Fix of rowsReduction in Tridiagonal matrix - wrong format of reduction lambda function. (9df662f9) · Commits · TNL / tnl-dev

Documentation/Examples/Matrices/TridiagonalMatrix/TridiagonalMatrixExample_rowsReduction.cpp

+1 −1

Original line number	Diff line number	Diff line
		@@ -46,7 +46,7 @@ void rowsReduction()
		/***
		* Reduce lambda return maximum of given values.
		*/
		auto reduce = [=] __cuda_callable__ ( double& a, const double& b ) -> double {
		auto reduce = [=] __cuda_callable__ ( const double& a, const double& b ) -> double {
		return TNL::max( a, b );
		};

Documentation/Examples/Matrices/TridiagonalMatrix/TridiagonalMatrixViewExample_forRows.cpp

+10 −10

Original line number	Diff line number	Diff line
		@@ -10,11 +10,11 @@ void forRowsExample()
		* Set the following matrix (dots represent zero matrix elements and zeros are
		* padding zeros for memory alignment):
		*
		* 0 / 1 3 . . . \ -> { 0, 1, 3 }
		* \| 2 1 3 . . \| -> { 2, 1, 3 }
		* \| . 2 1 3 . \| -> { 2, 1, 3 }
		* \| . . 2 1 3 \| -> { 2, 1, 3 }
		* \ . . . 2 1 / 0 -> { 2, 1, 0 }
		* 0 / 2 1 . . . \ -> { 0, 2, 1 }
		* \| 3 2 1 . . \| -> { 3, 2, 1 }
		* \| . 3 2 1 . \| -> { 3, 2, 1 }
		* \| . . 3 2 1 \| -> { 3, 2, 1 }
		* \ . . . 3 2 / 0 -> { 3, 2, 0 }
		*/
		TNL::Matrices::TridiagonalMatrix< double, Device > matrix(
		5, // number of matrix rows
		@@ -29,11 +29,11 @@ void forRowsExample()
		*
		* 0 1 2 <- localIdx values
		* -------
		* 0 / 1 3 . . . \ -> { 0, 1, 3 }
		* \| 2 1 3 . . \| -> { 2, 1, 3 }
		* \| . 2 1 3 . \| -> { 2, 1, 3 }
		* \| . . 2 1 3 \| -> { 2, 1, 3 }
		* \ . . . 2 1 / 0 -> { 2, 1, 0 }
		* 0 / 2 1 . . . \ -> { 0, 2, 1 }
		* \| 3 2 1 . . \| -> { 3, 2, 1 }
		* \| . 3 2 1 . \| -> { 3, 2, 1 }
		* \| . . 3 2 1 \| -> { 3, 2, 1 }
		* \ . . . 3 2 / 0 -> { 3, 2, 0 }
		*
		*/
		value = 3 - localIdx;

