1/3 5.A.34 By definition, 8x + null T 2 V /null T . We have T /null T (x + null T ) = Tx + null T Then T /null T is injective is equivalent to x 2 null T , null T = x + null T , Tx + null T = null T , Tx 2 null T Remark 1. it can be viewed as l.h.s is x = 0, r.h.s is Tx = 0. Then Tx 2 null T , x 2 null T . Suppose x 2 null T \ range T , then 9v; Tv = x ) Tv 2 null T ) Tv = 0. Similarly, if null T \ range T = f0g, suppose x 2 null T ) x = 0 ) Tx = 0 2 null T . Suppose Tx 2 null T also notice that Tx 2 range T , thus x = 0 ) x 2 null T . 2/3 5.A.36 Take 1; x; x2; : : : l.i in PR. Consider V = P (R); U = P (R) ¡ P1(R). Define Tp = xp. We can see that T is linear and U is invariant under T . Then (T /U )(1 + U ) = x + U = 0 + U Since 1 + U = / 0 + U , thus 0 is an eigenvalue of T /U with eigenvector 1 + U . But 0 is not a eigenvalue of T , assume 9p 0 = / 0 2 P (R) s.t xp = 0, which is impossible since x / 0 on R. 3/3 5.B.3 Idea: Split V into two eigenspace null(T + I) and null(T ¡ I). Thus 1 1 1 1 v = (v ¡ Tv) + (v + Tv) ) (T + I) (v ¡ Tv) = (T + I) (v + Tv) = 0 2 2 2 2 V = null(T + I) + null(T ¡ I) Since ¡1 is not an eigenvalue of T , then null(T + I) = 0, otherwise T + I is not surjective thus ¡1 is an eigenvector (5.6). Hence V = null(T ¡ I) ) 8v 2 V ; (T ¡ I)v = 0 ) T = I