Quote:
Originally Posted by whosnext
If you love factorials, I think I have the formula for the general problem!!
Let there be L unique letters (in OP L was 11) and let TOTAL be the total number of letters (in OP TOTAL was 23). Let N(J) be the number of the Jth letters (so in OP the N's are 4,2,2,2,2,2,2,2,2,2,1). As above, we will use the index T for the Turn on which the winner is determined.
For simplicity of the formula below, suppose we reorder the letters so that we are interested in finding the probability of the 1st letter.
Then P(1) is a nested sum of L sums:
= Sum of (T from N(1) to TOTAL-(L-1))
Sum of (I(2) from 0 to N(2)-1)
Sum of (I(3) from 0 to N(3)-1)
.
.
.
Sum of (I(L-1) from 0 to N(L-1)-1)
The term that gets summed is a product of three terms (and an indicator term):
Term1 = T!/((N(1)-1)!*(T-N(1)-Sum of (K from 2 to L-1) I(K))! * Prod of (K from 2 to L-1) I(K)!)
Term2 = N(1)!*N(L)! / (N(L)-(T-N(1)-Sum of (K from 2 to L-1) I(K)))! * Prod of (K from 2 to L-1) N(K)!/(N(K)-I(K))!
Term3 = (TOTAL-T)! / TOTAL!
Term = Term1 * Term2 * Term3 * [0 <= T-(N(1)-Sum of (K from 2 to L-1) I(K)) <= N(L)-1]
where the bracket term is 1 if the condition is true and 0 otherwise
So there is a general formula for this type of problem. Whether you call this an "elegant" solution is up to you!
Thank you everybody for your contributions, including the unnamed former mod. Yes, this is the type of solution I was looking for, it's been so long since I've taken that intro to probability course that I was pretty much lost. I spent around a half hour trying to come up with the solution laid out in your prior post, with a similar method, to no avail. So thank you for that, and also your simulation results. At first I thought I overlooked something simple (after statmanhal's post), but felt by my earlier enumeration efforts that P(A) was much less than 1/4 of P(K). I couldn't prove it though so that was bugging me. Thanks again for the insightful post.