[gmx-users] MSD and merging trajectory files

[David van der Spoel]

On Wed, 16 Jan 2002, Lars Poulsen wrote:

>Hi!
>Since I'm not an expert at C code I was wondering, if anyone on the list 
>could enlighten me about the MSD routine in GROMACS: g_msd.
>I looks to me like g_msd is calculating <|r(t)-r(0)|>^2 averaged over 
>the molecules/atoms in the given group - which is all fine. However, it 
>seems to operate with only one time origin (t=0), which is not so fine 
>when you want to improve statistics. (from the output: the 'noise level' 
>is constant over time and it has sharp jumps)  Is this observation 
>correct?
>Is there a flag for g_msd I have overlooked? Or should I use g_analyse 
>instead? Or somethingelse?

This is a common misunderstanding. Using multiple origins (by splitting up 
the trajectory) does not give you any improved statistics. You just add up 
the numbers in different order. The best way to extract diffusion 
constants is to take the slope of the MSD, after disregarding the first 
few ps. You can however split the trajectory in bits, and calculate 
diffusion from that. This may enable you to make an error estimate if the 
pieces of trajectory are sufficiently independent. It will however not 
change the average D.

 >Because of some instabilities in my linux system, I had to restart my 
>md-simulation several times using tpbconv. Initially, I thought mdrun 
>would then continue writing in the same trajectory files as before, but 
>this appears not to be the case. So now I have several traj. files, that 
>I would like to merge into one. Is there an utility for that in GROMACS?

trjcat

Groeten, David.
________________________________________________________________________
Dr. David van der Spoel, 	Biomedical center, Dept. of Biochemistry
Husargatan 3, Box 576,  	75123 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://zorn.bmc.uu.se/~spoel
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++