this is for holding javascript data
Hui Li edited excess mass.tex
over 9 years ago
Commit id: b7d7e6c1d60509d478f063d0f65d8a83097ff852
deletions | additions
diff --git a/excess mass.tex b/excess mass.tex
index bc309d9..f772ec3 100644
--- a/excess mass.tex
+++ b/excess mass.tex
...
where the supremum is taken over all families $\{C_j:j=1, ..., m\}$ of pairwise disjoint connected sets. In practice, the distribution $f(x)$ cannot directly obtained from samples, a more convenient way of dealing with the above expression is to convert it into integral space:
\begin{equation}
E_m(\lambda)=\rm sup \sum_{j=1}^{m} H_\lambda(C_j),
\end{equation}
where $H_\lambda=F-\lambda\times\rm Leb$, $F$ is the cumulative distribution and $\rm Leb$ is the Lebesgue measure.