Intel run-time check failure in call to eeq_chrgeq

[dftd4]

Intel run-time check failure in call to eeq_chrgeq

[awvwgk]

Which version of the Intel compiler? I remember a discussion on false positives Intel runtime checks while working together with the DFTB+ team.

[ajmay81]

The latest (2020.4.304):

> ifort -v
ifort version 19.1.3.304

The case of passing undefined memory address has been reported since forever with every version of the compiler I can remember, I don't really know whether that is actually valid or not if the routine never uses it, but we nonetheless inserted zero size allocations to avoid those ourselves because of the otherwise general usefulness of the run-time checking. The bounds issue I really don't know about, it could well be a false positive.

[awvwgk]

Sorry, I was mixing up passing alloctable arrays to optional dummy arguments here, which were buggy with Intel runtime checks at some point. At least this is how I implemented this part of D4 in DFTB+.

I will see if I can backport the DFTB+ D4 implementation to this repository.
See: https://github.com/dftbplus/dftbplus/blob/a076cd3/prog/dftb%2B/lib_dftb/encharges.F90#L940-L1019