PDA

View Full Version : birthday paradox



Jackie Chan
06-12-2008, 05:47 AM
I am back because of an interesting thing called Birthday Paradox

Let's see how many of us share a common birthday.

I will edit this post to add to the list, just post your replies in this format
UserName MonthName DayOfMonth



1) Jackie Chan April 7
2) Prelude Nov 11
3) Neo1 October 11
4) brewbuck February 7
5) ahluka June 7
6) ping December 15
7) anon October 25
8) indigo0086 June 13
9) Mario F. August 23
10) BMJ May 29
11) maxorator March 25
12) SlyMaelstrom August 10
13) abh!shek January 10
14) P4R4N01D 28 February
15) QuantumPete 11th August

Prelude
06-12-2008, 07:58 AM
Cboard has a calendar, you know. It's probably going to give you a better sample than a few people on a single thread.

Prelude Nov 11

Neo1
06-12-2008, 08:07 AM
Neo1 October 11

brewbuck
06-12-2008, 09:22 AM
brewbuck February 7

cboard_member
06-12-2008, 09:34 AM
ahluka June 7

Officially depressed about turning 20 :(

PING
06-12-2008, 09:37 AM
ping December 15

anon
06-12-2008, 09:38 AM
anon October 25

We could probably get 365 replies before we find 2 people that have birthday on the same day (it would be great to disprove the paradox :))

indigo0086
06-12-2008, 09:48 AM
June 13....

I'm Serial.

Mario F.
06-12-2008, 10:00 AM
Alas. The theory falls.

August 23.

brewbuck
06-12-2008, 10:15 AM
Well, it doesn't help if a person with a matching birthday deliberately doesn't post theirs :)

BMJ
06-12-2008, 10:17 AM
BMJ May 29

:)

Mario F.
06-12-2008, 10:32 AM
Well, it doesn't help if a person with a matching birthday deliberately doesn't post theirs :)

That's the beauty of these probability theories; they explain nothing because they can't be disproven. Only proven. That is, they are true because that's just the nature of probabilities.

Or better yet, for every set of results that disprove the theory, one can always argue a bigger set can prove it.

I'm not sure why we should waste time on it :)
The exact same theory can immediately be applied to people's height, weight, cars with the same two initial letters in their license plates, poking my nose at the same time as someone else...

maxorator
06-12-2008, 10:36 AM
March 25

whiteflags
06-12-2008, 12:34 PM
That's the beauty of these probability theories; they explain nothing because they can't be disproven. Only proven. That is, they are true because that's just the nature of probabilities.


I wouldn't say that the likelihood of an event occuring (even if in real world demonstration it is very unlikely because only a subset of humanity will ever produce an evenly distributed sample) makes the statement any less of a fact. You could simulate a number of assumptions by computer:



// Nobody in the sample was born on February 29
// Everyone's birthdays in the sample are evenly distributed

double bd_paradox (int n)
{
return n == 1 ? 1.0 : (365.0 - (n - 1)) / 365.0 * bd_paradox(n - 1);
}

#include <iostream>
#include <ostream>
#include <iomanip>
int main ()
{
for ( int n = 1; n <= 365; n++ ) {
std::cout << n << ": " << std::scientific << 1.0 - bd_paradox(n) << '\n';
}
return 0;
}


1: 0.000000e+000
2: 2.739726e-003
3: 8.204166e-003
4: 1.635591e-002
5: 2.713557e-002
6: 4.046248e-002
7: 5.623570e-002
8: 7.433529e-002
9: 9.462383e-002
10: 1.169482e-001
11: 1.411414e-001
12: 1.670248e-001
13: 1.944103e-001
14: 2.231025e-001
15: 2.529013e-001
16: 2.836040e-001
17: 3.150077e-001
18: 3.469114e-001
19: 3.791185e-001
20: 4.114384e-001
21: 4.436883e-001
22: 4.756953e-001
23: 5.072972e-001
24: 5.383443e-001

// and so on ...


But it certainly has applications. The birthday paradox shows that even if the number of bins is much greater than the number of items, the probability that two items will randomly be put into the same bin is still very high. The lesson to be learned in relation to computer science is that efficient handling of the bins containing lots of items is always necessary.

Mario F.
06-12-2008, 12:37 PM
Hmm... interesting point.

SlyMaelstrom
06-12-2008, 01:34 PM
SlyMaelstrom August 10

brewbuck
06-12-2008, 01:41 PM
SlyMaelstrom August 10

Ding ding ding ding ding!

That's my son's birthday.

Mario F.
06-12-2008, 01:55 PM
Oh. Well then, in that case there's been another two hits. On my late mother's and my sister birthdays.

brewbuck
06-12-2008, 02:13 PM
Oh. Well then, in that case there's been another two hits. On my late mother's and my sister birthdays.

Yeah, I suppose according to the strict version of the Birthday Paradox it really is restricted to people in the room (or in this case, on this thread), not people who those people know, etc etc...

abh!shek
06-12-2008, 11:05 PM
January 10

P4R4N01D
06-13-2008, 02:37 AM
P4R4N01D 28 February
Note: I was not born on a leap year.

