Weighted Latin Hypercube Sampling (WLHS)

In some cases, a very small probability of failure (), such as in the order of 10-6, is of interest in engineering design. It would require simulations/runs of regular LHS. To reduce the number of runs, WLHS can be used to generate biased/more samples in the region.

In WLHS, the input variables are discretized unevenly/unequally in their design space. The cell (or hypercube, in multiple dimensions) size of probability of occurrence is evaluated according to topology of output/response. The cell size of probability of occurrence is discretized such that it is smaller (relatively) around minimum and maximum of output/response. As evaluation of cell size is output/response oriented, WLHS is somewhat unsymmetrical/biased.

In general, WLHS is intended to stretch distribution farther out in the tails (lower and upper) with less number of runs than LHS. This means that given same number of runs, WLHS is expected to reach a smaller than LHS. Due to biasness, however, the evaluated can be subject to some difference compared to LHS.


Release 18.2 - © ANSYS, Inc. All rights reserved.