[Insight-developers] about the itk::fem(post again)

[Yixun Liu]

Hi,

I post the problem abou itk::fem several times, but no any reply. I need your help very much and any reply will be appreciated!

1. may be a bug in the itk::fem::applyBC

The extracted code from the code\numerics\fem\femsolver.cxx is shown below:


  // Step over each nonzero matrix element in a row
  for(LinearSystemWrapper::ColumnArray::iterator c=cols.begin(); c!=cols.end(); c++)
  {
    // Get value from the stiffness matrix
    Float d=this->m_ls->GetMatrixValue(fdof, *c, matrix);

    // Store the appropriate value in bc correction vector (-K12*u2)
    //
    // See http://titan.colorado.edu/courses.d/IFEM.d/IFEM.Ch04.d/IFEM.Ch04.pdf
    // chapter 4.1.3 (Matrix Forms of DBC Application Methods) for more info.
    this->m_ls->AddVectorValue(*c,-d*fixedvalue,1);
  }


the method which itk use to apply BC is in the http://titan.colorado.edu/courses.d/IFEM.d/IFEM.Ch04.d/IFEM.Ch04.pdf 
However, the code in the Bold format may be wrong. It should be :
Float d=this->m_ls->GetMatrixValue(*c,fdof,  matrix);

I need your affirmation!


2. I fix some DOFs with specific values when applying the BC, however, the solution of the linear system is different at DOF which I fix with the specific value. 


Thank you in advance!

Regards,

Yixun Liu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.itk.org/mailman/private/insight-developers/attachments/20040806/3079a124/attachment.html