[gmx-users] problem regarding "mdrun" and "g_hbond"

[Mark Abraham]

> I think a warning is a good idea, but maybe 10 is even on the low side
> for that. Since temperature is macroscopic, you'll need enough atoms
> not to deviate too much from the target temperature. I have no
> evidence on which I can make a proper suggestion... My bidding will be
> 50! Going once, going twice... (maybe someone has looked more into
> this matter. David, Berk?).

I added a bugzilla enhancement-request for this feature, suggesting that
if the user tries 3+ T-coupling groups where the smallest is less than 10%
of the smaller of the other two then we probably have a case of bad
planning from the user, and should give a warning. Doubtless there are
better ideas out there... it's after midnight here!

Mark