I recently performed calculations on some Iridium complexes using wB97X-D/def2-svpd with a def2-ecp and a PCM using CPCM and SWIG. In looking through the output file, I noticed the warning “Gaussian blur charges not well tested with ECPs”.
I’m hoping for some advice about this warning. Does this mean I shouldn’t really trust the results of such calculations and avoid using SWIG/iSWIG with ECPs? Or does this mean that I could encounter some numerical instabilities and/or problems that I would recognize if they appeared?
I noticed that warning recently and almost removed it. (Not sure who put it there; wasn’t my group.) I’ve only done a few calculations using ECP with PCM but have not encountered any problems.
Unrelated to PCM, do be careful that you are turning on the ECP explicitly for the def2 basis sets. That really ought to be done automatically for the heavier elements because for heavy elements those basis sets are only designed to be used with ECPs, but right now it’s not automatic and it’s not checked.
Update: There is a very old ticket about incompatibility of ECPs with Gaussian blurring, in the context of QM/MM but SwiG also uses Gaussian blurring and that’s the reason for the warning. Meanwhile the ECP code has been completely overhauled since that ticket was posted so it’s not clear this is an issue. I stand by my previous statement that I’ve not had problems but also I’ve not done a whole lot combining ECPs with PCM. If I can find time I will look into it. Meanwhile make sure to be on the lookout for weird behavior and if you see something unusual, please post about it.