[gmx-users] Dynamical atom groups in 'index.ndx'

[Vitaly Chaban]

Hi Berk, 

BH> I think this would require complete rewriting of all the (required) analysis tools.
BH> Currently all tools assume that you have a static, fixed length list of atoms.
BH> You can not "simply" make this list dynamic.
BH> It would be very useful to have such a functionality, but one would like to
BH> have this implemented at a higher level.

You are absolutely true here. It is more convenient to firstly treat a trajectory and then apply analysis tools.

Does GROMACS team plan to add such functionality in the nearest future? No matter in what way it will be implemented.


BH> There is currently a (limited) similar functionality via trjorder.
BH> trjorder can order atoms based on distance. 
BH> This effectively makes
BH> you index group dynamic, but the size is fixed.
BH> (although not for g_velacc, which needs time correlation, but there
BH> it becomes difficult to define what you want to calculate exactly
BH> with dynamic groups)

Why not? One can want to calculate a diffusion constant in the solvation shell of an ion,
inside and outside any confinements, like zeolites, nanotubes, etc. To do this, we need molecules which are
currently located there. If the molecule goes out, it should be simply removed to no longer take
part in the calculation. When the other molecule comes to the area of interest the calculation (for example, VACF)
should just start for it from the beginning.

The 'trjorder' as I understand can only resort the trajectory file to have the needed particles at the beginning. The problem
is that when one particle replaces another in the area of interest the analysis tool "thinks" it is the same molecule. I don't know 
if it can lead to significant inaccuracy. Maybe not.


Best,
Vitaly



>> Date: Wed, 25 Jun 2008 15:35:46 +0300
>> To: gmx-users at gromacs.org
>> From: vvchaban at gmail.com
>> Subject: [gmx-users] Dynamical atom groups in 'index.ndx'
>> 
>> Dear Colleagues,
>> 
>> Did not anybody try to implement a possibility to use a changeable
>> 'index.ndx' when analyzing a trajectory (g_rdf, g_velacc, etc)?
>> I mean that the atoms numbers in 'index.ndx' could change,
>> being different for different frames.
>> 
>> This trick can be used for evaluating properties of molecules located
>> in the definite area at the current moment. Now we can control the
>> particles but how can we check if these particles are still located
>> in the same area where they were before they started to move?
>> 
>> Of course, one is able to do it by visualizing trajectory and then
>> edit the particles number by hand but it is quite unlucky solution, as
>> for me.
>> 
>> Probably I'll try to make something like it if time permits.
>> If somebody has got any considerations on this topic, please let me know.



-- 
Vitaly V. Chaban
School of Chemistry
National University of Kharkiv
Svobody sq.,4, Kharkiv 61077, Ukraine
email: chaban at univer.kharkov.ua
skype: vvchaban