when I make sparse matrix in C-MEX , eigs fuctions call doesn't work

when I make sparse matrix in C-MEX , eigs fuctions call doesn't work

(1) I made a sparse matrix M in C-MEX

(2) I called eigs function with the sparse matrix M

     [V,T] = eigs(M,1,'sm');

(3) it doesn't work with following error. >> eigs(M,1,'sm');

??? Error using ==> eigs>checkInputs/LUfactorAminusSigmaB at 1136 (A-sigma*I)is singular. The shift is an eigenvalue. Try to use some other shift please.

Error in ==> eigs>checkInputs at 899 [L,U,pp,qq,dgAsB] = LUfactorAminusSigmaB;

Error in ==> eigs at 94 [A,Amatrix,isrealprob,issymA,n,B,classAB,k,eigs_sigma,whch, ...

If I make a sparse matrix, then eigs function works well. what is problem with my sparse matrix made in C-MEX?

my C-MEX sourse is...... ---------------------------------------------------------- percent_sparse=0.10; nzmax = (mwSize)ceil(3000*3000*percent_sparse);

plhs[0] = mxCreateSparse(3000, 3000, nzmax ,mxREAL);

    T  = mxGetPr(plhs[0]);
    irs = mxGetIr(plhs[0]);
    jcs = mxGetJc(plhs[0]);

. . . . jcs[x-1] = k;

         for(y=0 ; y< (*num_states) ; y++)  
         {
             Tp=1;

             for (i=0 ; i< (*num_nodes) ; i++)
             {
                 if(ADotS[i] ==0)
                 {
                     if(next_state[i]==current_state[i])                           Ts = 1-(*c);
                     else
                         Ts = (*c); 
                 }
                 else
                 {
                     if (next_state[i]==1)                           Ts = 0.5 + (0.5 * tanh(5*ADotS[i]));
                     else        
                         Ts = 0.5 - (0.5 * tanh(5*ADotS[i]));      
                 }
                 Tp=Tp*Ts;
             }

             if( Tp > 0)
             {
                 if( k>=nzmax )
                 {
                     mwSize oldnzmax = nzmax;
                     percent_sparse += 0.01;                     
                 nzmax = (mwSize)ceil((3000)*(3000)*percent_sparse);

                     if(oldnzmax==nzmax) /* make sure nzmax increases atleast by 1 */
                         nzmax++;

                     mxSetNzmax(plhs[0],nzmax);
                     mxSetPr(plhs[0],(double*)mxRealloc(T,nzmax*sizeof(double)));
                     mxSetIr(plhs[0],(mwIndex*)mxRealloc(irs,nzmax*sizeof(mwIndex)));
                     T=mxGetPr(plhs[0]);
                     irs=mxGetIr(plhs[0]);
                 }

                 if( x-1 == y )
                     T[k] = -1;
                 else
                     T[k] = Tp;   
                 irs[k]=y;
                 k++;      
             }
         }

-----------------------------------------------------------------

plhs[0] = mxCreateSparse(3000, 3000, nzmax ,mxREAL);

I think that when i make sparse matrix in C-MEX, it needs nzmax. it is the number of non-zero value. It would be incorrect, because I don't know how many non-zeros there exists. so the sparse matrix is a little bit strange, eigs fuctions doesn't understand the sparse matrix...

please help me!!

[Kaustubha Govind]

What is the error that you get? Also, why do you run the line:

plhs[0] = mxCreateSparse(3000, 3000, nzmax ,mxREAL);

at the end of your code? This overwrites all the values you assign to the sparse matrix. Is this a typo? Please format your code for better readability.

Answer 1

[James Tursa]

See this FEX submission by Tim Davis to help you diagnose sparse matrix formation problems in a mex routine:

http://www.mathworks.com/matlabcentral/fileexchange/20186-spok-checks-if-a-matlab-sparse-matrix-is-ok