Most people know "birthday puzzle" (sometimes called "
birthday paradox", I have no idea where "paradox" part came from): how big should be a group of people that there is 50% probability that at least two of them have the same birthday date (answer is 42 23). In my previous group people used to celebrate birthdays (with cake and candles and singing
(
Read more... )
there are N people, there are M days of the year, there are M^N ways to distribute the birthdays. Now, let's pick a day and distribute the birthdays of these people among (M-1) days: you get exactly (M-1)^N ways to distribute them. Pick another day... so you got M*(M-1)^N ways... divide that by M^N.. you get
M-1
M*(---)^N == 0.5. Substitute M==365 (yes I know about leap years)
M
ln 365 + N ln (1-1/365) == ln 0.5
N ~= 365*(ln 730)
Reply
Reply
Reply
Reply
мне кажется я нашел ошибку - ты некоторые разбиения считаешь дважды (когда положим все дни рождения попались в один день ты посчитаешь это M раз, а надо только один).
Reply
Reply
Reply
M-1
(---)^N == Р
M
Для любого другого дня вероятность такая же.
Вот у нас 365 дней, какова вероятность что хоть раз за год произойдет событие, ежедневная вероятность которого Р? Такая же как один минус вероятность того, что событие с вероятность 1-Р произойдет 365 раз.
M-1
(1 - (---)^N)^M = 0.5
M
В общем с несколькими округлениями (тейлор два раза) у меня 2287. КОнсенсус.
Reply
Reply
Правильное решение твоей формулы на W-A дает 2284
Reply
Reply
Leave a comment