Doubt about hybridization and choice of fragments for water molecule.

[Oscar Ac]

Hi,

I've been trying to understand by myself how the XMVB program works, since it's crucial for my project. However, I've found some features which I can't still fully understand with the information provided on the manual and tutorials. This time, I'll use the water molecule to exemplify some of my doubts. I excuse myself beforehand in case my questions are too basic.

1. For the water molecule I calculated (I adjoin the LOG and XMO files), I obtain a lower energy using the hybridization scheme "spx" than with the "spy" and "spz". I also tried "spxpypz" just to play a bit. My question is: 

Why does this happen? I know the orientation of my molecule plays a role but still the coefficients of the MOs in my LOG file don't show any particular benefit for the x axis, as far as I understand.

2. I decided to choose the fragments as SAO because I don't know the criteria of choosing some specific basis functions for the fragmentation per atom. I know that I should choose those which must contribute to every shell and electron pairs but I'm not sure whether this will be efficient for bigger molecules such as coordination compounds, which are my target. So my question would be:

Which is the criteria to propose an adequate fragmentation to create the Valence Bond orbitals from the molecular orbitals in the LOG file?

3. And last but not least, for the time being:

When is it pertinent to choose the option GUESS=MO? So far, I obtain the same result when I use it with the GUS section and when I use GUESS=AUTO.

Thank you for your kind help. Any comments are welcome! =)

I'm a bit lost, so I'm sorry again if I'm missing something from the Valence Bond theory one can find in textbooks in order to create chemically meaningful inputs.

Oscar