seeking advice for lammps software example from elastic and elastic_T

Dear friend,

In the example, whether the single crystal Si model has gone through the relaxation process, and then the elastic constant has been calculated. If the model I established is relatively complex, such as the single crystal aluminum has been established first,and then inserted a certain amount of copper atoms by random. Because of the addition of copper atoms, the model structure of aluminum and copper does not reach the equilibrium state. whether it is necessary to add a relaxation step in the original program , and then calculate the elastic constant.if you know, please tell me. Have a good day.

I don’t know if I would call this “relatively complex.” The number of instances of simulations such as this in the literature are abundant…

If you are going to calculate elastic constants using molecular statics then you need to relax at both the initial and final steps. If you replace Si with Cu then there will be some induced lattice strain which must be relaxed. If you read the LAMMPS example you can see that minimization is performed both before and after deformation already…

I would take some more time to get familiar with reading LAMMPS scripts and the documentation.


Hi Dylan. Thank you for your replay. In the example of elasctic there are minimization, but in the example of elastic_T ? After establishing the model of aluminum and copper, whether it is necessary to insert the minimum treatment or relaxation into the original program to make the structure of the model reach a stable state.

before continuing the discussion, please read:
…and also read publications explaining what is summarized there.

there is a fundamental difference between the two folders in what they compute and how it is done.
there are also comments in the scripts themselves. for example in the ELASTIC folder it says explicitly, that care must be taken that each minimization is properly converged and it even suggests how this can be tested.
how you get to a converged minimized state at T=0K is up to you. it is just logical, that if you start from a well relaxed initial configuration, it will be easier to get to the desired minimum, that from a high potential energy configuration. also, you have to consider how probable it is that a deformation will nudge your system into different (local) minima (and thus invalidate results).

please note, that in research there are rarely cases where there are “do this, not that” rules that can be blindly followed. most of the time one has to check for each case at each step that what is done is reasonable and consistent with the purpose of the study.