A DISTRIBUTED-MEMORY PARALLEL ALGORITHM FOR THE EFFICIENT COMPUTATIONOF SENSITIVITIES OF DIFFERENTIAL-ALGEBRAIC SYSTEMS

An efficient algorithm for the evaluation of the parametric sensitivit ies for mixed systems of differential and algebraic equations (DAEs) o n computers involving multiple processors operating in parallel is pre sented. The algorithm derives its efficiency by decoupling the integra tion of the sensitivity equations from that of the original DAE system , and by allowing tasks associated with the evaluation of sensitivitie s at multiple time points to overlap instead of bring carried out in s equence. Numerical experiments demonstrating the efficiency of the pro posed algorithm with systems of more than 850 DAEs and 45 parameters a re presented. In all the cases studied, the simultaneous integration o f the original DAEs and their sensitivity equations is carried out in less than 10% more time than that of the original DAEs alone. (C) 1998 IMACS/Elsevier Science B.V.