| 2:05:34 AM PST - Tue, Jan 16th 2018 |
|
I'm experiencing an issue with NWChem 6.6 (completely fixed, worked for LINUX64 with Intel MKL) to perform CDFT counts. In particular, at whatever point I endeavor to apply a charge limitation utilizing the Becke populace conspire, the estimation crashes amid the principal cycle with the mistake message "CDFT: multipliers go over farthest point."
Hunting on the web down this mistake raises an old string from the mailing list where this blunder message is credited to the objective diabatic state being astoundingly unphysical. In any case, I don't trust this to be the situation here; this mistake has been reproducible for an assortment of frameworks, including for frameworks worked from the fitting section surmises substantial partition.
These frameworks meet fine and dandy when Lowdin populaces are utilized.
The negligible framework for which I can duplicate this mistake is a He dimer with a net charge of +1. This will unite fine and dandy when Lowdin populaces are utilized, however comes up short for Becke populaces. Another illustration is the CDFT test incorporated into the NWChem QA suite (QA/tests/cdft_jarv1/cdft_jarv1.nw). This is another situation where meeting is fine when the default Lowdin populaces are utilized, however the "multipliers go over cutoff" mistake happens in the primary cycle if Becke populaces are utilized.
|
|
|