The SampleWeighting system variable defines a global relative weighting for sample points. It is implicitly or explicitly indexed by Run. It may contain additional indexes, but if it does, these will appear in every uncertain result.

By default, SampleWeighting := 1, meaning that every sample has equal weight.

To change the SampleWeighting, access its object window from the Definition Menu. First, click in a diagram so that you are not editing a definition, then select Definition → System Variables → SampleWeighting. From there, you can fill in its definition.

By setting SampleWeighting system variable appropriate, it is possible to perform various forms of Importance Sampling and Bayesian inference. This can be particularly useful when modeling extreme events, allowing more samples to be concentrated in the tails of the distributions, but weighting the samples so that the viewed distributions and results are not biased.

See Also


You are not allowed to post comments.