[cclib-devel] Fwd: about cclib v1.5.2, HOMO/LUMO orbitals and the use of ECP

[Karol L. <kar...@gm...>]

---------- Forwarded message ----------
From: Otero Calvi <ao...@is...>
Date: Tue, Sep 5, 2017 at 7:55 AM
Subject: about cclib v1.5.2, HOMO/LUMO orbitals and the use of ECP
To: Karol Langner <kar...@gm...>


Hello Karol
I have some comments about cclib v1.5.2.
Normally I use pseudopotentials (ECP) in my calculations with transition
metal in GAMESS, at this point I have some comments.

1. When cclib obtain the HOMO, the value that cclib shows correspond with
an incorrect HOMO. Other programs like Gabedit or Molden shows the correct
HOMO. cclib obtain the core electrons but not use this information to
update the HOMO.

2. In this version of cclib, the inclusion of molden and wfx format to
export, have this problem, when you compare the Molden file generated by
Molden and by cclib have differences, and of course, wfx format too.

I have made a few modification of the gamessparser code to extract
HOMO/LUMO orbitals that correct this problem and is valid to restricted and
unrestricted wavefunction.
I'm not a developer, I send to you this code attached at this mail, if you
consider that it is OK, you can replace this in the original gamessparser
file. Whit this solution the molden format is exported correctly. In the
case of wfx format I think that other corrections are needed. You can use
CdtetraM1B3LYP.log in the regression list of cclib to check the code and
other output of GAMESS that not use ECP.

Best regards,
    Alexis