1
$\begingroup$

Let $S^K$ be the unit sphere embedded in $R^{K+1}$.

$v \in S^K$ is randomly chosen from a uniform distribution over $S^K$.

$A \subseteq S^K$ is a $d$-dimensional sub-manifold ($d \leq K$). Think of A as a patch of $S^K$. $A$ may or may not have a boundary.

I want to compute the distrbution of the inner-products of $v$ and the points in $A$, i.e., the distribution $P(q)$ such that $q = <x,v>, x \in A$. What can we say about the concentration properties of $P(q)$? How does it depend on $vol(A)$ ? By $vol(A)$, I mean the $d$-dimensional volume measure of $A$.

When $A = S^K$, it is well studied; the distribution concentrates very sharply around $q=0$, because most of the points are orthogonal w.r.to $v$. but how does one go about reasoning for smaller sub-manifolds (especially when $d << K$)?

$\endgroup$
1
$\begingroup$

Since you have the same concentration properties when $d=0,$ and $A$ is a point (by rotation invariance), the answer is Yes, there is always concentration.

$\endgroup$

Your Answer

By clicking "Post Your Answer", you agree to our terms of service, privacy policy and cookie policy

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