install.packages(“combinat”) require(combinat)
From Exercise 10, we know that the CDF for the minimum function for the sum of n random varialbes is
P(min(\(X_1\),…,\(X_n\) > x) = (P(X1 > x))\(^n\)
since the probability density for X1 to Xn are all equal.
Converting this equation to exponential form by substituting X1 > x with X1 > e\(^(-x/u)\), we get
P(min (e\(^(-x/µ)^n\))) = P(e\(^-(n/u)*x\)).
Solving for PDF of the CDF, we get the deriative f(x) = n/u*e\(^(-(n/u)*x)\).
The mean of the minimum function for the exponential distribution is then
1/\(\lambda\) = 1/(n/u) or u/n.
The expected time then for the first (minimum) of the 100 (n) lighbulbs to burn out where u = 1000 hours (mean) is
u/n = 1000/100 = 10 hours.
First we note that Z = X1 - X2 is the same as Z = X1 + (-X2) or setting X2 = Y, Z = X + (-Y).
This means that for the convulution formula fz(z) = \(\int_{-\infty}^{\infty} fx(x) f-y(z-x) dx\) = \(\int_{-\infty}^{\infty} fx(x) fy(x-z) dx\) since f-y(z-x) = fy(x-y)
Computing for the CDF then of an expontial density function given the equation above -
for z < 0, fx(x) = [0 for x< 0 and \(\lambda\)e^(-\(\lambda\)x) for x >= 0]
= \(\int_{0}^{\infty} \lambda e^(-\lambda*x) \lambda e^(-x(x-z)) dx\) = \(\lambda\)e^(\(\lambda\)z) * (-1/2 * e^(-2x)\(\lambda\))
= \(\lambda\)/2 * e^(\(\lambda\)*z) for z < 0
for z >= 0
we note that Z = X - Y and thus -Z = Y - X, or the density function fz(z) = -f-z(z)
Which means for z>=0, the CDF density function is the symmetrical opposite of Z < 0
thus for z >= 0
fz(z) = \(\lambda\)/2 * e^(-\(\lambda\)*z)
joining the 2 equations then all z values =
(1/2)\(\lambda\)e^(-\(\lambda\)|z|)
min(1,(100/3)/2^2)
## [1] 1
min(1,(100/3)/5^2)
## [1] 1
min(1,(100/3)/9^2)
## [1] 0.4115226
min(1,(100/3)/20^2)
## [1] 0.08333333