A MatrixMap is a form of Mapping which performs a general linear transformation. Each set of input coordinates, regarded as a column-vector, are pre-multiplied by a matrix (whose elements are specified when the MatrixMap is created) to give a new column-vector containing the output coordinates. If appropriate, the inverse transformation may also be performed.
A value of zero indicates that a full NOUT x NIN matrix of values will be supplied via the MATRIX argument (below). In this case, the elements should be given in row order (the elements of the first row, followed by the elements of the second row, etc.).
A value of 1 indicates that only the diagonal elements of the matrix will be supplied, and that all others should be zero. In this case, the elements of MATRIX should contain only the diagonal elements, stored consecutively.
A value of 2 indicates that a
" matrix is required, whose diagonal elements are set to unity
(with all other elements zero). In this case, the MATRIX argument is not used.
In general, a MatrixMap
’ s forward transformation will always be available (as indicated by its
TranForward attribute), but its inverse transformation (TranInverse attribute) will only be available if
the associated matrix is square and non-singular.
As an exception to this, the inverse transformation is always available if a unit or diagonal matrix is specified. In this case, if the matrix is not square, one or more of the input coordinate values may not be recoverable from a set of output coordinates. Any coordinates affected in this way will simply be set to the value zero.
A null Object pointer (AST__NULL) will be returned if this function is invoked with STATUS set to an error value, or if it should fail for any reason.