[gmx-users] Re: gmx-users Digest, Vol 59, Issue 44

[Vitaly V. Chaban]

Hi Flo,

Yes, I remember our talk about this problem about a year ago.

So what should be the correct notation now?

I just tried
g_velacc_402f -acflen 1001 -nonormalize -n an.ndx
or
g_velacc_402f -acflen 1001 -nonormalize -n an.ndx -mol
and then g_analyze -integrate -f vac.xvg

but they give me a value close to 0...

Thank you,
Vitaly


>>Please let me know if the g_velacc programme contains any changes
>>since gmx-3.3.3?
>
> Hello,
>
>  yes there have been changes in the code of g_velacc, to allow
>  calculating the diffusion coefficient of molecules and not just the
>  momentum autocorrelation function ( we talked about this already, do
>  you remember ? )
>
>>
>> I try to calculate the diffusion constant with Green-Kubo equation in
>>the way used in 3.3.3:
>>$GMXPATH/g_velacc -acflen 1001 -nonormalize -mol -n an.ndx -o vac1_an.xvg
>>$GMXPATH/g_analyze -f vac1_an.xvg -integrate  > ksd1_an.xvg
>>where the index file contains the numbers of the needed molecules (not atoms).
>>
>>If one makes the same call in 4.0.2 it gives an error:
>>Fatal error:
>>Need a topology to determine the molecules
>>
>>If the topology is given one obtains:
>>Program g_velacc_402f, VERSION 4.0.2
>>Source code file: gmx_velacc.c, line: 77
>>Fatal error:
>>The index group does not consist of whole molecules
>>
>>Why "whole molecules"? - The programme asks for the molecules number
>>initially...
>>
>
> This is a wrong description and has to be changed, because the you have
> to supply an atom index and from the topology file the molecules are
> found.
>
>>Trying
>>g_velacc_402f -acflen 1001 -nonormalize -n an.ndx
>>it works fine but after integration gives an incorrect value of
>>0.00836 (very close to zero) while it has to be about 2.5 (with
>>g_msd).
>
> This sounds strange, because only some parts of the code were changed to
> deal with molecules, but the routine calculating the VACF stayed
> untouched.
>
>>
>>What have been changed since 3.3.3? And what should the correct call
>>be to find the VAC?
>
> Flo