Speeding up DSI with an algebraic multigrid method

Thomas Jerome and Mary Ford. ( 2001 )
in: $21^st$ Gocad Meeting Proceedings

Abstract

The current speed of convergence of DSI is directly linked to the ratio between the number of constraints and the number of nodes of the interpolated object. If this ratio is high, the best approximate solution is quickly reached. If it is low, DSI converges slowly. To speed up the convergence, the resolution of the local equation of DSI is modi¯ed by adding an algebraic multigrid method. This method, applicable to any Gocad object (atomic or gridded), is extremely fast as it has the advantage to speed up the DSI convergence: The number of iterations is then no longer dependent on the mesh size.

Download / Links

    BibTeX Reference

    @INPROCEEDINGS{Jerome2001a,
        author = { Jerome, Thomas and Ford, Mary },
         title = { Speeding up DSI with an algebraic multigrid method },
     booktitle = { $21^st$ Gocad Meeting Proceedings },
          year = { 2001 },
      abstract = { The current speed of convergence of DSI is directly linked to the ratio between the
    number of constraints and the number of nodes of the interpolated object. If this ratio is
    high, the best approximate solution is quickly reached. If it is low, DSI converges slowly. To
    speed up the convergence, the resolution of the local equation of DSI is modi¯ed by adding
    an algebraic multigrid method. This method, applicable to any Gocad object (atomic or
    gridded), is extremely fast as it has the advantage to speed up the DSI convergence: The
    number of iterations is then no longer dependent on the mesh size. }
    }