Regarding extraction of structural parameter from lammps trajectory

Dear all,
I want to extract bond distance, angle and dihedral angle between particular type of atoms during my observation period. Can anybody suggest me a tool that I can used to extract those information? Thank you.

Maybe the compute bond/local ( ) and compute property/local ( ) commands are useful in this case.

Hope it helps

Στις Τετ, 27 Μαΐ 2020 στις 6:54 μ.μ., ο/η Shrabanti Bhattacharya <[email protected]…24…> έγραψε:

Thank you Evangelos, for you suggestion. Can I use those command to extract the property from a lammps trajectory I already have?


Yes you can. You can “rerun” a simulation (see the command ) and compute several properties in a post-process step.
The rerun command will read a snapshot , compute any quantity of interest based on the configuration read (without performing a real time integration) and proceed with reading the next snapshot.


Στις Τετ, 27 Μαΐ 2020 στις 7:14 μ.μ., ο/η Shrabanti Bhattacharya <[email protected]…24…> έγραψε:

Hi Evangelos,

I tried as you suggested but it is not printing any output regarding the dihedral I want as a average. my input is like below. Could you please check if I am doing correct

a compute will do nothing unless there is something that will use it. that can be, for example with dump local
however, there is no averaging happening. that would require post-processing of the dump file.

on top of that, you need to read the documentation in more detail to understand what the compute does.

i don’t think that you will get the kind of information you seem to be after with any existing “canned” software, but will need to write your own custom post-processing tool. this is because compute dihedral/local will work on “defined” dihedrals in the topology, if you, however, want to compute the (averaged) dihedral angle between all(?) water HO bonds and a pentanol C-C bonds you definitely will have to program your own analysis code. this is too complex and something that a human can understand, but not a software with a couple of lines of input.



Thank you Axel for your reply. I understand that it is always better to write your own post processing code but I tried to use a already developed code first as I am new in the field of MD and not so expert in writing a code. I am learning it although. I will read the compute documentation more carefully again. Thanks for your suggestion.