Using Theorem 3.3 show that the optimal latent vector at corresponds to the smallest latent root of Eq. (10.7).
THEOREM 3.3
(Rayleigh Quotient) Let Σ be a covariance matrix with latent roots λt ≥ λ2 ≥. . . ≥ λp Apand let

for some vector U≠ 0. Then

if and only if U is, respectively, the latent vector of Σ corresponding to the first and last latent root.

