Kowariancja funkcji zmiennych losowych
: 25 cze 2018, o 18:55
\(\displaystyle{ X_{i}, i=1, ..., n, n \ge 1}\) — niezależne zmienne losowe \(\displaystyle{ U\left(0, 1 \right)}\),
\(\displaystyle{ S_{n}= \sum_{i=1}^{n}X_{i} ,\\
Z_{n}=\max \left(X_{i}, ..., X_{n}\right)}\).
Obliczyć \(\displaystyle{ Cov\left(S_{n}, Z_{n}\right)}\).
Mamy
\(\displaystyle{ Cov\left(S_{n}, Z_{n}\right)=E\left(S_{n}Z_{n}\right)-ES_{n}EZ_{n}}\)
\(\displaystyle{ E\left(S_{n}Z_{n}\right)=\sum_{i=1}^{n}E\left(X_{i}\max \left(X_{1}, ..., X_{n}\right) \right)=nE\left(X_{1}\max \left(X_{1}, ..., X_{n}\right)\right)}\)
Niech \(\displaystyle{ U=\max \left(X_{1}, ..., X_{n}\right), V=X_{1}}\). Dla \(\displaystyle{ 0 \le v \le u \le 1}\) mamy
\(\displaystyle{ F\left(v,u\right)=P\left(V\le v, U\le u\right)=P\left(U\le u\right)-P\left(U\le u, V>v\right)=\\=P\left(X_{1}\le u, ..., X_{n}\le u\right)-P\left(v<X_{1}\le u, X_{2}\le u, ..., X_{n}\le u\right)=vu^{n-1}}\)
\(\displaystyle{ f\left(v, u\right)=\frac{\partial^{2}}{\partial v \partial u}vu^{n-1}=\left(n-1\right)u^{n-2}}\)
\(\displaystyle{ E\left(VU\right)= \int_{0}^{1} \int_{0}^{u} vu\left(n-1\right) u^{n-2} \mbox{d}v \mbox{d}u= \frac{n-1}{2\left( n+2\right) }}\)
\(\displaystyle{ E\left(S_{n}Z_{n}\right)= \frac{n\left(n-1\right)}{2\left(n+2\right)}}\)
\(\displaystyle{ ES_{n}=nEV=\frac{n}{2}}\)
\(\displaystyle{ F\left(u\right)=u^{n}, f\left(u\right)=nu^{n-1}}\)
\(\displaystyle{ EZ_{n}=EU=\frac{n}{n+1}}\)
Co tu jest źle?
\(\displaystyle{ S_{n}= \sum_{i=1}^{n}X_{i} ,\\
Z_{n}=\max \left(X_{i}, ..., X_{n}\right)}\).
Obliczyć \(\displaystyle{ Cov\left(S_{n}, Z_{n}\right)}\).
Mamy
\(\displaystyle{ Cov\left(S_{n}, Z_{n}\right)=E\left(S_{n}Z_{n}\right)-ES_{n}EZ_{n}}\)
\(\displaystyle{ E\left(S_{n}Z_{n}\right)=\sum_{i=1}^{n}E\left(X_{i}\max \left(X_{1}, ..., X_{n}\right) \right)=nE\left(X_{1}\max \left(X_{1}, ..., X_{n}\right)\right)}\)
Niech \(\displaystyle{ U=\max \left(X_{1}, ..., X_{n}\right), V=X_{1}}\). Dla \(\displaystyle{ 0 \le v \le u \le 1}\) mamy
\(\displaystyle{ F\left(v,u\right)=P\left(V\le v, U\le u\right)=P\left(U\le u\right)-P\left(U\le u, V>v\right)=\\=P\left(X_{1}\le u, ..., X_{n}\le u\right)-P\left(v<X_{1}\le u, X_{2}\le u, ..., X_{n}\le u\right)=vu^{n-1}}\)
\(\displaystyle{ f\left(v, u\right)=\frac{\partial^{2}}{\partial v \partial u}vu^{n-1}=\left(n-1\right)u^{n-2}}\)
\(\displaystyle{ E\left(VU\right)= \int_{0}^{1} \int_{0}^{u} vu\left(n-1\right) u^{n-2} \mbox{d}v \mbox{d}u= \frac{n-1}{2\left( n+2\right) }}\)
\(\displaystyle{ E\left(S_{n}Z_{n}\right)= \frac{n\left(n-1\right)}{2\left(n+2\right)}}\)
\(\displaystyle{ ES_{n}=nEV=\frac{n}{2}}\)
\(\displaystyle{ F\left(u\right)=u^{n}, f\left(u\right)=nu^{n-1}}\)
\(\displaystyle{ EZ_{n}=EU=\frac{n}{n+1}}\)
Co tu jest źle?