Exercise 6.6.7

Answers

1.
It comes from Theorem 6.25(c).
2.
If
T0 = Tn = i=1kλ inT i.

Now pick arbitrary eigenvector vi for the eigenvalue λi. Then we have

0 = T(vi) = ( i=1kλ inT i)(vi) = λinv i.

This means that λin = 0 and so λi = 0 for all i. Hence we know that

T = i=1kλ iTi = T0.
3.
By the Corollary 4 after the Spectral Theorem, we know that Ti = gi(T) for some polynomial gi. This means that U commutes with each Ti if U commutes with T. Conversely, if U commutes with each Ti, we have
TU = ( i=1kλ iTi)U = i=1kλ iTiU
= i=1kλ iUTi = U( i=1kλ iTi) = UT.
4.
Pick
U = i=1kλ i1 2 Ti,

where λi1 2 is an arbitrary square root of λi.

5.
Since T is a mapping from V to V , T is invertible if and only if N(T) = {0}. And N(T) = {0} is equivalent to that 0 is not an eigenvalue of T.
6.
If every eigenvalue of T is 1 or 0. Then we have T = 0T0 + 1T1 = T1, which is a projection. Conversely, if T is a projection on W along W, we may write any element in V as u + v such that u W and v W. And if λ is an eigenvalue, we have
u = T(u + v) = λ(u + v)

and so

(1 λ)u = λv.

Then we know that the eigenvalue could only be 1 or 0.

7.
It comes from the fact that
T = i=1kλ iTi¯.
User profile picture
2011-06-27 00:00
Comments