Move limits of decision variables, X1 and X2, during design iterations in a single execution of the PANDA2 processor, PANDAOPT

Fig. 83 from the paper, 1987.basic. This slide shows a schematic of the evolution of a design with two decision variables, X1 and X2. With each design iteration the optimizer, ADS, establishes a “window” of permitted excursion of the decision variables. In PANDA2 this “window” shrinks by a factor of 0.8 for each design iteration.

Upon re-execution of PANDAOPT the “window” is re-expanded to its original size, which depends upon lower and upper bounds of the decision variables supplied by the PANDA2 user and certain strategies used by the gradient-based optimizer, ADS.

Because of the "shrinking window" it is always necessary to execute the PANDA2 processor, PANDAOPT, several times in succession in order to determine an optimum design. This multiple execution of PANDAOPT is automatically performed by the "global" optimizers, SUPEROPT and SUPERDUPEROPT.

(from Computers & Structures, Vol.25, No.4, pp. 469-605, 1987)

Page 37 / 49