Dear Steve,

I was simulating a box of carbon dioxide with fixed bonds and fixed angles. ‘fix shake’ was used for bond and angle constraints. It is very strange that the output SHAKE statistics gives a ‘NAN’ for the averaged angle in the initial configuration. The simulation then broke down with an error information ‘Out of range atoms - cannot compute PPPM’. I’ve checked my input initial coordinates. The triples in CO2 are all in a line or nearly in a line. So I just don’t know how SHAKE produces a ‘NAN’ for the angle.

This is my input data:

http://www.princeton.edu/~bhong/CO2.zip

Thank you

Binster

Dear Steve,

I was simulating a box of carbon dioxide with fixed bonds and fixed angles.

'fix shake' was used for bond and angle constraints. It is very strange that

the output SHAKE statistics gives a 'NAN' for the averaged angle in the

initial configuration. The simulation then broke down with an error

information 'Out of range atoms - cannot compute PPPM'. I've checked my

input initial coordinates. The triples in CO2 are all in a line or nearly in

a line. So I just don't know how SHAKE produces a 'NAN' for the angle.

because the constraint forces are divergent for a 180 degree angle.

you'd have to use one of the fix rigid integrators to keep co2 straight.

please search through the mailing list archives, there have been several

long discussions on the subject.

cheers,

axel.