Array Processors Optimal mapping of algorithms onto various architectures is studied. Techniques that have been developed in the Laboratory, allow the automatic composition of a finite number of application-specific processors onto arrays that optimally implement the parallel execution of algorithm fragments. Computer Arithmetic Research focuses on the properties of number systems and their potential use in designing high-performance processors for computationally intensive tasks. Techniques for performance improvement in speed, hardware complexity, and power dissipation through the exploitation of number system properties are developed. VLSI Neural Networks Study of computational structures that mimic neural networks with application in pattern recognition. Various topologies are studied and both digital and analog implementations are considered. VLSI Architectures for control applications Research on parallelizing control algorithms is carried out.