Dynamic Programming for partation and backtracing

CSE 3318 Lab Assignment 3

Due March 27

Goal:

1. Understanding of dynamic programming.

2. Understanding of subset sums.

Requirements:

1. Design, code, and test a C program that uses dynamic programming to partition (if possible) a

sequence of n positive integers into three subsequences such that the sum of each subsequence is the

same. For example, if the input were (10, 20, 30, 40, 40, 50, 80), with a total of m = 270, the three

m/3 = 90 subsequences could be (10, 80), (20, 30, 40), and (40, 50). If the input were (20, 20, 30, 50),

then no solution is possible even though the values yield a sum (m = 120) divisible by 3 (m/3 = 40).

The input should be read from standard input (which will be one of 1. keyboard typing, 2. a shell

redirect (<) from a file, or 3. cut-and-paste. Do NOT prompt for a file name!). The first line of the

input is n, the length of the sequence. Each of the remaining lines will include one sequence value.

Your program should echo the input sequence in all cases. The dynamic programming table should be

output when m/3 < 10, but in no other cases. Error messages should be displayed if m is not divisible

by 3 or if the problem instance does not have a solution. When a solution exists, it should be

displayed with each subsequence in a separate column:

i 0 1 2

1 10

2 20

3 30

4 40

5 40

6 50

7 80

2. Submit your C program on Canvas by 3:45 p.m. on Wednesday, March 27. One of the comment lines

should include the compilation command used on OMEGA (5 point penalty for omitting this).

Another comment should indicate the asymptotic worst-case time in terms of m and n.

Getting Started:

1. If you wanted two sequences summing to m/2, then the backtrace part of subsetSum.c could easily

be modified. By finding one subsequence that sums to m/2, the remaining elements would be another

subsequence that sums to m/2. Similarly, your program should use dynamic programming to find two

subsequences that each sum to m/3 and then take the leftover values as the third subsequence. Thus,

this is a two-dimensional DP situation, not one-dimensional like ordinary subset sums in Notes 7.F.

2. Dynamic programming is the only acceptable method for doing this lab.