compute group/group

thank you very much for you respond.

there are some restrictions for this command which I don’t understand well. and I think it is very complicated and needs numerous runs to be found how it exactly works.
could you please explain a little more? there in no any keyword called bond, coulomb, etc. so how they should be defined?
and you mean I can’t calculate the whole interaction caused by Lennard-jones potential?
so what should I do ? is there any command suitable for this?

HI,

Please keep in mind that compute can calculate the type of interaction energies, e.g. pair, coulomb, bond, dihedral, etc.

Shafat

thank you very much for you respond.

there are some restrictions for this command which I don’t understand well. and I think it is very complicated and needs numerous runs to be found how it exactly works.

i disagree with your assessment. what this compute does is pretty straightforward. the problem here seems to be, that you are making (incorrect) assumptions for what it does instead of paying attention to what the documentation says and thus you are confused that there is a mismatch.

could you please explain a little more? there in no any keyword called bond, coulomb, etc. so how they should be defined?

there cannot be a “bond” command, because this looks only at the non-bonded interactions (based on the neighbor list). it also computes lj (or rather non-coulomb) and coulomb pairwise components together, since that is how this is implemented. if you want to evaluate those separately, you need to record a trajectory first with all interactions enabled and then use the rerun command to get a decomposition into the contributions that you are interested in.

and you mean I can’t calculate the whole interaction caused by Lennard-jones potential?

you need to explain what you mean by the “whole” Lennard-Jones interaction.

so what should I do ? is there any command suitable for this?

unless you can explain more clearly what it is that you want to compute and how this differs from what the compute group/group documentation says, there is no way to give a recommendation. help that can be given is limited by the quality and preciseness of the questions being asked, and your questions are not very well posed.

axel.

thank you so much, yes that is what I exactly did. I use the rerun command and I need to calculate the z component of lj interactions between two group of atoms. in fact the assumption of all pairwise interactions caused by lj potential along the z direction.
just because the results aren’t logical, I want to make sure that these commands give what I expect. here is the command I use in rerun section.

reset_timestep 0
pair_style lj/cut 50.0
pair_coeff * * 0.002967 3.4
compute group ext group/group int


thermo_style custom step c_group[3]
rerun dump.lammpstrj every 1 dump x y z vx vy vz
clear
quit

thank you very much for you respond.

there are some restrictions for this command which I don’t understand well. and I think it is very complicated and needs numerous runs to be found how it exactly works.

i disagree with your assessment. what this compute does is pretty straightforward. the problem here seems to be, that you are making (incorrect) assumptions for what it does instead of paying attention to what the documentation says and thus you are confused that there is a mismatch.

could you please explain a little more? there in no any keyword called bond, coulomb, etc. so how they should be defined?

there cannot be a “bond” command, because this looks only at the non-bonded interactions (based on the neighbor list). it also computes lj (or rather non-coulomb) and coulomb pairwise components together, since that is how this is implemented. if you want to evaluate those separately, you need to record a trajectory first with all interactions enabled and then use the rerun command to get a decomposition into the contributions that you are interested in.

and you mean I can’t calculate the whole interaction caused by Lennard-jones potential?

you need to explain what you mean by the “whole” Lennard-Jones interaction.

so what should I do ? is there any command suitable for this?

unless you can explain more clearly what it is that you want to compute and how this differs from what the compute group/group documentation says, there is no way to give a recommendation. help that can be given is limited by the quality and preciseness of the questions being asked, and your questions are not very well posed.

axel.

you are still not making much sense here.
…as i recommended before. set up a simple test system and compute what you want to determine manually and compare to what you get from compute group/group and what the documentation of compute group/group says. i am quite confident that the latter two parts are consistent, but i cannot tell if the first is in agreement.
most importantly, you are failing to give a decent explanation for what it is exactly that you are aiming to compute. so i maintain, that the mismatch you are seeing is due to your incorrect thinking and making assumptions that are not backed by the documentation.

unless you significantly improve your explanations for what you do. i have nothing to add to this discussion.

axel.