Can anyone help in solving the recurrence relationship of a divide and conquer algorithm with the following equation? I am pretty sure you can't use master theorem here because it is not in the form T(n/b) but may be forgetting a simple math rule here. Please help.
T(n)=T(√n)+logn.
Notice that for some k>0
we have
T(n) = log n + log n^{1/2} + log n^{1/4} + ... + log n^{1/2^k} =
= log n + (1/2)*log n + (1/4)*log n + ... + (1/k) * log n
= (1 + 1/2 + 1/4 + ... + 1/2*k) log n
= (1 + 2^{-1} + 2^{-2} + ... + 2^{-k})log n
<= 2 log n
from which it follows that T(n) = O(log n)
. The bound <= 2 log n
follows because 1+1/2+1/4+1/8+1/16+...=2
in the limit.