[MMTK] pMoldyn scripting problems

hinsen at cnrs-orleans.fr hinsen at cnrs-orleans.fr
Mon Aug 21 09:06:15 CEST 2006


On 18.08.2006, at 00:49, ALEJANDRO MAXIMILIAN SUAREZ wrote:

> Thank you for the quick insight regarding the default units for the  
> nMoldyn
> calculations. However, in my testing I found some very strange  
> behavior with
> rerunning incoherent scattering function generation for the same  
> trajectory
> multiple times.

The incoherent scattering function, like all functions that depend on  
the scattering vector, is not deterministic, i.e. you should expect  
it to yield different results if you run it several times. The reason  
is that nMoldyn uses Monte-Carlo integration to perform an average  
over the orientations of the scattering vector. The reason for doing  
this average is that typical simulated systems are much too small to  
be sufficiently isotropic. Averaging is thus essential to obtain  
agreeement with experiment.

The number of scattering vector orientations for averaging is a  
parameter in the dialog boxes of all q-dependent quantities. The  
default is 50, which is rather small. Higher values will give better  
results, and smaller fluctuations among multiple evaluations, but  
will also take more time; the CPU time is in fact roughly  
proportional to that number.

> The first inconsistency I found actually involved similarity  
> between my results.
> I tried generating incoherent scattering spectra for a single  
> trajectory twice:
> once setting the q_units variable to be inverse Angstroms, and once  
> for inverse nanometers.

After a quick look at the source code, I assume that the units are  
respected only for results written to text files, not for results  
written to netCDF files. The latter are always nm (length), 1/nm  
(wave number), and 1/ps (frequency).

> I'm completely at a loss as to why runs 1 & 3 are so different from  
> 2 & 4 yet
> so similar to eachother. Timewise, I ran runs 1 & 2 independently  
> at separate

I can't say either; this would depend on the implementation details  
of the random number generator.

> generated the ISF spectrum for this trajectory 4 times however I  
> used coherent
> weights instead of incoherent due to the exception:

The incoherent scattering length of argon is zero, which explains the  
exception. A decent error message would be nicer, I'll suggest this  
to the current developers.

Konrad.
--
---------------------------------------------------------------------
Konrad Hinsen
Centre de Biophysique Moléculaire, CNRS Orléans
Synchrotron Soleil - Division Expériences
Saint Aubin - BP 48
91192 Gif sur Yvette Cedex, France
Tel. +33-1 69 35 97 15
E-Mail: hinsen at cnrs-orleans.fr
---------------------------------------------------------------------





More information about the mmtk mailing list