Cosmology Simulations' softwares

[MathematicalPhysicist]

I have always read about simulations being done by computer softwares concerning cosmology and astrophysics, do you know if any of the simulations softwares are offered (hopefully, free ) on the web?

thanks in advance.

p.s
do these softwares work on assembly environment (machine language)?

 

[neurocomp2003]

take a look at computational astrophysics professors webpages, this is a tip i got from one.
"There's also quite an interesting web
project called 'The Art of Computational Science' by Piet Hut & Jun
Makino that has some insight into what programs like STARLAB do"

some are free.
McMaster has a computational group. But if your american you may want to look at the an american school first.

 

[MathematicalPhysicist]

thanks!
(-: (-;

 

[Allday]

I am analyzing some time slices from a dark matter simulation done by a group called the Virgo Consortium. I don't know if the code for that is on the web-site but I'm sure at least one simulation code (GADGET) is available through this link

http://www.mpa-garching.mpg.de/NumCos/

unfortunately you need a massivley parallel cluster of computers to run it.

 

[ohwilleke]

Yes, my sense is that the big limitation of cosmology simulations is not so much that the software is profound. It takes some cleverness to turn mathematical equations into computable algorhythms, but not all that much. But, that the amount of processing power necessary to do a good simulation is immense.

 

[setAI]

you need a few teraflops of processing power to run a decent simulation- so your going to have to wait until the Playstation 3 comes out!

 

[ohwilleke]

You made me look: http://www.answers.com/topic/playstation-3

 

[Allday]

You do need a lot of power to run these cutting edge simulations, but a big thing (at least in dark matter simulations) is the problem of resolution. A smaller simulation with a lower mass per particle may be kind of interesting. I don't know all the ins and outs but I took a class on MPI programming. Its a group of subroutines that can be used with C or Fortran to write parallel code. If you could scrape together 4 gigahertz machines and get them connected and get a PBS (I think that's the acronym) scheduler running on them you could run a decent size simulation in a couple days. I've seen websites about building your own cluster, but I never had the spare computers or time to experiment with.