|
||
Title: Lightbulb Race Post by william wu on Feb 9th, 2006, 2:01am The lifetime of a Brand A lightbulb is distributed as an exponential random variable with parameter 2. A Brand B lightbulb's lifetime is distributed as an exponential random variable with parameter 1. You have two options: to buy both brand A bulbs and chain their lifetimes together (meaning you replace the first bulb with the second as soon as it goes out), or just buy the brand B bulb. What is the probability that the brand B bulb outlives two brand A bulbs? What happens in the limit, if Brand A's parameter is n times the parameter of Brand B's? Note: This problem can be solved with rote dirty integration of conditional densities, but there exists a very simple and clean solution. Source: John Gill |
||
Title: Re: Lightbulb Race Post by desi on Feb 9th, 2006, 2:11pm In the limit [hideb]1/e[/hideb] |
||
Title: Re: Lightbulb Race Post by desi on Feb 9th, 2006, 2:16pm Explanation [hideb] If X(1), X(2) ... X(n) are i.i.d exp r.v's with mean 1. A B bulb has life X(i)/n n B bulbs have life Z(n)=(X(1) + ... X(n))/n ==> a.s. 1 Hence, A - Z(n) ==> a.s A - 1 P(A > Z(n) ) ==> P(A > 1) = 1/e [/hideb] |
||
Powered by YaBB 1 Gold - SP 1.4! Forum software copyright © 2000-2004 Yet another Bulletin Board |