 |
| Chris Johnson 2003-07-04 05:46:45 | I like the new galaxy simulation screenshots. Any chance of some info on how the algorithm works? From the graniness it looks like particle-based rather than grid-based simulation, but are there any other considerations (relativistic motion, realistic damping, heat flow etc.) which you've used? As people might have seen at the time, a fairly large scale (750 hours on an 8Gflop Cray) simulation of this sort of thing was done a short while ago: http://zeus.ncsa.uiuc.edu:8080/Summers/galmerge.html http://hubblesite.org/newscenter/archive/2001/22/video/b |
| Stephen Brooks 2003-07-15 07:01:58 | Well that reminded me to (eventually) upload the program properly so that people can download it. I suspect that adding gas clouds and the like (as well as the stars) would require temperature and pressure to come into it. That requires solving a fully-general fluid dynamics equation with some strange pV=nRT stuff in each cell, which is the sort of thing I would normally only do if I was being payed for it... Today's weather in %region is Sunny/(null), max. temperature #NAN°C |
| Chris Johnson 2003-07-15 16:36:00 | Regarding the newly posted galaxy formation program: From the fact you're using the FFT I assume that this is a Newtonian (or at least non-GR) simulation, and the FFT is used for fast convolution of the star densities in each grid square with a single gravity potential to create the total potential field and thus the forces on each star. Is this right, or am I being too simplistic...? |
| Stephen Brooks 2003-07-16 03:23:25 | That is how it works. For some reason even the big simulations talk about the number of "FFT cells" so it looks like even they do something similar. The one additional factor the cosmological models will have in them is the expansion rate of the universe, which effectively just scales the forces and velocities a bit. I'm trying to figure out which of SR and GR is least irrelevant, but stars don't typically go at more than 1% of the speed of light relative to their local neighbourhood (probably much less), making SR effects tiny. GR is more worrying, as a few people have said there are galaxies or clusters around with something like 10% of the density needed to become a black hole. I've not heard of anyone simulating this with GR included though. From what we see in the sky, it doesn't look as if GR effects are significant on the large scale, though there have got to be some cluster configurations around somewhere where they are fairly big. Today's weather in %region is Sunny/(null), max. temperature #NAN°C |
