Design and Analysis of Algorithms
Assignment 3
Muhammad Anas Qureshi
BCS213186
Matrix Chain Multiplication Problem (Dynamic Programming)
Understanding:
Goal: Find the most efficient way to multiply a sequence of matrices, minimizing scalar
multiplications.
Overlapping Subproblems: Optimal solutions for larger sub chains often reuse those for smaller ones.
Optimal Substructure: The optimal solution for a larger problem can be built from optimal solutions
of smaller ones.
Solution:
Recursive Structure:
a. m[i][j] represents the minimum cost to multiply matrices from i to j.
b. m[i][j] = min(m[i][k] + m[k+1][j] + p[i-1]*p[k]*p[j]), for k from i to j-1
Memorization or Tabulation:
a. Store results of subproblems to avoid redundant calculations.
b. Fill a table m[i][j] in a bottom-up manner.
Steps:
Initialize: Create tables m (n x n) for minimum costs and p (n+1) for dimensions.
Base Cases: m[i][i] = 0 (single matrix costs 0).
Fill Table: Iterate through m, filling entries using the recursive formula.
Find Minimum Cost: The minimum cost is in m[1][n].
Explanation:
The recursive formula explores different splits of the matrix chain to find the most efficient one.
Memorization/tabulation avoids recalculations.
The bottom-up approach builds solutions for larger subproblems from smaller ones.
Time Complexity: O(n^3)
Space Complexity: O(n^2)
Run on random data:
Find an optimal parenthesizing of a matrix-chain product whose sequence of dimensions is: (5, 10, 3, 12, 5, 50,
6).
We have:
p0 = 5
p1 = 10
p2 = 3
p3 = 12
p4 = 5
p5 = 50
p6 = 6
The corresponding matrices are:
A1 – 5x10 A2 – 10x3 A3 – 3x12 A4 – 12x5 A5 – 5x50 A6 – 50x6
From the algorithm, we have, for all x, m[x,x] = 0. m[1,2] = m[1,1] + m[2,2] + p0 * p1 * p2
m[1,2] = 0 + 150
m[1,2] = 150
m[3,4] = m[3,3] + m[4,4] + p2 * p3 * p4 m[3,4] = 0 + 180
m[3,4] = 180
m[4,5] = m[4,4] + m[5,5] + p3 * p4 * p5 m[4,5] = 0 + 3000
m[4,5] = 3000
m[5,6] = m[5,5] + m[6,6] + p4 * p5 * p6 m[5,6] = 0 + 1500
m[5,6] = 1500
m[1,3] = min of { m[1,1] + m[2,3] + p0 * p1 * p3 = 750 }
{ m[1,2] + m[3,3] + p0 * p2 * p3 = 330 }
m[2,4] = min of { m[2,2] + m[3,4] + p1 * p2 * p4 = 330}
{ m[2,3] + m[4,4] + p1 * p3 * p4 = 960}
m[3,5] = min of { m[3,3] + m[4,5] + p2 * p3 * p5 = 4800}
{ m[3,4] + m[5,5] + p2 * p4 * p5 = 930 }
m[4,6] = min of { m[4,4] + m[5,6] + p3 * p4 * p6 = 1860 }
{ m[4,5] + m[6,6] + p3 * p5 * p6 = 6600 }
m[1,4] = min of { m[1,1] + m[2,4] + p0 * p1 * p4 = 580 }
{ m[1,2] + m[3,4] + p0 * p2 * p4 = 405 }
{ m[1,3] + m[4,4] + p0 * p3 * p4 = 630 }
m[2,5] = min of { m[2,2] + m[3,5] + p1 * p2 * p5 = 2430 }
{ m[2,3] + m[4,5] + p1 * p3 * p5 = 9360 }
{ m[2,4] + m[5,5] + p1 * p4 * p5 = 2830 }
m[3,6] = min of { m[3,3] + m[4,6] + p2 * p3 * p6 = 2076 }
{ m[3,4] + m[5,6] + p2 * p4 * p6 = 1770 }
{ m[3,5] + m[6,6] + p2 * p5 * p6 = 1830 }
m[1,5] ={ m[1,1] + m[2,5] + p0 * p1 * p5 = 4930 }
{ m[1,2] + m[3,5] + p0 * p2 * p5 = 1830 }
{ m[1,3] + m[1,4] + p0 * p3 * p5 = 6330 }
{ m[1,4] + m[1,5] + p0 * p4 * p5 = 1655 }
m[2,6] ={ m[2,2] + m[3,6] + p1 * p2 * p6 = 1950 }
{ m[2,3] + m[4,6] + p1 * p3 * p6 = 2940 }
{ m[2,4] + m[5,6] + p1 * p4 * p6 = 2130 }
{ m[2,5] + m[6,6] + p1 * p5 * p6 = 5430 }
m[1,6] ={ m[1,1] + m[2,6] + p0 * p1 * p6 = 2250 }
{ m[1,2] + m[3,6] + p0 * p2 * p6 = 2010 }
{ m[1,3] + m[4,6] + p0 * p3 * p6 = 2550 }
{ m[1,4] + m[5,6] + p0 * p4 * p6 = 2055 }
{ m[1,5] + m[6,6] + p0 * p5 * p6 = 3155 }
M 1 2 3 4 5 6
6 2010 1950 1770 1840 1500 0
5 1655 2430 930 3000 0
4 405 330 180 0
3 330 360
2 150
1 0 The minimum cost is therefore
2010 and the optimal
And using this, we construct the S table:
parenthesizing is:
((A1 * A2) * (A3 * A4) * (A5 *
S 1 2 3 4 5 A6)).
6 2 2 4 4 5
5 4 2 4 4
4 2 2 3
3 2 2
2 1
