[MMTK] Problems adding configuration to Universe Copy

Jaroslaw Kalinowski jak at biogeo.uw.edu.pl
Wed Feb 8 21:00:11 CET 2006


   Hi,

The exception is exacly due to what is says: gradients and configurations 
are instances of ParticlePropery class (or derived classes). Now, for any 
arithmetical operation, at first performed is a test wheather both 
particle properties are related to the same Universe. If not exception is 
raised. Thus when you do:

         energy, gradient = universeCopy.energyAndGradients()
         Dphi0 = gradient.dotProduct(direction)

direction is a property related to original universe, but gradient is a 
property of a new universe being a copy of the first one. You can think 
about ParticlePropety (~Scalar, ~Vector, Configuration) as arrays that 
remember universes they belong to.

The question is why do you need copies of universe? I suppose that what 
you really need is ability to evaluate energy and gradients for a new 
point. So why not just make a copy of the configuration object using 
copyConfiguration() method and after doing calculations revert changes 
using setConfiguration()?

BTW, addToConfiguration method was broken in recent stable versions, see 
http://starship.python.net/pipermail/mmtk/2005/000907.html.

   Best regards,

Jaroslaw Kalinowski




More information about the mmtk mailing list