ReaxFF with hybrid pair style: hbondchk failed

LAMMPS LAMMPS Mailing List Mirror
1

Ray,

thank you for the suggestion. Increasing the skin distance to 4 A actually solved the problem for the test input, but it did not help for a more complex system (attached). Considering that hbondchk fails only if one of the atom types is declared as NULL, I can’t help thinking that there is something funny in how reax/c handles NULL types in hbondchk. Or what do you think?

Thanks,

Kimmo

ffield.reax.FC (25.1 KB)

in.test2 (400 Bytes)

test_input2.dat (1.4 KB)

2

Kimmo,

Yes, this does indicate some problem in the source code. Thank you very much for providing a simple input deck that reproduces the problem. I will take a look.

Thanks,
Ray

3

Kimmo,

There was a bug in how reax/c generates NULL atoms’ hydrogen bond list. The attached patch will fix the problem, thanks. Type “make package-update” after applying the patch.

Thanks again for the simple and complete input deck that helped in fixing this bug.

Ray

hbondchk-null-fix.patch (515 Bytes)

4

Kimmo,

There was a bug in how reax/c generates NULL atoms' hydrogen bond list. The
attached patch will fix the problem, thanks. Type "make package-update"
after applying the patch.

please note that you have to go to src/USER-REAXC to apply the patch.
if you apply it in src/, the subsequent make package-update will
overwrite the installed copy with the old version.

axel.

5

added this change to the main distro - will be in the next patch.

Thanks Ray,
Steve