Scribd
Upload
33 views

Alternate Formulation of The Master Method

This document presents an alternate formulation of the Master Method for solving recurrences of the form T(n) = Θ(1) for n ≤ c, T(n) = aT(n/b) + f(n) for n > c, where f(n) = Θ(n^l(log n)^k). There are three cases: 1) l < log_b a, T(n) = Θ(n^log_b a); 2) l = log_b a, T(n) = Θ(f(n)log n); 3) l > log_b a, T(n) = Θ(f(n)).

Uploaded by

Nishant Mishra
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
33 views

Alternate Formulation of The Master Method

This document presents an alternate formulation of the Master Method for solving recurrences of the form T(n) = Θ(1) for n ≤ c, T(n) = aT(n/b) + f(n) for n > c, where f(n) = Θ(n^l(log n)^k). There are three cases: 1) l < log_b a, T(n) = Θ(n^log_b a); 2) l = log_b a, T(n) = Θ(f(n)log n); 3) l > log_b a, T(n) = Θ(f(n)).

Uploaded by

Nishant Mishra
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

CSE 241 Algorithms and Data Structures Spring Semester, 2006

Master Method: An Alternate Formulation


January 26, 2006 Handout 1

Alternate formulation of the Master Method


The formulation presented here will solve recurrences of the form:

T (n) = Θ(1) for all n ≤ c


T (n) = aT (n/b) + f (n) for all n > c

where f (n) = Θ(n` (log n)k ) for constants, c ≥ 0, a ≥ 1, b > 1, ` ≥ 0, and k ≥ 0.

Case 1: For ` < logb a, T (n) = Θ(nlogb a )

Case 2: For ` = logb a, T (n) = Θ(f (n) log n) = Θ(nlogb a (log n)k+1 )

Case 3: For ` > logb a, T (n) = Θ(f (n)) = Θ(n` (log n)k )

To get some practice with this, solve the problems in the text book on pages 74 and 75.
Check that you get the same answers as given in the text.

Some Things to Remember


For recurrences of the form T (n) = aT (n/b) + f (n), when the base case occurs for inputs of
size 1, there are exactly alogb n = nlogb a applications of the base case of the recurrence. So
Θ(nlogb a ) time is spent at the base of the recursive calls. In Case 1, the time spent here is
as much (asymptotically) as all the time spent dividing and combining.
Just as we saw when analyzing the time complexity of the divide-and-conquer algorithm
for finding the closest pair of points, in Case 2, there are logb n levels of recursion, across
each of which f (n) time is spent splitting and combining.
Finally, in Case 3, the f (n) time spent splitting and combining for the top-level call is
as much (asymptotically) as all the rest of the computation.

Reason for Presenting This Formulation


I have selected this formulation to present since students generally find it easier to use than
the one in the text book. Also, the version in the text book only considers when k = 0. The
formulation in the textbook is more general in that it can applies for recurrences of the form
T (n) = aT (n/b) + f (n) where f (n) is not of the form Θ(n` (log n)k ). However, it is rare to
find any divide-and-conquer algorithms where the cost to split and then combine is not of
the form Θ(n` (log n)k ).

You might also like