Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I want to know "if the metric projection onto one dimensional (closed) subspace in $L_p (p\neq 2)$ is linear? I think it is not linear, but I can not give a strict proof.

Thanks for any answer!

share|improve this question
    
Welcome to MSE! How did you come to the conclusion? It helps to add those details as that might be as important to answering the question. Regards –  Amzoti Apr 3 '13 at 15:15

1 Answer 1

up vote 0 down vote accepted

No. Consider space $\mathbb{R}_1^2$ and its one dimensional subspace $X=\{(t,t): t\in \mathbb{R}\}$. Then for all $a\in \mathbb{R}$ the distance from $(a,0)$ to $X$ is given by $$ d((0,a), X) =\min_{x\in X} d((a,0), x) =\min_{t\in\mathbb{R}}d((a,0),(t,t)) =\min_{t\in\mathbb{R}}(|a-t|+|t|) $$ You can see that this function is minimized when $t\in[0,a]$. Hence solution of this minimization problem is not unique. So you can not even define metric projection, not to mention its linearity.

share|improve this answer
    
Thanks for your answer, I do hope there are some examples in $L_p (p\neq 2)$. –  user70795 Apr 4 '13 at 8:17
    
By strict convexity there is a unique point that minimizes the distance for $1 \lt p \lt \infty$, so you can define a nearest point projection. The issue is its linearity if $p \neq 2$. –  Martin Apr 6 '13 at 21:31
    
@Martin fixed $\phantom{}$ –  userNaN Apr 6 '13 at 22:09
    
Well, that was a cheap fix :-) I don't think it actually answers the question since it leaves the more interesting cases $p \neq 1,2, \infty$ open. However, since OP seems satisfied, I won't insist... –  Martin Apr 7 '13 at 7:14
    
@Martin I thought a little, it is linear. If you want I can write the proof –  userNaN Apr 7 '13 at 8:15

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.