CS110: Matrix operations in C

Lecture 10
V. Kamakoti
29th January 2008
 Arrays
Matrix is
A two dimensional array
Array of one dimensional arrays
Stored in memory
Row-wise (Row Major order)
Column-wise (Column Major order)
 Storage of matrix
A - 2X3 matrix has elements
A[0][0],A[0][1],A[0][2];
A[1][0],A[1][1],A[1][2];
A[2][0],A[2][1],A[2][2];
Assume it is an integer matrix and
A[0][0] is stored at 1000
 Row-major order
Store the first row, then, the second,
then, the third and so on
A - 2X3 matrix has elements
A[0][0],A[0][1],A[0][2]; stored at 1000-1011
A[1][0],A[1][1],A[1][2]; stored 1012 - 1023
A[2][0],A[2][1],A[2][2]; stored 1024 - 1035
 Column-major order
Store the first column, then, the second
column, then, the third and so on..
A - 2X3 matrix has elements
A[0][0],A[1][0],A[2][0]; stored at 1000-1011
A[0][1],A[1][1],A[2][1]; stored 1012 - 1023
A[0][2],A[1][2],A[2][2]; stored 1024 - 1035
The C compiler assumes Row major order.
 Lab - 3
Write a program that reads in the
entries of a 3 by 3 matrix, and prints it
out in the form of a matrix. The entries
could be floating point entries too.
 Solution
#include<stdio.h>
main() {
int i,j, a[3][3];
for (i = 0; i <= 2; i++) {
for (j = 0; j <=2; j++) {
printf(Enter a[%d][%d] \n,i,j);
scanf(%d,&a[i][j]);
}
}
}
 What Happens
for (i = 0; i <= 2; i++) {
for (j = 0; j <=2; j++)
{
printf(Enter a[%d][%d] \n,i,j);
scanf(%d,&a[i][j]);
}
}
For every i, 0 <= i <= 2, j repeats thrice taking values
0,1 and 2. Therefore when i = 0 and j = 0, a[0][0]
shall be read in, when i = 0 and j = 1, a[0][1] shall be
read in and so on .
 To Print
for (i = 0; i <= 2; i++) {
for (j = 0; j <=2; j++)
{
printf(%d ,a[i][j]);
}
printf(\n);
}
Print all elements of a given row on the same
line with a space in between. After one row is
printed go to the next line.
 You have to
Repeat above for Floating point numbers
The effort would be to
Format the output properly
Let p be a floating point number
printf(%16f,..) prints a floating point with 16
spaces - if less then blanks are introduced.
Use the tab escape sequence \t to align.
This would be true for integers too.
 Next in Lab - 3
Write a program that reads in orders of two
matrices and decides whether two such
matrices can be multiplied. Print out the
decision.
main() {
int row1,col1,row2,col2;
printf(Enter number of rows and columns in
Matrix 1 \n);
scanf(%d %d,&row1,&col1);
//Similarly for Matrix 2
if (col1 != row2) { printf(Not possible\n); exit(0);
} // To use exit() we need to include <stdlib.h>
 Next in Lab - 3
Write a program that reads in two
arbitrary matrices and multiplies them.
Your output should be the two matrices
and the resulting product matrix.
 Matrix Multiplication : Outline
main() {
1. declare all variables required
2. read in Matrix A
3. read in Matrix B
4. check if A and B are compatible to be multiplied
5. initialize Matrix C to have zeroes to begin with
6. multiply A and B to give C
7. print Matrix C
}
Of course, all the steps above have to
follow C languages syntax rules
 Using Matrix Operations :
Read a Matrix
main(){
int a[11][11], b[11][11], c[11][11]; / *max size 10 by 10 */
int i,j,k;
int aRows, aCols, bRows, bCols, cRows, cCols;

scanf("%d%d", &aRows, &aCols);

for(int i = 1; i <= aRows; i++)
for(int j = 1; j <= aCols; j++)
scanf("%d", &a[i][j]);

/*continued on next slide */

 Read the other Matrix; Initialize the product
scanf("%d%d", &bRows, &bCols);
for(i = 1; i <= bRows; i++)
for(j = 1; j <= bCols; j++)
scanf("%d", &b[i][j]);

