Chapter 5: Probabilistic Analysis

Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

Average-Case Analysis

Chapter 5: Probabilistic Analysis


Average-Case Analysis Finding the Maximum Example Randomizing an Array Example

In practice, many algorithms perform better than their worst-case. The average case is analyzed by: 1. construct a probabilistic model of the input 2. determine the probabilities and running times (or costs) of alternate executions 3. calculate expected running time (or cost) Through randomization, one can often ensure that the probabilisitic model is true.
CS 5633 Analysis of Algorithms Chapter 5: Slide 2

Example 1: Finding the Maximum


Average-Case Analysis . . . . . . . . . . . . . . . . Example 1: Finding the Maximum . . . . . . . . Example 1: Probablistic Model . . . . . . . . . . Example 1: Analysis . . . . . . . . . . . . . . . . . . Example 2: Random Permutation of an Array Example 2: Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 3 4 5 6 7 Maximum(A) max A[1] for i 2 to length(A) do if max < A[i] then max A[i] return max Problem: How many assignments to max? Best-case: 1 (When does this happen?) Worst-case: n (When does this happen?) Average-case: (n + 1)/2 is incorrect
CS 5633 Analysis of Algorithms Chapter 5: Slide 3

Example 1: Probablistic Model


Assume A has n distinct numbers. (What is the eect of duplicates?) Assume each permutation of the numbers is equally likely. (How can randomization guarantee this?) How many permutations are there? What is the probability of the best case? What is the probability of the worst case?
CS 5633 Analysis of Algorithms Chapter 5: Slide 4

Example 1: Analysis
On iteration i, max is assigned a value i A[i] is the maximum of the rst i numbers. Probability that A[i] is the maximum of the rst i numbers = 1/i Probability of assignment = 1/i, cost = 1 Prob. of no assignment = (i 1)/i, cost = 0 On iteration i, the expected cost is: (1/i)(1) + ((i 1)/i)(0) = 1/i Over the initial assignment and n 1 iterations, the expected cost is: n 1 which is between ln n and 1 + ln n i=1 i
CS 5633 Analysis of Algorithms Chapter 5: Slide 5

Example 2: Random Permutation of an Array


Randomize-In-Place(A) n length(A) for i 2 to n do swap A[i] A[Random(1, i)] Random(a, b) returns an integer r, a r b. r is equally likely to be any integer between a and b
CS 5633 Analysis of Algorithms Chapter 5: Slide 6

Example 2: Analysis
Loop invariant: Before iteration i, A[1] through A[i 1] is a random permutation of the rst i 1 values. For any permutation of the rst i values, there is exactly one way to permute the rst i 1 values, and then swap A[i] into the correct position. So after iteration i, the rst i values have been randomly permuted, making the loop invariant true the next iteration.
CS 5633 Analysis of Algorithms Chapter 5: Slide 7

You might also like