From owner-chemistry@ccl.net Thu Jan 20 01:44:00 2011 From: "Georg Lefkidis lefkidis++physik.uni-kl.de" To: CCL Subject: CCL: spin contamination Message-Id: <-43705-110120014042-21599-VbpTwCKsLUDflnEjpc8CtA(-)server.ccl.net> X-Original-From: "Georg Lefkidis" Content-Transfer-Encoding: 7bit Content-Type: text/plain; format=flowed; charset="utf-8" Date: Thu, 20 Jan 2011 07:40:15 +0100 Mime-Version: 1.0 Sent to CCL by: "Georg Lefkidis" [lefkidis a physik.uni-kl.de] Dear Prakash, the problem of the UHF is that it still neglects correlations. That means that when virtual excitations are included you will most probably get a wavefunction with lower energy for a clear cut expectation value of . UHF tends to increase because it considers not "completely paired" electrons. However even more unpaired electrons when treated in a CI way can give lower multiplicities than the maximum possible. So you can obtain a singlet with two unpaired electrons \frac{(|ab>-|ba>)}{\sqrt{2}}, and this cannot be found with a single-determinant method like RHF, ROHF or UHF. Therefore high spin contamination can be an indication, but is to be treated with caution. A typical example is a molecule with antiferromagnetic coupling. It is always better to check it with an adequate method before making a statement. Besides the more unpaired electrons the more possible that UHF will converge to a high multiplicity value. For triplets a value of around 2 is a strong indication, but a value of 2.5 still does not mean much if not checked with a CI (or CI-derivative) method. Cheers George prakash surya tarzan11_11^yahoo.com writes: > > Sent to CCL by: "prakash surya" [tarzan11_11^-^yahoo.com] > Hi all > a small doubt. > The wavefunction obtained as a slater determinant under UHF is essentially not > a eigen function of the spin operator s^2. so, higher the value of the , > it signifies the mixing of higher exited states contributed by higher spin > states and the wave functions has lesser physical significance. right ...!!! > so my question is how much of is OK , since during my gaussuian > calcualtions i some times observe a as much as 2.0 > > thanx in advance > pls let me know if i am wrong any way ..:) > email: tarzan11_11_._yahoo.com> > --------------------------------------------------------------- Dr. Georg Lefkidis E-mail:lefkidis(at)physik(dot)uni(dash)kl(dot)de University of Kaiserslautern Tel. +49 631 205 3207 Department of Physics Fax: +49 631 205 3907 PO Box 3049 67653 Kaiserslautern Germany ---------------------------------------------------------------