_HIST file

Documentation, Web site and code modifications

Moderators: baguetl, routerov

Locked
Jordan
Posts: 282
Joined: Tue May 07, 2013 9:47 am

_HIST file

Post by Jordan » Mon May 04, 2015 5:24 pm

Dear all,

The _HIST file contains very useful data for MD simulation but lacks some information by itself.
Indeed, without the _OUT.nc file, there is no way to know what kind of atoms are studied and what type of atom each atom is.

I would like to add the ntypat, typat and znucl variables in the _HIST file as well as some other relevant variables for MD like dtion which again is in _OUT.nc rather than in _HIST.
Those data should be written only once at the beginning (when there are actually written in the _OUT.nc).

Of course, those variables would remain in the _OUT.nc file.

Do you have any objection ?

Cheers,

Jordan

mverstra
Posts: 655
Joined: Wed Aug 19, 2009 12:01 pm

Re: _HIST file

Post by mverstra » Mon May 04, 2015 7:12 pm

exquisite idea. Go for it!

The file has some overlap with the GSR and other .nc files but serves its own purpose. It is convenient, but incomplete as you mentioned.

ciao!
Matthieu Verstraete
University of Liege, Belgium

User avatar
gmatteo
Posts: 291
Joined: Sun Aug 16, 2009 5:40 pm

Re: _HIST file

Post by gmatteo » Tue May 05, 2015 10:49 am

Hi Jordan,

I've already implemented this modification in my 7.11 branch but the code must be merged by
Xavier in trunk. Before doing so, however, I have to remove all the calls to the etsf-io library
used to write the data in netcdf format since, as we agreed in Liege, Abinit8 won't rely on etsf-io anymore.

Cheers,
Matteo

Jordan
Posts: 282
Joined: Tue May 07, 2013 9:47 am

Re: _HIST file

Post by Jordan » Tue May 05, 2015 11:22 am

Great !

If you need help, we can split the work.

Jordan

User avatar
torrent
Posts: 127
Joined: Fri Aug 14, 2009 7:40 pm

Re: _HIST file

Post by torrent » Tue May 05, 2015 11:44 am

Matteo, Jordan,

I didn't look at Matteo's modifs (yes, sometimes, I don't spy!) but I suppose you have stored these time-independent data once for all in the HIST file and not at every time step...
Is it the case ?

Also, if we decide to suppress the necessary use of the OUT.nc file, it is important to store in the HIST file all the quantities needed to analyse a trajectory (not only the ones you need now); for instance, the time step dtion.

A good starting point could be APPA to list all these needed quantities; they are not numerous...
Looking at APPA src files, I found the following variables: natom, typat, amu, dtion, znucl
Could you please add these when committing your changes?

A+
Marc
Marc Torrent
CEA - Bruyères-le-Chatel
France

User avatar
gmatteo
Posts: 291
Joined: Sun Aug 16, 2009 5:40 pm

Re: _HIST file

Post by gmatteo » Tue May 05, 2015 12:38 pm

didn't look at Matteo's modifs (yes, sometimes, I don't spy!) but I suppose you have stored these time-independent data once for all in the HIST file and not at every time step...
Is it the case ?


Yes

A good starting point could be APPA to list all these needed quantities; they are not numerous...
Looking at APPA src files, I found the following variables: natom, typat, amu, dtion, znucl
Could you please add these when committing your changes?


OK

Jordan
Posts: 282
Joined: Tue May 07, 2013 9:47 am

Re: _HIST file

Post by Jordan » Wed May 06, 2015 2:27 pm

torrent wrote:Matteo, Jordan,
Looking at APPA src files, I found the following variables: natom, typat, amu, dtion, znucl


natom is already in the _HIST file ;o)

Ok ok I --->

Locked