Lenard Jones potential for specific atom or between pair of atoms

[Fernando Davó Miralles]

Is there a possibility to add a LJ potential for a specific type of atom or for specific atom pairs?

[David Koes]

No, but it was easy enough to add. I've pushed a commit ([7bbb14]). Let me know how it works.

[Fernando Davó Miralles]

Thanks a lot for the effort.
I tried it but adding the function with this sintax didn't work:
100.0 atom_type_lennard_jones(t1=N,t2=Fe,o=0,_w=3,_c=8)

[David Koes]

Call smina --print_atom_types to see the available atom types. Nitrogen has four different types: Nitrogen, NitrogenXSDonor, NitrogenXSDonorAcceptor, NitrogenXSAcceptor. These are non-overalapping so if you want all nitrogens you need four terms. Fe is Iron.

[Fernando Davó Miralles]

I think the problem is not on the atom types as "100.0 atom_type_gaussian(t1=N,t2=Fe,o=0,_w=3,_c=8)" works perfectly.

I tried this and also didn't work:
"Error with scoring function specification.
Unknown term [atom_type_lennard_jones(t1=Nitrogen,t2=Iron,o=0,_w=3,_c=8)]"

[David Koes]

Did you rebuild smina with the code changes I committed?

[David Koes]

Also, you can see the name of the terms with smina --print_terms
The format is atom_type_lennard_jones(t1=,t2=,o=0,_^=100,_c=8)
Where o is the offset, _^ is the cap and _c is the cutoff.

[Fernando Davó Miralles]

Yes! The version I have is from 30 April 2020

[Fernando Davó Miralles]

Okay, that's true I changed cap by w. Anyway I changed it and gave me the same error:

Error with scoring function specification.
Unknown term [atom_type_lennard_jones(t1=Nitrogen,t2=Iron,o=0,_^=100,_c=8)]

[Fernando Davó Miralles]

When printing terms the function: atom_type_lennard_jones(t1=,t2=,o=0,_^=100,_c=8)
appears

[Fernando Davó Miralles]

When printing terms the function: atom_type_lennard_jones(t1=,t2=,o=0,_^=100,_c=8)
appears

[Fernando Davó Miralles]

Is it possible that w and ^ variables are the same in the code or some problem related to this?

[David Koes]

My fault. This is what happens when you think a code change is simple enough you don't have to test it - you are wrong. I've pushed the fixes (988094a). Main problem was a typo in the regular expression matching the term name, but there was a second issue that is fixed as well.

[Fernando Davó Miralles]

Now, it works perfectly. Thank you very much