next up previous contents
Next: 1.5 Implementation of Basic Up: 1.4 Redistributing and Duplicating Previous: 1.4.3 Reducing vectorsmatrix

1.4.4 Terminology

 

We explicitly make a distinction between a row or column of a matrix and a vector. Whenever a vector exists as a row or column of a matrix, we will refer to it as a (matrix) row     or (matrix) column.     Whenever a vector is distributed like a row or column of a matrix, without being part of a matrix, we will refer to it as a projected row vector or projected column vector.             If a vector is distributed like a row of a matrix, but a copy exists within every row of nodes, we will refer to it as a duplicated (projected) row vector (projected row vector duplicated to all rows of nodes).           Similarly, if a vector is distributed like a column of a matrix, but a copy exists within every column of nodes, we will refer to it as a duplicated (projected) column vector (projected column vector duplicated to all columns of nodes).        


next up previous contents
Next: 1.5 Implementation of Basic Up: 1.4 Redistributing and Duplicating Previous: 1.4.3 Reducing vectorsmatrix

rvdg@cs.utexas.edu