1.We are given an array A containing n numbers, and an integer k that divides n. We want to find k numbers x1,x2, …,xk of A such that xi is the ( n /k i)-th smallest element of A. Develop a divide and conquer algorithm that finds the xi ’s that runs in time O(n lg k). Prove its correctness and running time.

2.Recall that the Fibonacci numbers are defined by the following recurrence: F0 = 0, F1 = 1 and Fn = Fn-1 + Fn-2 for n > 2. 1. Using the above recurrence, give a simple algorithm to compute Fn, given n. The complexity of the algorithm in terms of the number of arithmetic operations should be O(n 2 ). Show that the algorithm indeed has this complexity. Recall that an arithmetic operation is the cost of adding or multiplying two single digit numbers. (Hint: How many digits are there in Fn? Use the fact that Fn is T(( 1+v 5 2 ) n ). 2. Our goal is to use the following matrix identity to compute the nth Fibonacci number more efficiently. F1 F2 ! = 0 1 1 1! · F0 F1 ! (5.1) (a) Show that Fn Fn+1! = 0 1 1 1!n · F0 F1 ! (5.2) (b) Show that O(log n) matrix multiplications are enough to compute Fn. (c) Using the fast algorithm for multiplying two integers (Karatsuba’s algorithm) show that Fn can be computing in O(n log2 3 log n) arithmetic operations. (d) Can you improve your analysis of

(c) and show a cost of O(n log2 3 ) arithmetic operations ?