Hello Users,
I’m using the MARTINI force field, which uses the LINCS algorithm to constrain the aromatic ring bonds (composed of 3 atoms). I’m running into difficulties doing this. I’m not finding the fix for LINCS so instead I’ve tried fix shake, however, I get the error:
ERROR: Shake clusters are connected
The fix is:
fix 1 all shake 0.0001 10 0 b 1
where bond 1 is the bonds between the 3 atoms of the ring:
Researching this topic previously to try to implement TraPPE-EH in Lammps I found that it is not possible to do fully bond length constrained chains in Lammps due to the difficulty in parallelizing the code (as far as I understood). I suspect a similar thing might be happening with your constrained atoms.
Salomon
Hello Users,
I'm using the MARTINI force field, which uses the LINCS algorithm to
constrain the aromatic ring bonds (composed of 3 atoms). I'm running into
difficulties doing this. I'm not finding the fix for LINCS so instead I've
tried fix shake, however, I get the error:ERROR: Shake clusters are connected
The fix is:
fix 1 all shake 0.0001 10 0 b 1
where bond 1 is the bonds between the 3 atoms of the ring:
\* b1 \* \* \*----* b1
* *
b1 *
*Is there anyway I can constrain those 3 bonds?
it might be possible using fix rigid,
but that will likely limit your time step.
the second option would be to try extending
fix shake to handle your specific case (small triangles).
a third option would be to implement LINCS into LAMMPS.
i am certain that there would be a lot of people that
would appreciate it.
and finally, you might just use gromacs, since that
is what martini was developed for, or use a different
coarse grain force field that doesn't suffer from this
limitation.
axel.
Maybe I'm writing something stupid, but a triangle can also be
enforced using 2 bonds and 1 angle, and this can be treated by SHAKE
(cf. rigid water models).
Maybe I'm writing something stupid, but a triangle can also be
enforced using 2 bonds and 1 angle, and this can be treated by SHAKE
(cf. rigid water models).
i don't think that this is stupid.
in some sense, that is the second
option that i was mentioning. fix
shake categorically rejects all
connected bonds, but there should
be an exception for triangles for
the reason you just mentioned.
i don't know the shake code well
enough to say, whether this would
require a lot of extra coding, or could
be done rather trivially. i almost
expect the latter.
axel.
What Laurent is saying, is that LAMMPS can already
make the 3 atoms in a triangle rigid via fix shake if
you specify it as 2 bonds and 1 angles, instead of
3 bonds. This is how it does rigid water. The angle
is an equivalent constraint to the 3rd bond (and no angle).
To do this with your system, you may need to add an
artificial angle (one per triangle) of a special type,
so that you can specify it with fix shake. And you
may need to delete the 3rd bond. Neither of which
will matter if you are holding it rigid.
Steve