MKL with C++

          C++ program with MKL 



i have  written a  c++ example for Matrix eigenvalue/eigenvector problem.
In the Documentation MKL all the examples are given in FORTRAN/C .
To find all the eigenvalue and optionally all the eigenvector of real symmetric matrix i have used the lapack function dsyev from MKL .
In This example i have created a matrix which correspond to 3D Tight Binding Hamiltonian. This is quit common in Condensed Matter Physics community.

#include
#include
#include "mkl.h"
#define dm 12
#define N dm*dm*dm
#define LDA N

int main()
{
    int n = N, lda = LDA, info, lwork;
    double wkopt;
    double w[N];
    double *a =new double [N*N];


    int i,j,k,l,m,o,pd[dm];
    for(i=0;i        pd[i]=(i+1)%dm;
    for(i=0;i    for(j=0;j        a[i*N + j] =0.00;


//******Matrix :: 3D Tight Binding Hamiltonian ***************
       for(i=0;i       for(j=0;j       for(k=0;k        {
                l=i*dm*dm + j*dm + k;
                m=pd[i]*dm*dm + j*dm + k;
                a[l*N + m] =-1.0;
                a[m*N + l] =-1.0;

                m=i*dm*dm + pd[j]*dm + k;
                a[l*N + m] =-1.0;
                a[m*N + l] =-1.0;

                m=i*dm*dm + j*dm + pd[k];
                a[l*N + m] =-1.0;
                a[m*N + l] =-1.0;
        }

//*******************Diagonalization*************************************
    lwork = -1;
    dsyev( "N", "U", &n, a, &lda, w, &wkopt, &lwork, &info );
    lwork = (int)wkopt;
    double *work= new double[lwork];
    dsyev( "N", "U", &n, a, &lda, w, work, &lwork, &info );
    delete[] a;
    delete[] work;
//*****************Write eigen values to a file :-------------------------
    std::ofstream dfile("eigen.dat");
    for(i=0;i    dfile << w[i] << std::endl;
    dfile.close();
//************************************************************************
    return 0;
}

To compile this program you need to use this links .
 icpc filename.cpp  -O2 -C -Wl,--start-group /export/apps/ics_2012.0.032/ics/2012.0.032/mkl/lib/intel64/libmkl_intel_lp64.a  /export/apps/ics_2012.0.032/ics/2012.0.032/mkl/lib/intel64/libmkl_core.a -Wl,--end-group -lpthread  -openmp -lmkl_intel_ilp64 -lmkl_intel_thread -lmkl_core -liomp5   -mcmodel large -shared-intel -traceback

I assume that you have save this program in  filename.cpp file . You should change the above linking path depending upon where you have installed the Intel libraries.


For Newer version just do this : 

icpc filename.cpp -mkl