From af4e1d3f5f8c7bb1555f8f47e2772bf46573939d Mon Sep 17 00:00:00 2001 From: Martin Kronbichler Date: Tue, 11 Jul 2017 08:42:15 +0200 Subject: [PATCH] Add changelog --- doc/news/changes/major/20170711MartinKronbichler-b | 6 ++++++ doc/news/changes/minor/20170711MartinKronbichler | 9 +++++++++ 2 files changed, 15 insertions(+) create mode 100644 doc/news/changes/major/20170711MartinKronbichler-b create mode 100644 doc/news/changes/minor/20170711MartinKronbichler diff --git a/doc/news/changes/major/20170711MartinKronbichler-b b/doc/news/changes/major/20170711MartinKronbichler-b new file mode 100644 index 0000000000..91cacbff77 --- /dev/null +++ b/doc/news/changes/major/20170711MartinKronbichler-b @@ -0,0 +1,6 @@ +Improved: The constructor of MappingQGeneric used a naive inversion of a cell +Poisson matrix, which became very slow for high approximation degrees in +3D. The implementation has been switched to the fast diagonalization method +implemented by TensorProductMatrixSymmetricSum. +
+(Martin Kronbichler, 2017/07/11) diff --git a/doc/news/changes/minor/20170711MartinKronbichler b/doc/news/changes/minor/20170711MartinKronbichler new file mode 100644 index 0000000000..011d5d911a --- /dev/null +++ b/doc/news/changes/minor/20170711MartinKronbichler @@ -0,0 +1,9 @@ +New: A new class TensorProductMatrixSymmetricSum that represents a tensor +product of a mass and derivative matrix has been added. It implements both the +matrix-vector operation (TensorProductMatrixSymmetricSum::vmult()) as well as +the inverse operation (TensorProductMatrixSymmetricSum::apply_inverse()) +operation in tensorial form with optimal complexity of +sizedim+1 operations. The inverse operation uses the fast +diagonalization method. +
+(Martin Kronbichler, 2017/07/11) -- 2.39.5