Multiwfn forum

Hello,

I wanted to calculate transition dipole moments using the NTOs, so I used Gaussian to output me a formchk containing all the NTOs, then I can construct the transition density and do a simple integral. But, this gave me a wrong answer. After checking my code, and not finding anything, I decided to use Mutltiwfn and feed it the original formchk and the log file from Gaussian, so that it gives me back the NTOs, and of course I took the IOp into account.
Surprisingly, the NTOs constructed by the formchk outputted by Multiwfn gave the right answer that I was waiting, and yet Gaussian didn't. Now, this is already enough for me, as I can continue my work.

Though, I'm just curious on how does Gaussian outputs its NTOs in the formchk? Because to my knowledge, it should just replace the MO coeffcients by the NTO coefficients. Yet, apparently, this is not the case.

Thank you.