Is there an efficient Eigen way of doing the following?

I need to examine a matrix coefficientWise for values great than zero and assign a zero or 1 to a corresponding element in a new matrix

Code: Select all

for (x = 0; x < numRows; x++)
   {
     for (x = 0; x < numRows; x++)
        {
       if (mtxfCold( x, z ) > 0)
       {
         mtxfBinary( x, z ) = 1;
       }
       else
       {
         mtxfBinary( x, z ) = 0;
       }
    }
  }

It seems I read some where something like this might be possible:

mtxfBinary = (mtxfCold > 0).somefunction();

Thanks

for instance:

MatrixXd mtxfBinary;
mtxfBinary = (mtxCold > 0).cast<double>()

replace double by the scalar type you like, (mtxCold > 0) returns an expression of double.

Thx ggael,

I tried this since both matricies were floats

mtxfBinary = ( mtxfCorrelation > 0).cast<float>();

However I'm getting compile errors after that implementation:

error C2784: 'bool std::operator >(const std::unique_ptr<_Ty,_Dx> &,const std::unique_ptr<_Ty2,_Dx2> &)' : could not deduce template argument for 'const std::unique_ptr<_Ty,_Dx> &' from 'Eigen::MatrixXf'
1> c:\Program Files\Microsoft Visual Studio 10.0\VC\include\memory(2602) : see declaration of 'std::operator >'


First in a series of about 60 of them. Would this be a MSVS 2010 compiler issue?

oh, you need to go to the array world for such comparisons: dest = (mat.array() > 0).cast<float>();

Thank you ggael, works great.