🚀 go-pugleaf

		 Vendredi 4 novembre - 14h
	    Salle de conferences Michel Metivier

	    ************************************

			Edward A. LEE
		  University of California
		       Berkeley - USA

	     ***********************************

	     Dataflow Process Networks and their 
	      Application to Signal Processing

	     ***********************************



This talk reviews a class of models of computation called dataflow.
These models are used in industrial practice in signal processing software
environments and experimentally in other contexts.  The basic model
can be shown to be a special case of Kahn process networks where each
process consists of repeated atomic "firings" of a dataflow "actor".
An actor defines a (usually functional) quantum of computation. The
dataflow model bears interesting relationships to functional languages,
and can serve as a basis for language design.  It is compatible with
such modern language concepts such as higher-order functions and
polymorphism, and is particularly well suited to a graphical syntax.
Various experiments with this model have been conducted using the
Ptolemy environment from Berkeley.