[ESPResSo-devel] RATTLE

[Axel Arnold]

Hi all,

I got recently a comment from Aimee Bailey on our RATTLE. I am not too
familiar with the method, but if someone is, can you confirm it? Would be
interesting to know...

Axel

> I have been looking at the rattle.c file and can tell you that what is
> coded is not actually the RATTLE algorithm.  The algorithm that's
> coded in espresso, if I'm reading it correctly, calculates all the
> corrections to the bond vectors simultaneously and then subsequently
> applies them to the unconstrained positions.  The RATTLE algorithm,
> however, identifies the first bond, calculates the correction, and
> updates the unconstrained particle positions with this correction
> immediately. Then it moves on to the next bond. The correction in the
> next bond uses the updated positions from the previous bond. I can
> think of a few scenarios where RATTLE would converge but espresso's
> RATTLE wouldn't.