(a) Define Householder reflections and show that a real Householder reflection is symmetric and orthogonal. Moreover, show that if H,A∈Rn×n, where H is a Householder reflection and A is a full matrix, then the computational cost (number of arithmetic operations) of computing HAH−1 can be O(n2) operations, as opposed to O(n3) for standard matrix products.
(b) Show that for any A∈Rn×n there exists an orthogonal matrix Q∈Rn×n such that
QAQT=T=⎣⎢⎢⎢⎢⎢⎢⎡t1,1t2,10⋮0t1,2t2,2t3,2⋱⋯t1,3t2,3t3,3⋱0⋯⋯⋯⋱tn,n−1t1,nt2,nt3,n⋮tn,n⎦⎥⎥⎥⎥⎥⎥⎤
In particular, T has zero entries below the first subdiagonal. Show that one can compute such a T and Q (they may not be unique) using O(n3) arithmetic operations.
[Hint: Multiply A from the left and right with Householder reflections.]