jl 发帖数: 398 | 1 向量 Y = X + E
在 L_2 空间, X 和 E 正交;
|Y|^2 = |X|^2 + |E|^2 ;
有没有哪个定理说:
在L_p, p>2, 如果 |X| >0
|Y|^p > |E|^p
多谢!
|
x******a 发帖数: 6336 | 2 what is 正交 in L_p?
【在 jl 的大作中提到】 : 向量 Y = X + E : 在 L_2 空间, X 和 E 正交; : |Y|^2 = |X|^2 + |E|^2 ; : 有没有哪个定理说: : 在L_p, p>2, 如果 |X| >0 : |Y|^p > |E|^p : 多谢! :
|
jl 发帖数: 398 | 3 The inner product is zero in L2 space.
【在 x******a 的大作中提到】 : what is 正交 in L_p?
|
l********e 发帖数: 3632 | 4 你没有想过?
Y=1
X=-1
E=2
你说的不是显然不成立吗?
【在 jl 的大作中提到】 : 向量 Y = X + E : 在 L_2 空间, X 和 E 正交; : |Y|^2 = |X|^2 + |E|^2 ; : 有没有哪个定理说: : 在L_p, p>2, 如果 |X| >0 : |Y|^p > |E|^p : 多谢! :
|
jl 发帖数: 398 | 5 X, E are linear dependent in your case.
【在 l********e 的大作中提到】 : 你没有想过? : Y=1 : X=-1 : E=2 : 你说的不是显然不成立吗?
|
l********e 发帖数: 3632 | 6 you didn't get the point and you didn't specify you condition on X and E in
your question.
if you still insist, simply let E=2+i and X=-1-i
【在 jl 的大作中提到】 : X, E are linear dependent in your case.
|
n***p 发帖数: 7668 | 7 The answer is no. Here is an counterexample.
Consider the 2D Euclidean space and let
X =(2,1) and E=(-a, 2a). Then X is perpendicular to E.
Then Y=(2-a, 1+2a). For p=4,
|Y|_4^4 = (2-a)^4 + (1+2a)^4
= 17 - 24a + C_1 a^2 +C_2 a^3 +C_3 a^4
<17 = |X|_4^4
if we choose a>0 to be small enough.
The same idea can be applied to construct counterexamples
for other spaces like L_p function spaces for p=4. I choose
p=4 because I can easily calculate the norms explicitly.
For other values of p, we can use taylor expansions.
【在 jl 的大作中提到】 : 向量 Y = X + E : 在 L_2 空间, X 和 E 正交; : |Y|^2 = |X|^2 + |E|^2 ; : 有没有哪个定理说: : 在L_p, p>2, 如果 |X| >0 : |Y|^p > |E|^p : 多谢! :
|
n***p 发帖数: 7668 | 8 I don't get the point either and appreciate it if you may
clarify further. To me I don't see in what definition
E=2+i and X=-1-i can be orthogonal.
in
【在 l********e 的大作中提到】 : you didn't get the point and you didn't specify you condition on X and E in : your question. : if you still insist, simply let E=2+i and X=-1-i
|
l********e 发帖数: 3632 | 9 The point is there is no inner which is compatible with the norm for the L^p
(p>2) space. They are only Banach spaces, so it does not make any sense to
talk about orthogonality in such a space.
So basically there is no restriction on X and E in the decomposition in OP's
question.
【在 n***p 的大作中提到】 : I don't get the point either and appreciate it if you may : clarify further. To me I don't see in what definition : E=2+i and X=-1-i can be orthogonal. : : in
|
n***p 发帖数: 7668 | 10 Since L^p functions are naturally in L^2, it is OK to talk about
orthogonality of L^p functions in L^2. When talking about functions
in L^p\cap L^2, there are naturally two norms, one from L^p, and
another from L^2. It is OK to discuss the relation between these
two norms.
^p
to
's
【在 l********e 的大作中提到】 : The point is there is no inner which is compatible with the norm for the L^p : (p>2) space. They are only Banach spaces, so it does not make any sense to : talk about orthogonality in such a space. : So basically there is no restriction on X and E in the decomposition in OP's : question.
|
l********e 发帖数: 3632 | 11 no, the claim "L^p functions are naturally in L^2" depends on the measure of
the whole space to be finite, which is apparently not assumed by OP.
【在 n***p 的大作中提到】 : Since L^p functions are naturally in L^2, it is OK to talk about : orthogonality of L^p functions in L^2. When talking about functions : in L^p\cap L^2, there are naturally two norms, one from L^p, and : another from L^2. It is OK to discuss the relation between these : two norms. : : ^p : to : 's
|
n***p 发帖数: 7668 | 12 Come on. Apparently OP implicitly made such assumptions.
of
【在 l********e 的大作中提到】 : no, the claim "L^p functions are naturally in L^2" depends on the measure of : the whole space to be finite, which is apparently not assumed by OP.
|
l********e 发帖数: 3632 | 13 哪里显然?这里毫无上下文你也能分析出这个条件,算你牛。
我都怀疑LZ是指不知道这个条件。
【在 n***p 的大作中提到】 : Come on. Apparently OP implicitly made such assumptions. : : of
|