problem with a sparse Matrix function CSR (Compressed Sparse Row)

This is my first time using Matlab , I tried to write a function that gives you CSR (Compressed Sparse Row)

the result are not correct.

I want the result of IA and JA as a vector and NNZ in my matrix is 14

this is my test Marix M = [0, 0, 0, 0, 1 ; 5, 8, 0, 0, 0 ; 0, 0, 3, 0, 0 ; 0, 6, 0, 0, 1]

the result should be

M =

0 0 0 0 1

5 8 0 0 0

0 0 3 0 0

0 6 0 0 1

A = [ 1 5 8 3 6 1 ]

IA = [ 1 2 4 5 7 ]

JA = [ 5 1 2 3 2 5 ]

this the function I wrote or to be precise the code that i took it from python and i tried to write it in matlab

function sparse_CSR2(M)
m = length(M);
i = 1 ;
val = 0  ;
if isempty(M)
    n = 0 ; 
else
    % n = height(M);
    [n,~] = size(M);
end
NNZ = 0;
%val = 0;
while i < m
    j = 1;
    
    while j < n
        if M(i,j)~=0
            val=M(i,j);
            JA = [j] ;
            NNZ = NNZ + 1 ;
        end
        j=j+1;
    end
    IA = [NNZ];
    i=i+1;
    disp(M);
    disp(val);
    disp(IA);
    disp(JA);
end

[soufiane]

this is my last code

function sparse_CSR3(M)
m = length(M);
i = 1 ;
%val = 0  ;
k=1;
if isempty(M)
    n = 0 ; 
else
    % n = height(M);
    [n,~] = size(M);
    [row,col,value]=find(M);
end
NNZ = 0;
JA = zeros(m,1);
IA = zeros(n,1);
val = zeros(n,1);
%val = 0;
while i < n+1
    j = 1;
    
    while j < m+1
        if M(i,j)~=0
            val(k)=M(i,j);
            JA(k) = j ;
            NNZ = NNZ + 1 ;
            k=k+1;
        end
        j=j+1;
    end
    IA(i) = NNZ;
    i=i+1;
    
    
end

disp(M);
disp(val.');
disp(IA.');
disp(JA.');

end

my result is :

0 0 0 0 1

5 8 0 0 0

0 0 3 0 0

0 6 0 0 1

1 5 8 3 6 1

1 3 4 6

5 1 2 3 2 5

the result should be

0 0 0 0 1

5 8 0 0 0

0 0 3 0 0

0 6 0 0 1

1 5 8 3 6 1

1 2 4 5 7

5 1 2 3 2 5

Answer 1

[Prathamesh]

Hi @soufiane,

I understand that you have a function that gives you a Compressed Sparse Row. And you want the result of “IA” and “JA” as a vector and “NNZ” in a matrix.

You can follow below steps to get the CSR(Compressed Sparse Row)

1. Count all “NNZ_total” (non-zero elements) in a separate loop.
2. Pre-allocate “A” and “JA” to size “1 x NNZ_total”.
3. Pre-allocate “IA” to size “1 x (num_rows + 1)”.
4. Get dimensions: “ [num_rows, num_cols] = size(M) ” .
5. Set IA(1) = 1.
6. Inside the row loop (for r = 1:num_rows):

• set “ IA(r) = k “ before the inner column loop for row r.
• Increment k only when a non-zero element is found and stored.

7. After all loops, set “ IA(num_rows + 1) = NNZ_total + 1 “.