[gmx-users] weird problem on small molecules with triple bonds

[David Mobley]

David and all,

> Sorry, my brain was on vacation (about the N). Anyway, having C2 as VS
> would mean the molecule remains perfectly linear as it should be, and
> you do not have to define angles of 180 degrees which gives you an
> ill-defined force.

Isn't there a robust implementation that wouldn't have this problem?
As I pointed out, I can run the molecules in question in AMBER with no
problems. Presumably you/we could look at what they did...

Another thing I just remembered that backs up this interpretation: I
previously ran one of these molecules containing a nitrile in a
binding site, where there was no problems, but when I simulated it in
vacuum and water I ran into these problems. Now I can make sense of
that: in a binding site it would tend to be strained and thus not be
nearly linear, and so not crash -- but in water/vacuum, it would tend
to adopt its preferred conformation and so be nearly linear.

Should I submit a bugzilla?

Thanks,
David