Recent content by redlegend

I was wondering whether you have the proof to the following lemma: "The Ritz values at step n of the Lanczos iteration are the stationary values of the Rayleigh quotient r(x)=(xT A x)/(xT x) if x is restricted to Kn "