Find a configuration for which this bound is achieved.

33 Views Asked by At

This is the problem from David Pollard Book" Convergence of Stochastic Process".

Can anyone give me some hint here??

Let $S_0$ be a set of N points in $S$. Suppose there is an integer $V\leq N$ such that $D$ shatters no set of V points in $S_0$. Then $D$ picks out no more that ${n \choose 0}+ {n \choose 1}+ .....+ {n \choose {V-1}}$ subset from $S_0$

The above theorem informs us that the class of quadrants in $R^2$ picks out at most $1+\frac{1}{2}N+ \frac{1}{2}N^2$ subsets from any collection of N points. Find a configuration for which this bound is achieved.

Hstrong textow to find Configuration for which this bound is achived??

Definition from the same book :

Let $D$ be a class of subsets of some space $S$. It is said to have polynomial discrimination (of degree v) if there exists a polynomial $\rho(.)$ (of degree $v$) such that, from every set of $N$ points in $S$, the class picks out at most $\rho(N)$ distinct subsets. Formally, if $S_0$ consists of $N$ points, then there are at most $\rho(N)$ distinct sets of the form $S_0\bigcap d $ with$ d$ in $D$. Call $\rho(.)$ the discriminating polynomial for $D$.

Thank you