get_morris_samples

pyapprox.sensitivity_analysis.get_morris_samples(nvars, nlevels, ntrajectories, eps=0, icdfs=None)[source]

Compute a set of Morris trajectories used to compute elementary effects

Parameters
nvarsinteger

The number of variables

nlevelsinteger

The number of levels used for to define the morris grid.

ntrajectoriesinteger

The number of Morris trajectories requested

epsfloat

Set grid used defining the Morris trajectory to [eps,1-eps]. This is needed when mapping the morris trajectories using inverse CDFs of unbounded variables

icdfslist (nvars)

List of inverse CDFs functions for each variable

Returns
trajectoriesnp.ndarray (nvars,ntrajectories*(nvars+1))

The Morris trajectories

Notes

The choice of nlevels must be linked to the choice of ntrajectories. For example, if a large number of possible levels is used ntrajectories must also be high, otherwise if ntrajectories is small effort will be wasted because many levels will not be explored. nlevels=4 and ntrajectories=10 is often considered reasonable.