The question about partial derivative of energy to calculate force

[AlienMarkWong]

Hi Nong:

I have a question about the partial derivative of energy to calculate force

the source code in src/ext/symmfunc.f90: the subroutine of sf_F1_ijk

I think the partial derivative of atom i  about j or k should is :

dFijk_dRj = prefactor*( -cost*( vecRij/Rij/Rij ) + vecRik/Rij/Rik )

dFijk_dRk = prefactor*( -cost*( vecRik/Rik/Rik ) + vecRij/Rik/Rij)

I have pull request in Github and I hope you can check again, thanks.

[Nongnuch Artrith]

Dear AlienMarkWong,

Thank you for the pull request.

We have a unit test that compares the analytic symmetry function derivatives with numerical derivatives, so I am surprised that you think the derivatives are incorrect.  However, I just confirmed that your modified version also passes the unit test, so either the versions are equivalent or the test needs hardening.  We will take a closer look at your pull request.

The unit test is "src/tests/test_symmfunc.f90" and all tests are run with "make tests -f makefile/<Makefile>" from the main "src" directory.

Thanks,
Nong

I have pull request in Github and I hpe you can check again, thanks.