CHARMM Development Project
i am wondering at this problem. i was told that, if the timestep is too small, then the simulation runs very slow,if it is too big, the the error will increase. so we should choose a right timestep with acceptable error and maximum speed. but what is the criteria? i have looked at some inp file of charmm, and found they mostly choose a timestep of 0.002 ps. but i do not know why?
thank you for your kind help.
In most cases, use of 0.002 ps is tolerable (but not ideal) if one constrains the bonds to hydrogens via e.g.

SHAKE BONH PARAM

This removes the high frequency C-H, N-H, and O-H vibrations (think of an IR spectrum), and provides some rationale for the larger 0.002 ps timestep. However, we've observed differences in calculated properties between simulations run at 0.001 and 0.002 ps, so we mainly use the smaller 0.001 ps value.
© CHARMM forums