If you keep drawing cards from a shuffled deck, what is the expected number of draws to complete at least one card from each suit?
The expected value is the average number of draws that it would take if you repeated the experiment many times.
The fewest draws is 4 and the greatest number possible is 40. Therefore
the solution for expected outcome is the finite series given by the
following:
4*P(4 cards) + 5*P(5 cards) + ... 40*P(40 cards)
Case 1: 4*(1 * 39/51 * 26/50 * 13/49)
In this case, each sucessive draw reveals a new suit. For all other
cases, let 'N' imply 'new suit' and 'O' imply 'old suit'. Each
successive case will include 39, 26, and 13 in its numerator for each
combination of Ns and Os. For example with case 2:
Case 2: 5*[(P(NONNN) + P(NNONN) + P(NNNON)]
Of course the first N refers to the first card which is always a 'new
suit' and the probability of its draw is identically 1 which I will
hereafter disregard.
Case 2: 5[(12/51 * 39/50 * 26/49 * 13/48) + (39/51 * 24/50 * 26/49 * 13/49) + (39/51 * 26/50 * 36/49 * 13/48)]
Factoring out common elements shows that the expression is:
(5*39*26*13*47!/51!)*(12+24+36)
And we see that each term in the series will be similarly constructed
given the proper indexing for 'old suits'. I am beginning to see
something with i-2 choose i-4 where i is the number of cards to make
our hand., but it is still fuzzy.
|
Posted by Eric
on 2006-06-06 12:08:06 |