What you see is partial filling of bands, try using only one k-point and you will see that your spin is integral.
2007/4/19, Rainer Hoft <[EMAIL PROTECTED]>:
Dear Siesta users, I have a question with regards to "spin contamination", yet I'm not sure if this is the correct terminology to use in my context. SIESTA gives as part of its output the "spin polarisation", SP= Q_up - Q_down. So for a singlet this is 0, for a doublet 1, a triplet 2 etc. But what if Q_up-Q_down is somewhere inbetween 0 and 1. Does this mean that the system is spin contaminated? And if Q_up-Q_down is what one expects, ie 0 for a singlet or 1 for a doublet, does this imply that the system is NOT spin contaminated? I realise that spin contamination means that the expectation value of S^2 is not s(s+1), but im not sure how to relate this to the spin polarisation output of SIESTA. Does siesta provide the expectation value of S^2 ? Any thoughts will be much appreciated. Regards, Rainer
