What do you need fix deposit for, if you don’t compute interactions?
Ans :
Thanks I am depositing molecules inside a polymer network, and I am going to computing pair wise interactions by “excluding certain atom types”, this however was going to be my next step. So create atoms won’t work. I wanted to see how efficient fix deposit command actually is by turning of all the pair wise interactions (also bond. angles and dihedral interactions) which is the maximum reduction in time that I could get.
Correct me If I am wrong:
I found that "neigh time " is really huge since lammps keeps track of the special and topological neighbors of the atoms at every step (assuming the deposition is also carried out every step) even though all the interactions are turned off.
So I do not want lammps to build the neighbor list when fix deposit is invoked. The atoms are static.
I guess there is no workaround for this ?
Arun
What do you need fix deposit for, if you don’t compute interactions?
Ans :
Thanks I am depositing molecules inside a polymer network, and I am going to computing pair wise interactions by “excluding certain atom types”, this however was going to be my next step. So create atoms won’t work.
That is not a convincing argument. You can create molecules just as well.
I wanted to see how efficient fix deposit command actually is by turning of all the pair wise interactions (also bond. angles and dihedral interactions) which is the maximum reduction in time that I could get.
Correct me If I am wrong:
Your logic doesn’t make sense.
I found that "neigh time " is really huge since lammps keeps track of the special and topological neighbors of the atoms at every step (assuming the deposition is also carried out every step) even though all the interactions are turned off.
If you turn off interactions, there is no difference between fix deposit and create_atoms. If you want to compute interactions, you need a neighbor list.
So I do not want lammps to build the neighbor list when fix deposit is invoked. The atoms are static.
I guess there is no workaround for this ?
It doesn’t make sense unless you plan to convert lammps into a monte carlo code. At that point, you should switch to a code like towhee.
Axel