From: bangerth Date: Thu, 21 Feb 2013 00:07:11 +0000 (+0000) Subject: Provide more explanation for the Tvmult function. X-Git-Url: https://gitweb.dealii.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=4427d734f418986cbb41b1db8d1db56c40286a88;p=dealii-svn.git Provide more explanation for the Tvmult function. git-svn-id: https://svn.dealii.org/trunk@28499 0785d39b-7218-0410-832d-ea1e28bc413d --- diff --git a/deal.II/examples/step-20/step-20.cc b/deal.II/examples/step-20/step-20.cc index 49bb88bd09..4bc9d5ef8a 100644 --- a/deal.II/examples/step-20/step-20.cc +++ b/deal.II/examples/step-20/step-20.cc @@ -607,8 +607,15 @@ namespace Step20 // step. Consequently, the class also does not have to store a pointer to an // inverse mass matrix object. // - // Since InverseIterate follows the standard convention for matrices, we - // need to provide a Tvmult function here as well. + // We will later use this class as a template argument to the + // IterativeInverse class which will in turn want to use it as a + // template argument for the PointerMatrix class. The latter class + // has a function that requires us to also write a function that + // provides the product with the transpose of the matrix this object + // represents. As a consequence, in the code below, we also + // implement a Tvmult function here that represents the + // product of the transpose matrix with a vector. It is easy to see + // how this needs to be implemented here. class ApproximateSchurComplement : public Subscriptor { public: