About linear momentum conservation

From: Kim Do Gyun <kdgcom_at_gmail.com>
Date: Fri, 23 Jan 2009 14:47:55 +0900


I'm using Gadget2 to simulate compound galaxy.
I use 5Mpc cube box and total 220000 particles to describe halo, bulge and
All particles have same mass.
In initial condition, the mean velocity of all particles is 10^-7(~0) in
km/s unit.

The problem is that this isolated galaxy moved to another place.
After about hubble time(~10^!0 year), this model galaxy moved about 70kpc(of
course, I mean the center of mass).

To build the Gadget2 executable, I used 'galaxy.Makefile' in example,
therefore I did't use
periodic boundary condition. I even used same softening length setting in
that file (halo-1.0, bulge&disk-0.4).

Is this result possible?
70kpc in 10^10 year means that the change of total mean velocity is about

I tested bigger box size, 20Mpc, and the movement was reduced to ~20kpc.
And 200Mpc cube didn't show better result.

I searched this mailing list and found that just 1 posting about linear
In that posting, Dr. Springel just said that Gadget proveds just good
approximation of true force and
some adjustment of gravity parameter will help.

Could anyone give me some detailed information about this problem?
From now on, I cannot believe that small force error of small fraction of
particles can make
this big error of linear momentum(because most force error arise near the
boundary of short and long
force range).

Thank you,
DG, Kim.

Kim, Do Gyun
Center for Space Astrophysics
Yonsei University, Seoul, Korea
TEL) 82-2-2123-4249
FAX) 82-2-362-5136
Received on 2009-01-23 06:47:57

This archive was generated by hypermail 2.3.0 : 2023-01-10 10:01:30 CET