bound
fun boundFor(label: String, weights: DoubleArray, components: List<ContinuousDistributionIfc>, types: List<RVParametersTypeIfc>, numRefinements: Int = 6, numPoints: Int = 8001): SeparabilityBound.Bound?(source)
The closest (k-1)-component mixture reachable by merging one adjacent pair.
The grid, the true density on it, and each true component's density on it are computed once per case. A candidate then costs one density evaluation for the merged component rather than a rebuilt grid and a re-evaluated mixture, which is the difference between this running in a minute and running for hours: the grid construction alone bisects for an effective range per component, and the search visits on the order of a quarter of a million candidates.
Parameters
case
the design case
num
how many coordinate-search passes to run on the merged moments
num
the quadrature resolution