source code of only
those programs, routines and include files that were revised at this
version. These files should overwrite those in the appropriate
directories of the original package, which can be achieved if tar is
executed in the directory "pckg". The lib14 and lib14_mpi libraries
will need to be rebuilt, and the programs recompiled by rerunning
'make'.
- Nov 21, 2014. The first released version of
subroutine loadup.f contained a serious bug that prevented
multiple populations of particles from being loaded properly into the
simulation code. The bug is fixed in this version, which should replace that in the directory
pckg/lib14