✔ 最佳答案
註:粗體色字表行向量(矩陣), u'表 u之transpose
dim[Im(T_i)]=1 => T_i= ui*vi' , for some nonzero vector ui, vi
(Im(T_i)={k ui})
(T_i)^2≠0 => inner product < ui, vi>≠0
(T_i)*(T_j)=0 => <ui, vj>=0 for i≠j
Claim1: u1, u2, u3,..., um為 linearly indep.
設 k1 u1+k2 u2+...+km um= 0
(dot vi )=>ki = 0, i=1,2,...,m
故 u1, u2, u3,..., um為 linearly indep.
而 u1, u2, u3,..., um為V( n-dim )之向量,
故 m<= n
2009-10-07 11:01:59 補充:
Co大好久不見!
2009-10-07 17:12:46 補充:
內積 = ui*vj', 是兩matrix相乘,內積只是好說明而已,
證明過程沒有用到inner product space的性質!
與是否finite field有關係嗎?
2009-10-07 21:35:33 補充:
1. T=u*v', T^2=(uv')*(uv')=u(v'u)v'=(v'u)uv'≠0 , so v'u≠0
2. T1*T2=0 ,(u1v1')*(u2v2')=0, k(u1v2')=0, so k=v1'u2=0
同理vi'*uj=0 (i≠j)
原始想法,並未用到內積性質,也未用||u||=0=> u=0
so, 與 finite field無關!
2009-10-07 22:41:37 補充:
im(T)= col. space of T= 1 dim, 例
[ 1, -2, 3] = T
[ 2, -4, 6]
[ 3, -6, 9]
u=[1, 2, 3]', v=[1, -2, 3]', 則
T=u*v'=
[1][1, -2, 3]
[2]
[3]
矩陣很難keyin, 不知可看否?