Abstracting symbolic matrices: Special issue on artificial intelligence and symbolic computation

Abstract

We present a procedure that allows the abstraction of elements in concrete symbolic matrices to obtain a more compact representation employing ellipses in order to expose homogeneous regions present in a matrix. We furthermore extend that procedure to allow for generalisations of concrete matrices to an abstract form that enables us to determine the generic type of a given matrix. The presented algorithms employ artificial intelligence techniques such as pattern recognition and constraint solving. © 2012 Springer Science+Business Media B.V.