You are not logged in.
Sorry for the ten posts.
Now - a little break
IPBLE: Increasing Performance By Lowering Expectations.
Offline
Have you used my model?
Acctually it has a little error-it sumes the amount of bins is no lesser than that of balls. Once the former exceeds the latter, the probability of the bins amount of seperate balls needs remodelling-after all you can only throw a ball in a repeated bin among all bins each already with one ball in it.
X'(y-Xβ)=0
Offline
No I haven't.
this doesn't depend of n. Using Pingenhole proncipe, my sum ends with b+1 balls, because it's thoughtless to go further.
I used something like this - I'm cuonting all the possible combinatins with k balls and computing the probability to make a combination of k balls, multyplying them, and summazing for all k form 2 balls to b+1 balls. This gives me the aritmetic mean of the wheighted elements in the list, so the expected number of balls. It's tricky to understand, that the probability to get a combination with mare balls is less than the probability to get a combination with less balls.
And, as you can see from the picture above, the formula works!
Last edited by krassi_holmz (2006-12-29 00:29:43)
IPBLE: Increasing Performance By Lowering Expectations.
Offline
The combination is okay, but the S,L,E functions are obtruse.
X'(y-Xβ)=0
Offline
The L function is the result.
In the above plot the ponts represent the information given by Ricky in his post. So the L functtion is the answer.
The S function was defined by me to help myself. It gives "the number of leafs of depth k in the characteristic tree of the problem".
I'm not sure you'll undersand this, because i can't explain it. I've made to pictures for me, but I'll post them.
The E function is a mathematical function:
http://mathworld.wolfram.com/En-Function.html
Last edited by krassi_holmz (2006-12-30 03:47:12)
IPBLE: Increasing Performance By Lowering Expectations.
Offline