ABSTRACT
Several numerical issues and techniques from numerical linear algebra together with a number of important applications of these ideas have been outlined. A key question in these and other problems in systems, control, and estimation theory is what can be computed reliably and used in the presence of parameter uncertainty or structural constraints (e.g., certain “hard zeros”) in the originalmodel, and rounding errors in the calculations. However, because the ultimate goal is to solve real problems, reliable tools (mathematical software) and experience must be available to effect real solutions or strategies. The interdisciplinary effort during the last few decades has significantly improved our understanding of the issues involved in reaching this goal andhas resulted in somehigh-quality control software based onnumerically reliable and well-tested algorithms. This provides clear evidence of the fruitful symbiosis between numerical analysis and numerical problems from control. We expect this symbiotic relationship to flourish, as control engineering realizes the full potential of the computingpowerbecomingmorewidely available inmultiprocessing systems and high-performanceworkstations.However, as in other applications areas, software continues to act as a constraint and a vehicle for progress. Unfortunately, high-quality software is very expensive.