/* initialize entries in Matrix c to 0 */

for(i = 1; i <= aRows; i++)
for(j = 1; j <= bCols; j++)
c[i][j] = 0; Remember
bRows must equal aCols
/*continued on next slide */ c is a aRows x bCols matrix
 Matrix multiplication

bRows
Multiply two numbers

aRows

aCols
Sum of N products
 Multiply Matrices and Print the
Result
/* multiply both the matrices and store in matrix c */
for(i =1; i <= aRows; i++)
for(j = 1; j <= bCols; j++)
for(k = 1; k <= aCols; k++)
c[i][j] += a[i][k]*b[k][j];

/* print matrix c */
for(i = 1; i <= aRows; i++){
for(j = 1; j <= bCols; j++) /* print a row */
printf("%d ", c[i][j]); /* notice missing \n */
printf("\n"); /* print a newline at the end a row */
}
} /* End of main program */
 Points to Ponder
Some repetition in the program
Reading in matrix A seems to be similar to reading
in matrix B
Except for changes in the number of rows and columns
You will learn how to write functions in C later
Functions are written to avoid repeated actions
Example C program would look like
readMat(A, aRows, aCols);
readMat(B, bRows, bCols);
Function readMat( ) must perform the operations
desired
 Celebrity Problem
(More efficient Solution)

There are n persons in a room. A person is called a celebrity

iff everyone knows him but he does not know anyone. So by
definition if at all there is a celebrity there can be only one.
The problem is to find if there is a celebrity among the people
in the room and if there is one identify him, by asking
questions. The only kind of question allowed is Does A know
B?.
 Celebrity Problem
(Naivest Solution)
'Naivest' Solution:
For each pair of persons A and B,
ask the questions, does A know B and does B know A. Build
the nxn matrix {aij}. is 1 if the ith person knows the jth person
and it is 0 otherwise. Now if there is a column i, which
contains all 1s, and if row i contains all 0s except the
diagonal, then the person i is a celebrity. If there is no such i,
then there is no celebrity.

Number of questions asked = n*(n-1)

 Celebrity Problem
(Better Solution)

Choose two persons A and B. Ask if

A knows B. If answer is yes, then A cannot be the celebrity. If
answer is no, then B cannot be the celebrity. So one person
gets eliminated. Now if at all there is a celebrity he will be in
the remaining n-1 persons. Keep eliminating in this fashion.
After n-1 questions, only one person X, will be left. For each
person Y X, ask if Y knows X and X knows Y. Hence we
can find if X is actually a celebrity.

Number of questions asked = n-1 + 2n-2 = 3n-3

 Celebrity Problem
(Best Solution)

Simplifying assumption: n = 2^k

Group the persons into pairs. So we will have n/2
pairs. In each pair (A, B) eliminate one person by asking if A
knows B. Now we will be left with n/2 persons. Repeat the
process till we are left with only one person.

Number of times this process in repeated is

obviously k, since each time the number reduces by half.

(cont.)
 Celebrity Problem
(Best Solution)
We can keep track of the tree thus obtained. Let us take a
case having 8 persons. The tree looks likes something below.

1 2 3 4 5 6 7 8

1 4 6 7

4 6 So totally 7 Qs asked.
This is a binary tree with a root,
leaves and children (at most 2).
4
Learn more about this later
 Generalizing above
Given N people
N/2 Qs in level 1
N/4 Qs in level 2
1 Question in level k, where, N = 2k
So total is N(1/2 + 1/4 + 1/2k)
N (1/2) ( 1 - 1/2k)/(1/2)
N - N/2k = N - 1
 Celebrity Problem
(Best Solution)
Clearly, if at all there is a celebrity, it can be only 4.
Also note that out of the 2n-2(= 6) questions asked to confirm
whether 4 is actually a celebrity, we have already asked
log2n (= 3) questions (Notice the yellow circles). So once we
maintain this tree, the number of additional questions that we
need to ask is 2n 2 log2n.
Therefore the total number of questions asked in this case is
n-1 + 2n 2 log2n = 3n 3 log2n.

There exists no algorithm which can do better than this, the

proof of which is beyond the scope of this lecture. Extending
it for n not a power of 2 is also interesting.
Thank You