How to extract a 2x2 submatrix from a bigger matrix

Question

Hello,

I am a very basic user and do not know much about commands used in C, so please bear with me...I cant use very complicated codes. I have some knowledge in the stdio.h and ctype.h library, but thats about it. I have a matrix in a txt file and I want to load the matrix based on my input of number of rows and columns

For example, I have a 5 by 5 matrix in the file. I want to extract a specific 2 by 2 submatrix, how can I do that ?

I created a nested loop using :

FILE *sample
sample=fopen("randomfile.txt","r"); 
for(i=0;i<rows;i++){
  for(j=0;j<cols;j++){
     fscanf(sample,"%f",&matrix[i][j]);
   }
 fscanf(sample,"\n",&matrix[i][j]);
}
fclose(sample);

Sadly the code does not work .. If I have this matrix :

5.00 4.00 5.00 6.00 
5.00 4.00 3.00 25.00 
5.00 3.00 4.00 23.00 
5.00 2.00 352.00 6.00

And inputting 3 for row and 3 for column, I get :

5.00 4.00 5.00
6.00 5.00 4.00
3.00 25.00 5.00

Not only this isnt a 2 by 2 submatrix, but even if I wanted the first 3 rows and first 3 columns, its not printing it correctly....

I need to start at row 3 and col 3, then take the 2 by 2 submatrix !

I should have ended up with :

4.00 23.00 
352.00 6.00

I heard that I can use fgets and sscanf to accomplish this. Here is my trial code :

fgets(garbage,1,fin);
sscanf(garbage,"\n");

But this doesnt work either :(

What am I doing wrong ?

Please help. Thanks !

Answer 1

OK, so you want to read a submatrix of size n x m, starting at positions x, y in the big matrix of size p x q. You need two things:

1. (verify that x + n <= p and y + m <= q)
2. skip to the first element of the matrix you want to read. This requires first skipping the first y - 1 rows
3. skip x - 1 elements from the next row, then read n elements into your submatrix. Repeat m times.

Your current implementation starts reading from the very first element of the matrix, then reads elements contiguously into the submatrix. An updated version:

FILE *sample = fopen("randomfile.txt", "r");
// skip the first y-1 rows
for (i = 0; i < y - 1; i++) {
  fscanf(sample, "%*[^\n]\n", &matrix[i][j]);
}
for (i = 0; i < m; i++) {
  // skip the first x-1 numbers
  for (j = 0; j < x - 1; j++) {
     fscanf(sample, "%*f");
  }
  // read n numbers
  for (j = 0; j < n; j++) {
     fscanf(sample, "%f", &matrix[i][j]);
  }
  if (x + n < p) {
    // consume the rest of the line
    fscanf(sample, "%*[^\n]\n");
  }
}
fclose(sample);

Update: to read the submatrix from an array instead is even simpler, just requires a bit more calculation. The gist is, a matrix of size p x q can be stored in a contiguous array of size p x q such that matrix[i,j] can be read from array[i*(j-1)+j] (approximately - there may be off-by-one errors and I am never sure which is the column and which is the row, but hopefully you get the idea :-)

So the code would be something like

for (i = 0; i < m; i++) {
  for (j = 0; j < n; j++) {
     submatrix[i][j] = array[(y + i) * p + x + j];
  }
}

Answer 2

Let's take this in stages. First a couple of minor fixes to your code:

for(i=0;i<rows;i++){
  for(j=0;j<cols;j++){
    float dummy;  /* this will make thing easier later */
    fscanf(sample,"%f",&dummy);
    matrix[i][j] = dummy;
  }
/* fscanf(sample,"\n",&matrix[i][j]); this isn't even legal */
}

Now we define what we want:

int startrow = 2; /* The starting index. Remember we index 0,1,2,3 */
int startcol = 2;
int resultrows = 2; /* How many rows we want in our answer */
int resultcols = 2;
float result[resultrows][resultcols];

Now we ignore what we don't want:

for(i=0;i<rows;i++){
  for(j=0;j<cols;j++){
    float dummy;
    fscanf(sample,"%f",&dummy);
    if(i >= startrow && i < startrow + resultrows &&
       j >= startcol && j < startcol + resultcols){
      matrix[i][j] = dummy;
    }
  }
}

Notice that now only the values we want are copied into matrix, the rest of matrix is uninitialized gibberish. Now write it into result instead:

for(i=0;i<rows;i++){
  for(j=0;j<cols;j++){
    float dummy;
    fscanf(sample,"%f",&dummy);
    if(i >= startrow && i < startrow + resultrows &&
       j >= startcol && j < startcol + resultcols){
      result[i-startrow][j-startcol] = dummy;
    }
  }
}

EDIT:
If you want to copy a submatrix from a larger matrix already in memory, the inner loop should be

for(j=0;j<cols;j++){
  if(i >= startrow && i < startrow + resultrows &&
     j >= startcol && j < startcol + resultcols){
      result[i-startrow][j-startcol] = matrix[i][j];
  }
}