quantlab.factor.kunquant_cs.RenormalizedCombine#

class quantlab.factor.kunquant_cs.RenormalizedCombine(values, weights)[source]#

Bases: CompositiveOp

Fixed-weight sum of several values, renormalized over the ones present.

sum(w_k * v_k) / sum(w_k) over the k whose v_k is finite, so a symbol missing one descriptor still gets the combination of the others, as USE4 combines descriptors into a style. NaN when none is present.

Parameters:

valueslist of OpBase: The values to combine, such as standardized descriptors.
weightslist of float: One positive weight per value.

Raises:

ValueError: If the lengths differ, no value is given, or a weight is not positive.

Examples

>>> Output(RenormalizedCombine([dastd, cmra, hsigma], [0.75, 0.15, 0.10]), "resvol")

decompose(options)[source]#: Expand into masked sums and one division.