One of my colleagues made code in .NET using the LU class.

LU lu = new LU(mtx_l);
if (lu.IsSingular) return false;
Matrix res_u = lu.Solve(mtx_u);

-cpede

http://www.bluebit.gr/matrix/version_31/LU.htm

It should not be difficult to convert the above code in C++.

Trifon Triantafillidis

Lead Developer

-cpede

".NET Matrix Library" (NML) is intended to be used with .NET languages such as VB.NET C# and managed C++ .

"Matrix ActiveX Component" (MaXC) is targeted to VB6 but it can be used by all languages supporting COM such as native C++.

I assume you want to use MaXC with C++. If yes it may help you the C++ project that is found in under the "Samples" directory.

I have modify the example in order to perform LU decomposition and here is the code:

#include "stdafx.h"
#include "iostream.h"
#import "bluebitmatrix31.dll" no_namespace
HRESULT MatrixLU()
{
 IMatrix *A;
 long N  = 6;
 HRESULT hr;
 hr = ::CoCreateInstance(__uuidof(Matrix), NULL, CLSCTX_INPROC, __uuidof(IMatrix), (void**)&A);
 if(FAILED(hr))
  return hr; 
 A->Size(N, N);
 A->FillRandom(0, 10, 0);

 cout << "===============================" << endl;
 cout << "Matrix A =" << endl;
 cout << A->GetString(3, _bstr_t("")) << endl;
 cout << "===============================" << endl;

 //Declare 2 matrix variables that will hold
 //the L and U parts of LU decomposition
 IMatrix *L, *U;
 hr = ::CoCreateInstance(__uuidof(Matrix), NULL, CLSCTX_INPROC, __uuidof(IMatrix), (void**)&L);
 if(FAILED(hr))
  return hr;
 hr = ::CoCreateInstance(__uuidof(Matrix), NULL, CLSCTX_INPROC, __uuidof(IMatrix), (void**)&U);
 if(FAILED(hr))
  return hr;
 VARIANT * perIndex ;
 perIndex = NULL;
 long perSign = 0;
 
 A->LU( &L, &U, perIndex, &perSign );
 
 cout << endl;
 cout << "===============================" << endl;
 cout << "Matrix L =" << endl;
 cout << L->GetString(3, _bstr_t("")) << endl;
 cout << "===============================" << endl;

 cout << endl;
 cout << "===============================" << endl;
 cout << "Matrix U =" << endl;
 cout << U->GetString(3, _bstr_t("")) << endl;
 cout << "===============================" << endl;

 if(A)
  A->Release();
 if(L)
  L->Release();
 if(U)
  U->Release();
 return hr;
}
int main(int argc, char* argv[])
{
 ::CoInitialize(NULL);
 MatrixLU();
 //cout << "Please press return to exit";
 //cin.get();
 CoUninitialize();    // clearup
 return 0;
}

Having the matrix to LU I can use:

http://en.wikipedia.org/wiki/LU_decomposition#Applications

It is true that our another product .NET Matrix Library supports this directly with its LU.Solve method.

 

Trifon Triantafillidis

Lead Developer