2
$\begingroup$

Suppose we have $N$ exponential random variables $X_i$ a with parameter $\beta$ $$P_{X_i}(x)=\beta e^{-\beta x}.$$ We define a random variable $Z_1$ which is the max of all $N$ exponential random variables $X_i$ as $$Z_1=\max\{X_1,\cdots,X_N\}.$$ The the PDF of $Z_1$ is $$P_{Z_1}(z)=N\beta e^{-\beta z}(1-e^{-\beta z})^{N-1}.$$ Using the binomial theorem, we will have at \begin{align} P_{Z_1}(z)=&N\beta\sum_{n=0}^{N-1}e^{-\beta z}\binom{N-1}{n}(-1)^n e^{-\beta zn} \\ = &N\beta\sum_{n=0}^{N-1}\binom{N-1}{n}(-1)^n e^{-\beta z(n+1)} \end{align} My question, can we simplify more than that.

What is the PDF of inverse of $Z_1$, $Y_1$ defined by $$Y_1=\frac{1}{Z_1}$$

$\endgroup$
1
$\begingroup$

Since $\mathbb P(Z_1>0)=1$, the distribution function of $Y_1$ is given by $$ F_{Y_1}(y) = \mathbb P(Y_1\leqslant y) = \mathbb P\left(Z_1\geqslant \frac1y\right) = 1 - F_{Z_1}\left(\frac1y\right), $$ where $F_{Z_1} = \left(1-e^{-\beta z}\right)^N$. Hence $$ F_{Y_1}(y) = 1 - \left( 1 - e^{-\frac\beta y}\right)^N, $$ and the density is obtained by differentiating: $$ f_{Y_1}(y) = \frac{\mathsf d}{\mathsf dy} \left[1 - \left( 1 - e^{-\frac\beta y}\right)^N\right] = \beta N e^{-\frac{\beta }{y}}y^{-2} \left(1-e^{-\frac{\beta }{y}}\right)^{N-1}. $$

$\endgroup$
  • $\begingroup$ thank you so much, So we have dirive $f_{y_1}(y)$ how we do simmulation and compar with analitical $\endgroup$ – Mokhtar Mar 31 '18 at 12:30

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.