Thanks, I learned how the index of the alpha/beta orbitals worked from the program itself and learned the negative index trick from the manual

I did some digging and it turns out the error was with the formchk utility from Gaussian. It mangled the beta orbitals and they were all out of order in the fchk file, but the log file was correct. Curiously, running formchk again on the same checkpoint file gave the correct formatted checkpoint file.

The error was not with Multiwfn, thanks for the help! You can clean this thread up if you want to prune it.

Thanks!
g

Please note that in Multiwfn the index of beta orbitals is after the alpha orbitals. Assume that there are 1000 basis functions, the 5th beta MO will have index of 1000+5 = 1005. Please check if you have mistaken the index. In addition, in the textbox of GUI of main function 0, you can directly input negative index to choose a beta orbital. For example, -5 corresponds to the 5th beta orbital.

For some of the results, the beta orbital energies reported by Multiwfn v3.7 from the .fchk files do not match the values reported by Gaussian in the .log files.

For example, a single point calculation at PW6B95-D3(BJ)/def2-QZVPP gives the beta HOMO and LUMO energies as -0.20128 and -0.07508 in the .log file; whereas they are reported as 10.244396 and 10.123182 by Multiwfn v3.7 (input "0 Show molecular structure and view orbitals" after loading fchk).

I cannot see any trend as to why some fchk files give the correct energies and others don't. There doesn't appear to be any trend with functional or basis set.

I'm using the newest (as of this post, 2019-Nov-17) development version on Ubuntu 19.04.

Please excuse me if this is the wrong place to report a bug.

PS
Thank you Prof. Lu and everyone else involved for writing, maintaining and distributing Multiwfn. It is quite the powerful tool, and I deeply appreciate it. I hope more people follow this example and continue to distribute code for everyone to use.