src/TNL/Matrices/TridiagonalMatrixView.hpp

+18 −18

Original line number	Diff line number	Diff line
		@@ -98,8 +98,8 @@ getCompressedRowLengths( Vector& rowLengths ) const
		auto fetch = [] __cuda_callable__ ( IndexType row, IndexType column, const RealType& value ) -> IndexType {
		return ( value != 0.0 );
		};
		auto reduce = [] __cuda_callable__ ( IndexType& aux, const IndexType a ) {
		aux += a;
		auto reduce = [] __cuda_callable__ ( const IndexType& aux, const IndexType a ) -> IndexType {
		return aux + a;
		};
		auto keep = [=] __cuda_callable__ ( const IndexType rowIdx, const IndexType value ) mutable {
		rowLengths_view[ rowIdx ] = value;
		@@ -275,23 +275,23 @@ rowsReduction( IndexType first, IndexType last, Fetch& fetch, Reduce& reduce, Ke
		Real_ sum( zero );
		if( rowIdx == 0 )
		{
		reduce( sum, fetch( 0, 1, values_view[ indexer.getGlobalIndex( 0, 1 ) ] ) );
		reduce( sum, fetch( 0, 2, values_view[ indexer.getGlobalIndex( 0, 2 ) ] ) );
		sum = reduce( sum, fetch( 0, 1, values_view[ indexer.getGlobalIndex( 0, 1 ) ] ) );
		sum = reduce( sum, fetch( 0, 2, values_view[ indexer.getGlobalIndex( 0, 2 ) ] ) );
		keep( 0, sum );
		return;
		}
		if( rowIdx + 1 < indexer.getColumns() )
		{
		reduce( sum, fetch( rowIdx, rowIdx - 1, values_view[ indexer.getGlobalIndex( rowIdx, 0 ) ] ) );
		reduce( sum, fetch( rowIdx, rowIdx, values_view[ indexer.getGlobalIndex( rowIdx, 1 ) ] ) );
		reduce( sum, fetch( rowIdx, rowIdx + 1, values_view[ indexer.getGlobalIndex( rowIdx, 2 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx - 1, values_view[ indexer.getGlobalIndex( rowIdx, 0 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx, values_view[ indexer.getGlobalIndex( rowIdx, 1 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx + 1, values_view[ indexer.getGlobalIndex( rowIdx, 2 ) ] ) );
		keep( rowIdx, sum );
		return;
		}
		if( rowIdx < indexer.getColumns() )
		{
		reduce( sum, fetch( rowIdx, rowIdx - 1, values_view[ indexer.getGlobalIndex( rowIdx, 0 ) ] ) );
		reduce( sum, fetch( rowIdx, rowIdx, values_view[ indexer.getGlobalIndex( rowIdx, 1 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx - 1, values_view[ indexer.getGlobalIndex( rowIdx, 0 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx, values_view[ indexer.getGlobalIndex( rowIdx, 1 ) ] ) );
		keep( rowIdx, sum );
		}
		else
		@@ -319,23 +319,23 @@ rowsReduction( IndexType first, IndexType last, Fetch& fetch, Reduce& reduce, Ke
		Real_ sum( zero );
		if( rowIdx == 0 )
		{
		reduce( sum, fetch( 0, 1, values_view[ indexer.getGlobalIndex( 0, 1 ) ] ) );
		reduce( sum, fetch( 0, 2, values_view[ indexer.getGlobalIndex( 0, 2 ) ] ) );
		sum = reduce( sum, fetch( 0, 1, values_view[ indexer.getGlobalIndex( 0, 1 ) ] ) );
		sum = reduce( sum, fetch( 0, 2, values_view[ indexer.getGlobalIndex( 0, 2 ) ] ) );
		keep( 0, sum );
		return;
		}
		if( rowIdx + 1 < indexer.getColumns() )
		{
		reduce( sum, fetch( rowIdx, rowIdx - 1, values_view[ indexer.getGlobalIndex( rowIdx, 0 ) ] ) );
		reduce( sum, fetch( rowIdx, rowIdx, values_view[ indexer.getGlobalIndex( rowIdx, 1 ) ] ) );
		reduce( sum, fetch( rowIdx, rowIdx + 1, values_view[ indexer.getGlobalIndex( rowIdx, 2 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx - 1, values_view[ indexer.getGlobalIndex( rowIdx, 0 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx, values_view[ indexer.getGlobalIndex( rowIdx, 1 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx + 1, values_view[ indexer.getGlobalIndex( rowIdx, 2 ) ] ) );
		keep( rowIdx, sum );
		return;
		}
		if( rowIdx < indexer.getColumns() )
		{
		reduce( sum, fetch( rowIdx, rowIdx - 1, values_view[ indexer.getGlobalIndex( rowIdx, 0 ) ] ) );
		reduce( sum, fetch( rowIdx, rowIdx, values_view[ indexer.getGlobalIndex( rowIdx, 1 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx - 1, values_view[ indexer.getGlobalIndex( rowIdx, 0 ) ] ) );
		sum = reduce( sum, fetch( rowIdx, rowIdx, values_view[ indexer.getGlobalIndex( rowIdx, 1 ) ] ) );
		keep( rowIdx, sum );
		}
		else
		@@ -487,8 +487,8 @@ vectorProduct( const InVector& inVector,
		auto fetch = [=] __cuda_callable__ ( const IndexType& row, const IndexType& column, const RealType& value ) -> RealType {
		return value * inVectorView[ column ];
		};
		auto reduction = [] __cuda_callable__ ( RealType& sum, const RealType& value ) {
		sum += value;
		auto reduction = [] __cuda_callable__ ( const RealType& sum, const RealType& value ) -> RealType {
		return sum + value;
		};
		auto keeper1 = [=] __cuda_callable__ ( IndexType row, const RealType& value ) mutable {
		outVectorView[ row ] = value;