This paper presents some of the practical issues that are encountered
when designing and implementing a two-dimensional shape optimization c
omputer program. The topics discussed will be focused on reducing the
number of stiffness inversions (decompositions), treating geometrical
dependencies between decision variables, processing mathematical relat
ions (needed for geometrical constraints or dependencies), the frequen
cy of evaluating gradients, and the steps that are involved in typical
shape optimization process. Copyright (C) 1996 Elsevier Science Ltd.