Domain decomposition with position and angle restraints

User discussions
1

GROMACS version: 2019.4
GROMACS modification: Yes/No
Here post your question

Hi,
I’m trying to run a protein simulation in water that is subjected to a couple of restraints. One is a position restraint that is applied on a virtual site by selecting some atoms (using virtual_sitesn). Also I am applying an [ angle_restraints_z ] calculated between a pair of atoms which are far apart in the protein.

The problem I am facing is that the restraints are giving an error since during domain decomposition the atoms required to evaluate the virtual_site and angle restraints are sometimes not in the same DD cell which leads to “atom missing errors”. Is there any way to solve this? I am more concerned about the angle restraints since for the virtual site I am able to choose atoms nearby and evade that error. But for the angle restraint, it has to be calculated by atoms which are far apart.

2

You can try setting the -rdd option of mdrun. I suppose the error message also gave this hint. Note that increasing the value of -rdd increases the communication cost.

3

Thanks for the tip! I have tried that and it definitely increases the computational cost. I was wondering is there any other way to do that.

4

Another option is to use the pull code instead for the angle restraint.