coexist.combiners.Product#
- class coexist.combiners.Product(weights=None)[source]#
Bases:
object
Class for combining multiple errors by multiplying them.
The multi-objective functionality allows multiple error values to be calibrated / optimised at the same time.
Examples
Use the
Product
combiner in an ACCES run:>>> import coexist >>> access = coexist.Access("<simulation_script>") >>> access.learn( >>> multi_objective = coexist.combiners.Product(), >>> random_seed = 42, >>> )
To test what a combiner will output, use its
.combine(errors)
method - this is what will be called by ACCES:>>> import coexist >>> comb = coexist.combiners.Product(weights = [2, 3]) >>> errors = [2, 2] >>> comb.combine(errors) 32.0
- Attributes:
- weights: float, optional
Raise each error to a power corresponding to its respective weight. If unset, all errors are multiplied with no exponentiation.
Methods
__init__
([weights])combine
(errors)Combine the given errors to a total error value.