Update README.md

Author: MMMarmelade

README.md

@@ -17,3 +17,5 @@ and
 		w(i,j)=sum(pk)+sum(ql)	i<=k<=j,i-1<=l<=j
 
 e[i,j] means the expected search cost of key[i] to key[j]
+
+w[i,j] comes when a subproblem yields, and the depth of each keys in the subproblem will be +1, so add the sum of all the relative probabilities with the left child's e[i,r-1] and the right child's e[r+1,j], we can get the e[i,j].