[gmx-users] Atom triplets in g_rotacf

[Yun Shi]

Hi all,

Sorry if there is duplicate message.

I want to calculate the rotational correlation time of a molecule like
UDP, which has many atoms. And I guess it would be reasonable to use
multiple triplets (multiple planes as defined by the cross product of
two vectors from a triplet) of atoms in the molecule, since I found
different triplets produced very different rotational correlation
times (by an order of magnitude).

However, using both triplet A and triplet B together as a sextet does
not give the arithmetic mean of using triplet A and triplet B
separately. Does anyone know why?

Practically, how many different triplets of atoms should be considered
for a molecule like UDP?

Thanks for any suggestion,
Yun