Processing math: 100%

 

 

 

Eigenvalues and Lanczos' method, defining the Lanczos' vectors

We have thus ˆAˆqk=βk1ˆqk1+αkˆqk+βkˆqk+1, with β0ˆq0=0 for k=1:n1. Remember that the vectors ˆqk are orthornormal and this implies αk=ˆqTkˆAˆqk, and these vectors are called Lanczos vectors.