Jackie Chan
06-13-2008, 02:47 AM
In probability theory, the birthday problem, or birthday paradox, pertains to the probability that in a set of randomly chosen people some pair of them will have the same birthday. In a group of 23 (or more) randomly chosen people, there is more than 50% probability that some pair of them will have the same birthday. For 57 or more people, the probability is more than 99%, tending toward 100% as the pool of people grows. The mathematics behind this problem leads to a well-known cryptographic attack called the birthday attack.

We do not have enough samples to really benefit from the paradox. Only persons visiting this board as members should be taken in our experiment.

QuantumPete
06-13-2008, 02:55 AM
QuantumPete 11th August

mike_g
06-13-2008, 05:34 AM
mike_g october 14

zacs7
06-13-2008, 06:12 AM
zacs7 October 17

foxman
06-13-2008, 07:10 AM
foxman October 15

Jackie Chan
06-13-2008, 07:28 AM
1) Jackie Chan April 7
2) Prelude Nov 11
3) Neo1 October 11
4) brewbuck February 7
5) ahluka June 7
6) ping December 15
7) anon October 25
8) indigo0086 June 13
9) Mario F. August 23
10) BMJ May 29
11) maxorator March 25
12) SlyMaelstrom August 10
13) abh!shek January 10
14) P4R4N01D February 28
15) QuantumPete August 11
16) mike_g october 14
17) zacs7 October 17
18) foxman October 15


I am unable to edit my original post now.

Jackie Chan
06-13-2008, 07:31 AM
Oops I almost overlooked it.
Very Happy Birthday wishes for indigo0086!!! :)

DavidP
06-13-2008, 08:09 AM
happy birthday!!!!!!

<Spooky> it's Friday the 13th today. </Spooky>

dra
06-13-2008, 04:02 PM
Sept. 28

idelovski
06-13-2008, 04:50 PM
idelovski, April 25

brewbuck
06-13-2008, 05:34 PM
idelovski, April 25

Another indirect match -- my wife's birthday.

(Actually, come to think of it, ever girl I've ever dated was born on that date. Spooky. But it makes it easier to remember the date :) )

zacs7
06-13-2008, 05:42 PM
> But it makes it easier to remember the date
Associative compression ;)

Always thought you were a computer :rolleyes:

idelovski
06-13-2008, 08:03 PM
(Actually, come to think of it, ever girl I've ever dated was born on that date. Spooky. But it makes it easier to remember the date :) )

So you dated Ren&#233;e Zellweger (http://www.imdb.com/name/nm0000250/)? How was she?

dwks
06-13-2008, 08:24 PM
Assuming Jackie Chan's list in post #27 was correct, we now have:

1) Jackie Chan April 7
2) Prelude November 11
3) Neo1 October 11
4) brewbuck February 7
5) ahluka June 7
6) ping December 15
7) anon October 25
8) indigo0086 June 13
9) Mario F. August 23
10) BMJ May 29
11) maxorator March 25
12) SlyMaelstrom August 10
13) abh!shek January 10
14) P4R4N01D February 28
15) QuantumPete August 11
16) mike_g October 14
17) zacs7 October 17
18) foxman October 15
19) dra September 28
20) idelovski April 25

Only three more to go and then we can buy into the theory that those sharing birthdays just aren't posting. :)

SlyMaelstrom
06-13-2008, 08:47 PM
Or we can just buy into the theory that half the time... a 50&#37; chance won't happen...

dwks
06-13-2008, 08:53 PM
Hmm, well, yes. :p

Spoiler: http://cboard.cprogramming.com/calendar.php?s=&c=1&week=&month=1&year=2008

zacs7
06-13-2008, 09:31 PM
Ahh great, all you've done now is confused me for my statistics exam :s

brewbuck
06-14-2008, 12:00 AM
Ahh great, all you've done now is confused me for my statistics exam :s

It is of course, random. The fact still remains that from a mathematical standpoint, the chance of a hit becomes greater than 0.5 when there are 23 or more people. This doesn't mean the hit will happen, but it becomes more and more likely.

In other words, given an infinite set of sets of 23 people, approximately half of such sets (only approximate, because the actual halfway point is not an integral number of people, but a non-integral number of people is impossible) will contain at least one coincident birthday. But you shouldn't be that surprised if any GIVEN set does not contain a hit, any more than you should be surprised by flipping 8 heads in 10 flips.

Now, if you had a set of 367 people and got no hits, you should be VERY surprised, since there are only 366 days in the year (worst case, counting leap years) :)

brewbuck
06-14-2008, 12:02 AM
So you dated Renée Zellweger (http://www.imdb.com/name/nm0000250/)? How was she?

Eh, she looks nice but isn't exactly my type... I only go for CS grads, preferably those who work at Intel and know all kinds of secrets they are under NDA not to tell me (eventually I married one) ;)

indigo0086
06-15-2008, 12:30 PM
Ahh great, all you've done now is confused me for my statistics exam :s

Hey, just had my stat exam wednsday, probably did well on it.