Periodic-boundary-conditions during relaxation

On sabato 14 giugno 2008, you wrote:

Hi,

On Sat, Jun 14, 2008 at 12:43 PM, Roberto Guerra
robguerra@unimore.it wrote:

Hello,

I wish to ask if the auto-optimization algorithm takes into account
the unit cell constraint, and if not, will this feature be included
in next releases? I think that’s quite simple to implement.

Do you mean the crystal unit cell?

yes I do

In any case:
At the moment the force fields (OpenBabel code) does not include any
periodic boundary conditions. However, in the future the AMBER force
field will be added which is much faster for (explicit solvent)
proteins compared to MMFF94. To simulate these “cells” I will
probably add cube based periodic boundary conditions. This will be
added around July-August I think/hope.

ok, great news

I guess we could also use this for the crystal unit cells. Not sure
how well this would work for small cells because the minimum image
convention requires the cut-off to be smaller than 1/2 cell edge.
(anyone on the list with a better knowledge of crystallography?)

one way to avoid truncation of the potential is to implement Ewald sum
for the potentials. But this is computatianally very heavy (scales as
N^3/2) and complicated.
The simple truncation+shift works well when the cell is big enough, and
that’s quite easy to implement. I think it sould definitely be
implemented.
Moreover if you evaluate the total force on every direction, you can
perform a relaxation also on the cell size. That would be another big
improvement with small efforts.

I Hope this helps…
thank you!
Tim

Rob

Hello,

I wish to ask if the auto-optimization algorithm takes into account the
unit cell constraint, and if not, will this feature be included in next
releases? I think that’s quite simple to implement.

thanks

On Jun 14, 2008, at 6:43 AM, Roberto Guerra wrote:

I wish to ask if the auto-optimization algorithm takes into account
the
unit cell constraint, and if not, will this feature be included in
next
releases?

No, it does not at the moment.

If you’d like to help include this, it would be greatly appreciated.
At the moment, we’ve been pretty busy with some other tasks, and I
don’t know if anyone else is working on anything related to this.

Hi,

On Sat, Jun 14, 2008 at 12:43 PM, Roberto Guerra robguerra@unimore.it wrote:

Hello,

I wish to ask if the auto-optimization algorithm takes into account the
unit cell constraint, and if not, will this feature be included in next
releases? I think that’s quite simple to implement.

Do you mean the crystal unit cell? In any case:

At the moment the force fields (OpenBabel code) does not include any
periodic boundary conditions. However, in the future the AMBER force
field will be added which is much faster for (explicit solvent)
proteins compared to MMFF94. To simulate these “cells” I will probably
add cube based periodic boundary conditions. This will be added around
July-August I think/hope.

I guess we could also use this for the crystal unit cells. Not sure
how well this would work for small cells because the minimum image
convention requires the cut-off to be smaller than 1/2 cell edge.
(anyone on the list with a better knowledge of crystallography?)

I Hope this helps…
Tim