class documentation
class EvaluationManager(object):
Constructor: SurrogatePopulation.EvaluationManager(X)
Manage incremental evaluation of a population of solutions.
Evaluate solutions, add them to the model and keep track of which solutions were evaluated.
Uses model.add_data_row and model.eval.
Details: for simplicity and avoiding the copy construction, we do not
inherit from list. Hence we use self.X instead of self.
| Method | __init__ |
all is based on the population (list of solutions) X |
| Method | __len__ |
should replace len(self.X) etc, not fully in use yet |
| Method | add |
add fitness(self.X[i]), not in use |
| Method | eval |
add fitness(self.X[i]) to model data, mainly for internal use |
| Method | eval |
evaluate unevaluated entries of X[idx] until number entries are evaluated overall. |
| Method | surrogate |
return surrogate values of model_eval with smart offset. |
| Instance Variable | evaluated |
Undocumented |
| Instance Variable | fvalues |
Undocumented |
| Instance Variable | last |
Undocumented |
| Instance Variable | X |
Undocumented |
| Property | evaluation |
Undocumented |
| Property | evaluations |
Undocumented |
| Property | remaining |
number of not yet evaluated solutions |
evaluate unevaluated entries of X[idx] until number entries are
evaluated overall.
Assumes that sorted(idx) == list(range(len(self.X))).
idx defines the evaluation sequence.
The post condition is self.evaluations == min(number, len(self.X)).