Counting heads - Dynamic Programming

Question

Problem:

Given integers n and k, along with p₁,p₂,..., p_n; where p_i ε [0, 1], you want to determine the probability of obtaining exactly k heads when n biased coins are tossed independently at random, where p_i is the probability that the i^th coin comes up heads. Give an O(n²) algorithm for this task. Assume you can multiply and add two numbers in [0, 1] in O(1) time.

Can somebody help me with developing the recurrence relation so that I may code it. (The question comes from back exercise of chapter Dynamic Programming in book "Algorithms By Dasgupta")

MBo · Accepted Answer · 2012-10-28T13:14:46.997

14

Consider the situation when (n-1) coins are tossed together and nth coin is tossed apart and take into account mutual independence.

Combine probabilities of simpler cases to get P(1..n, k) (where P(1..n, k) is the probability of obtaining exactly k heads when n coins)

Then apply this rule and fill all the cells in NxK table

Edit:

There are two possible ways to get exactly k heads with n coins -

a) if (n-1) coins have k heads, and Nth coin is tail, and

b) if (n-1) coins have k-1 heads, and Nth coin is head

so

P(n, k) = P(n-1, k) * (1 - p[n]) + P(n-1, k-1) * p[n]

edited Oct 28 '12 at 13:14

answered Oct 27 '12 at 13:17

MBo

77,366
5
53
86

So can we use this recurrence relation: P(N,K) = max { P(N-1,K), P(N-1,K-1)} + p_i And start with P(n,k) – Gaurav Oct 28 '12 at 09:59
sorry i dont know how to write subscript here. p_i is p with a subscript i – Gaurav Oct 28 '12 at 09:59
2

No, we don't need to choose maximum. You have to calc probability, not optimum of some kind. – MBo Oct 28 '12 at 12:32
So what will be the recurrence relation then? – Gaurav Oct 28 '12 at 12:54
I was hoping for your solution with some clues... OK, relation formula added – MBo Oct 28 '12 at 13:14

Counting heads - Dynamic Programming

1 Answers1