## Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

** Re^3: RHF vs ROHF in CPU TIME? **
Alex Granovsky

gran@classic.chem.msu.su

Dear Siddheshwar,>Thank you for the response. Actually I need to study the neutral molecules, but some have odd number of electrons. I need to study only singlet state. One way I understood is to add a charge, but I can't do that as I need to study the neutral molecules. Please help me understand how to go about it?

If a system has an odd number of electrons it cannot be singlet.

Singlets are possible only for systems with the even number of

electrons. I'd suggest you to double check your chemical structures

to find out any possible errors into them. Perhaps, the origin of

your problems is related to the terminating hydrogen atoms which

contributes additional electrons and thus can change the multiplicity.

On a separate note:

>>>Dear Firefly Users,

>>>Could anyone point out the differences between the RHF and ROHF on the basis of CPU times? I read on some other forum that both of these time scale by N^4.

This statement is not actually true. The N^4 scaling comes from

the formal counting of different four-center two-electron integrals.

This scaling holds indeed for small molecules. For larger ones, one

needs to take into account that for GTO-type basis sets most of two

electron integrals are nearly identically zero and can be safely discarded based on the efficient pre-screening techniques. This

reduces computational costs to about n^2 where n is the number of

atoms. See my previous post at the link below for additional information:

http://classic.chem.msu.su/cgi-bin/ceilidh.exe/gran/gamess/forum/?C35e9ea902bHW-8891-184+00.htm

> But I need to know say for all the parameters left same, how much do they differ in CPU time?

ROHF/RODFT requires from 1x to 2x as much time per iteration as RHF/RDFT.

> I wish to study the neutral molecules using RHF and ROHF (with same parameters) both. Can I compare their computational results: HOMO-LUMO etc. properties?

You cannot use ROHF/RODFT to compare HOMO-LUMO gaps. This is because

these methods do not have well-defined orbital energies. Use UHF/UDFT

instead.

Kind regards,

Alex Granovsky

Mon May 12 '14 2:29am

This message read **586** times