Direct Optimization of Experimental Designs
Jānis Auziņš · 10th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference · 2004
'criterion, entropy criterion, discrepancy criterion and others can be used . Univariate relaxation and coordinate exchange algorithm with improved multis tart is used for optimization. The effectiveness of this algorithm is shown for searching of D-optimal design s with continuous and 3 -level discrete parameters in 3 -15 dimensions with 10 to 300 runs (45 to 4500 optimization parameters) and for optimization of Latin hypercube designs according to several criteria. The optimized designs are compared on many metamodeling test problems. For the case of second order local polynomial approximation, the use of MSE -optimal Latin hypercube designs and modified Gauss ian weighting function is proposed.