Consider a unitary matrix ,
where
,
. Therefore,
given any vector
,
Given , we want to find
(and hence
) such that
is a multiple of the
first coordinate vector
. Therefore,
With the requirement that , we obtain
Given
, let
be the
unitary matrix
where is formed by the Householder
transformation for the column vector
. Thus,
It is clear that does not change the first column of
. Repeating this procedure,
is transformed by
orthogonal similarity transformations into an upper Hessenberg matrix.