Scribd
Upload
3 views14 pages

Dynamic Programming

Dynamic programming paradigm matrix chain multiplication

Uploaded by

Shrishu Ranjan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
3 views14 pages

Dynamic Programming

Dynamic programming paradigm matrix chain multiplication

Uploaded by

Shrishu Ranjan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 14

DYNAMIC PROGRAMMING

By

Prasanta K. Jana, IEEE Senior Member

Department of Computer Science and Engineering


Indian Institute of Technology (ISM), Dhanbad
E-mail: [email protected]
DYNAMIC PROGRAMMING

 What is Dynamic Programming (DP)?

 What are its application areas?

 Difference Between D&C and DP

 Similar to D&C, DP solves a problem by combining solutions of


subproblems
 Subproblems are independent for D&C so more works as it solves
common subproblems
 Subproblems are not independent rather overlapped in DP. So, DP
solves every subproblem just once and saves its answer in a table,
thereby avoiding recomputation of the answer every time the
subproblem is encountered.
MATRIX CHAIN MULTIPLICATION

Consider < A1 A2 A3 > chain where [A1]10 x 100, [A2]100 x 5, [A3]5 x 50.
 ((A1 A2) A3) requires 10 x 100 x 5 + 10 x 5 x 50 = 7500 scalar
multiplications.
 (A1(A2 A3)) requires 100 x 5 x 50 + 10 x 100 x 50 = 75000 scalar
multiplications.
So ((A1 A2) A3) is 10 times faster than (A1(A2 A3)).

Problem Statement: Given a chain < A1, A2, A2, …, An > of n matrices,
where Ai has dimension pi-1pi, for i = 1…, n, fully parenthesize the
product in a way that minimizes the number of scalar multiplications.

Brute force approach:

P(n)  Number of Alternative parenthesizations. Then,

1 , 𝑖𝑓 𝑛 = 1
P(n) = { 𝑛
∑𝑘=0 𝑃(𝑘)𝑃(𝑛 − 𝑘) , 𝑖𝑓 𝑛 >= 2

P(n) = C(n-1) is the solution of this recurrence equation where,


1
C(n) = (2𝑛
𝑛
)
𝑛+1

And C(n) = Ω (4n / n3/2)


Let Ai…j denote Ai A2 …Aj

Let m[i, j] be the minimum number of scalar multiplications needed to


compute the product Ai…j.

Time Complexity: O(n3)


An Illustration:
To be initially called with Print_Optimal-Parents(S, 1, n)

Hence the solution is ((A1(A2A3))((A4A5)A6))


Optimal Triangulation:
Exercise 1:
Consider the following matrix-chain multiplication: A17×6 A26×3 A33×10
A410×5 A55×8. Apply Dynamic programming (DP) to generate the DP
tables and compute the optimal sequence of parenthesization. Show each
step of the computation.
Solution:
ASSEMBLY LINE SCHEDULING PROBLEM
Background:
An automobile company produces a car in a factory that has two assembly lines,
shown in the following Figure.
An automobile chassis enters each assembly line, has parts added to it at a number
of stations, and a finished auto exits at the end of the line.

An Instance:
Problem Statement:

Si, j : jth station on line i ( where i is 1 or 2)


ti, j : time to transfer a chassis form line i gone through Si, j
ai, j : assembly time required at station Si, j
ei : time for the chassis to enter assembly line i
xi : an exit time for the completed auto to exit assembly line i
 S1, j performs same function as S2, j
 Time required at each station varies even between stations at the same
position on different lines
The problem is to determine which stations from line 1 and which stations from
line 2 so that total assembly time is minimized.
A recursive solution:
• Let fi [ j] denote the fastest possible time to get a chassis from the starting
point through station Si, j.
• Let f ∗ be the fastest time to get a chassis all the way through the factory.
• Then the chassis has to get all the way through station n on either line 1 or
line 2 and then to the factory exit.
• So, we have: f ∗ = min(f1[n] + x1, f2[n] + x2)
For j = 1
f1[1] = e1 + a1,1 And f2[1] = e2 + a2,1
For j = 2
f1[ 2] = f1[1] + a1,2 or f1[ 2] = f2[1] + t2, 1 + a1,2
And
f2[ 2] = f2[1] + a2,2 or f2[ 2] = f1[1] + t1, 1 + a2, 2

Thus, the fastest way through station S1, j is either

 The fastest way through station S1, j−1 and then directly through station S1, j,
OR
 The fastest way through station S2, j−1, a transfer from line 2 to line 1, and
then through station S1, j.

Using symmetric reasoning, the fastest way through station S2, j is either

 The fastest way through station S2, j−1 and then directly through station S2, j,
OR
 The fastest way through station S1, j−1, a transfer from line 1 to line 2, and
then through station S2, j.

Thus in General:
For j = 2, 3, . . . , n, we have,

f1[ j ] = f1[ j − 1] + a1, j or f1[ j ] = f2[ j − 1] + t2, j−1 + a1, j


And
f2[ j ] = f2[ j − 1] + a2, j or f2[ j ] = f1[ j − 1] + t1, j−1 + a2, j

Therefore, Recursive Equations:


Let li [ j] be the line number, 1 or 2, whose station j − 1 is used in a fastest way
through station Si, j.

Time Complexity: O(n)


An Illustration:

Procedure to print stations in decreasing order of station number

You might also like