[gmx-users] Re: molecule order optimization?

[Michael Shirts]

The changes in the order did indeed speed up the simulation, by about 30%.
One problem, though is that it is impossible to use settle with this ordering.
If you specify the oxygen in [settles], as so:

[ settles ]
; i     funct   doh     dhh
4       1       0.09572 0.15139

you get the error:

Fatal error: [ file "/home/mrshirts/gmx/share/top/tip4pgmx.itp", line 23 ]:
             Atom index (4) in settles out of bounds (1-4)
             Settle works on atoms 4, 5 and 6

Presumably, if the contraints are tight enough, then you'll get the same
answer whatever contraint method is used, but it might be useful for settle to
handle other molecular orders.

Cheers,
Michael Shirts
Stanford Chemisty