From chemistry-request@ccl.net Tue Mar 5 12:58:34 1991 Date: Tue, 5 Mar 91 12:45:20 EST From: bernhold@qtp.ufl.EDU Subject: Re: BASIS SETS AND POLARIZATION FUNCTIONS To: AHOLDER@VAX1.UMKC.EDU, CHEMISTRY@ccl.net Status: R Certainly many (most?) chemists like to think in terms of orbitals, but this is only a _model_ we have created -- it does not necessarily represent reality. (I'm sure that statement will generate some replies!) And very soon the appropriate definition of an "orbital" comes into question. Consider that to improve the quality of an ab initio calculation, you have two choices: improve the basis or improve the level of calculation. If you improve the basis beyond single zeta, how do you define the orbitals? Say I'm using a triple zeta basis. I get three times more orbitals than the orbital picture tells me how to deal with. Sure, I can consider only the valence orbtials, but there are properties & situations where the core orbtials are important too. And, of course, what if you use floating orbitals or magnetic field dependent orbitals? Improving the level of calculation is the big killer, though. As soon as I change from SCF to a correlated method, I only think of "orbitals" a place to excite electrons to or from. How does a molecule for which correlation effects are significant (there are many examples) fit into the orbital picture? If I tell you that the effect of triple excitations is very important in the description of a molecule, what does that say about the orbitals? IMHO (that's "in my humble opinion") it safest to always think of the basis set in ab inito calculations as mathematical functions which are used to describe the wavefunction. Mapping results into the orbital picture may or may not be appropriate depending on the case. Perhaps when we have enough experience with ab initio calculations we'll be able to develop a model which goes beyond the orbital picture by allowing for the description of correlation effects, etc. -- David Bernholdt bernhold@qtp.ufl.edu Quantum Theory Project bernhold@ufpine.bitnet University of Florida Gainesville, FL 32611 904/392 6365 ---