Leira Ruth A. F.
I have modified my input files a little bit and it improved the difference of the energies to about 1E-3 to 1E-6 decimal places. What I did is that in the $SYSTEM group I added kdiag=0 and fastf=.t. then in the $STATPT group, I change my method from QA to GDIIS and added ITBMAT=100 NOREG=5 OPTTOL=1E-5 (as suggested by another colleague). I applied these changes to all my files including the optimization, hessian, sadpoint and irc calculation such that they are uniform in these parts of the input. Attached is a screenshot of one of my output file and from the echo of the input cards you can see the changes I did. The geometries actually looked the same so it must be that the improvement was due to these changes. I'm not really sure how the improvement happened happened though...