> To make the OB force fields work with large amounts of atoms, > the non-bonded pairs should be detected dynamically.
Well, I'm not sure if that's really a performance win. Probably these need to be *updated* every 10 steps or something. > force field calculations. I noticed that periodic boundaries are one of the > version 3 goals; is this something that is underway? Looking at the code > this seems far from trivial to implement. You can get most of the effects fairly easily in the UFF class. The most important contributions would be the non-bonded ones, and with UFF, that's only the VDW term. To minimize the performance hit, you'd probably want to tag atoms near a cell boundary and have a separate VDW calculation with these atoms which determine whether the shortest distance is with an atom in "normal space" or through a cell boundary. You wouldn't want to do this with all VDW pairs, since the comparison "if" would be a huge performance hit. UFF would probably be OK with such an implementation, since there's no electrostatic term, and VDW falls off quickly. Something like MMFF94 or GAFF requires replicas for electrostatics, but even with the scheme I outlined above, you'd probably get most of the important interactions. The other caveat is bonds across cell boundaries. Again, you can add a bond-through-boundary term to handle this case and detect them at Setup(). > something I'm missing, or a better way to get the UFF atom types? What you describe suggests that we need to add a way to directly access the force field typing without the rest of Setup(). I think that's a generally useful thing. -Geoff ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ OpenBabel-discuss mailing list OpenBabel-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/openbabel-discuss
