IMP-dev February 2008

imp-dev@salilab.org

4 participants
61 discussions

IMP faq
by Daniel Russel 07 Feb '08

07 Feb '08

I started an IMP FAQ since I realized I was confused about memory management of some types of objects. It can be found at https://salilab.org/imp/wiki/faq and probably should be linked from somewhere.

2 1

Grid class
by Daniel Russel 06 Feb '08

06 Feb '08

Here is a proposed grid class (I haven't even tried to compile it). It can map between real space and voxels. The differences from the grid used in EMBed are: - The voxel index is actually a triple if ints. This makes computing a point in a voxel fast without the large tables used in EMBed. I think this is the right choice speed wise, but we can always test (and it is trivial to change). - the names are changed to be more consistent with IMP - the voxel data is a template argument - it is copyable - currently there is no resize function and I don't really know what would be meant by it and the obvious things involve lots of moving of data around.

3 4

Improved(?) nonbonded list
by Daniel Russel 06 Feb '08

06 Feb '08

Here is a patch for grid based non-bonded list (and the working grid code). All the tests pass, yada,yada. The interface of the nonbonded list does not change significantly. One issue is that as it currently stands, horrible things happen if two different bits of code iterate through the list of non-bonded particles with different distance cutoffs at the same time. The possible solutions to this problem are either to implement iterators which generate the list on the fly (slightly messy code but not bad), or pull the distance cutoff out of the iteration and into the ScoreState itself. I don't really like doing the latter as different restraints will have different distance cutoffs and it seems a waste to build different nonbonded lists for them solely because of that. So I will probably implement iterators at some point. Anyway, it is not a pressing concern. Also, the grid is rebuilt every iteration. I will eventually make it lazier. The grid has improved slightly: - there are now two types of grid indexes: Index, which always refers to a real voxel and VirtualIndex which count be something like (-10, -5, -6). These can be mapped onto real cells in various ways. There are also iterator which allow iteration over rectangular subregions of the grid.

2 1

MolecularHiearchyDecorator fix
by Daniel Russel 06 Feb '08

06 Feb '08

Keren found a bug in MHD. It is fixed here and some test code added to check for it. Various checks were added in decorator-related code in tracking down the bug.

1 0

ListRestraint
by Daniel Russel 06 Feb '08

06 Feb '08

I implemented a 1 particle analog of the new Nonbonded restraint. It stores a list of particles and applies a SingletonScore to each particle in the list. One SingletonScore is provided which applies a unary function to the distance to a fixed point. Other ones we might want are ones which restraint based on a single coordinate or pairs of coordinates. On a separate note, I would move that we remove SphericalRestraint immediately and DistanceRestraint eventually as both roles are better served by bulk restraints.

2 3

Nonbonded list
by Daniel Russel 05 Feb '08

05 Feb '08

The nonbonded list reports pairs of particles based on the distance between their centers. This is fine as long as all spheres have roughly the same radius. However, if some spheres are much larger than others, then we have problems as the sphere distance may be small when the center distance is large. Offhand, I don't have a simple solution to this problem.

1 0

Nonbonded list
by Daniel Russel 05 Feb '08

05 Feb '08

1 0

Nonbonded list
by Daniel Russel 05 Feb '08

05 Feb '08

1 0

Rewrite of connectivityrestraint
by Daniel Russel 05 Feb '08

05 Feb '08

I have a rewritten connectivity restraint. The key differences are that - it now looks like the other new restraints (it uses a PairScore and Particles) - the equivalence sets are described purely through the structure of the restraint, rather than, as before, through an attribute. This makes the code simpler and does not remove any documented functionality. - it is actually documented what it does - the test code is simpler I wanted to get people's comments on the new interface and all.

1 0

VRML tweaks
by Daniel Russel 05 Feb '08

05 Feb '08

The VRML log is added to the log documentation module and cosmetic code tweaks.

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

IMP-dev February 2008