|
||
|
Title: Birthday Paradox [variant] Post by THUDandBLUNDER on Apr 21st, 2003, 10:55am The so-called Birthday Paradox puzzle asks how many people need there be in a room so that the probability of two (or more) people sharing the same birthday is at least 1/2. Most people - including myself - are surprised to learn that the answer is 23, assuming a uniform distribution of birthdays, 365 days in a year, etc. [It is much less surprising when one considers the number of comparisons of birthdays. 23C2 = 253. Note that 253/365 ~ loge2]. VARIANT: Let there be 2 classrooms, each containing n people. What is the minimum value of n such that the probability of at least one person in EITHER classroom sharing the same birthday with at least one person in the OTHER classroom is more than 1/2? [I have put this in Medium, happy in the knowledge that I have never seen anybody complain, "Hey, dude. This belongs in Hard!"] :) |
||
|
Title: Re: Birthday Paradox (variant) Post by redPEPPER on Apr 23rd, 2003, 3:31pm Hey dude, this belongs in Easy! :) Either that, or I'm oversimplifying. [hide] The probability P1 that a person "a" in class A doesn't share a birthday with a person "b" in class B is obviously 364/365. b just needs to be born on any of the 364 days that aren't a's birthday. The probability P2(n) that a person "a" in class A doesn't share a birthday with n persons b1, b2,... bn in class B is (P1)n. Each person of class B needs to be born on another day than a's birthday. The probability P3(n) that n persons a1, a2,... an in class A don't share a birthday with n persons b1, b2,... bn in class B is (P2(n))n. Each person of class A needs to comply to P2(n). so P3(n) = ((364/365)n)n The probability P4(n) that a person in class A shares a birthday with a person in class B is 1 - P3. We're looking for a n such that P4(n) > 0.5, or P3(n) < 0.5. I have n = 16 as a very close answer: P4(16) = 50.457%) I expected this to be simpler than the original riddle but I didn't really expect the answer to be lower.[/hide] But I could be wrong. |
||
|
Title: Re: Birthday Paradox (variant) Post by THUDandBLUNDER on Apr 24th, 2003, 9:09am Quote:
Assume n = 366. Obviously, they can't all have different birthdays. Yet your formula produces a non-zero probability. |
||
|
Title: Re: Birthday Paradox (variant) Post by visitor on Apr 24th, 2003, 12:41pm Everyone in the first classroom certainly could have a different birthday from everyone in the second classroom no matter how large the classes are, because as your question is stated only a match from one room to the other counts, not matched birthdays within a room. So you could have a thousand in one room all born in the summer and a thousand in the other room all born in the winter (not likely if the rooms are random, but possible). |
||
|
Title: Re: Birthday Paradox [variant] Post by THUDandBLUNDER on Apr 24th, 2003, 6:39pm Yeah, I got my wires crossed there. I was disagreeing with the following: Quote:
I think the above claims that P3(n) = [P2(n)]^n = [P1(n)]^n2 But The probability that person 'a1" in class A does not share a birthday with any person "bn" in class B and the probability that person 'a2" in class A does not share a birthday with any person "bn" in class B are not independent. So we can't just multiply them. A simple example: let A(1), A(2), B(1), and B(2) each toss a fair coin (p = 1/2). The probability that A(1) doesn't share the same result as either B(1) or B(2) is (1/2)2 = 1/4. Similarly for A(2). But the probability that neither A(1) nor A(2) share the same result as either B(1) or B(2) is not (1/4)2 = 1/16, but 1/8. That is to say, there are only 2 outcomes out of 16 where this happens: (A(1), A(2), B(1), B(2)) = (H,H,T,T) and (T,T,H,H). |
||
|
Powered by YaBB 1 Gold - SP 1.4! Forum software copyright © 2000-2004 Yet another Bulletin Board |