QHyper.util#

This module contains utility functions that are used across the project.

Functions

weighted_avg_evaluation(results, score_function, penalty=0, limit_results=None, normalize=True)[source]#

Calculate weighted average evaluation of results.

Example:

results = solver.solve()
score = weighted_avg_evaluation(
    results.probabilities, solver.problem.get_score, penalty=3,
    limit_results=100, normalize=True)
Parameters:

  • results (np.recarray) – Results to evaluate. It should contain variables and probability.

  • score_function (Callable[[np.record, float], float]) – Function to evaluate results. Most often it’s a problem’s get_score method.

  • penalty (float, optional) – Penalty for the constraint violation, by default 0

  • limit_results (int, optional) – Number of results to evaluate, by default None

  • normalize (bool, optional) – Normalize the score, by default True, applicable when the limit is set

Returns:

Weighted average evaluation of results.

Return type:

float

sort_solver_results(results, limit_results=None)[source]#

Sort solver results by probability.

Example:

results = solver.solve()
sorted_results = sort_solver_results(results.probabilities, 100)
Parameters:

  • results (np.recarray) – Results to sort. It should contain variables and probability.

  • limit_results (int, optional) – Number of results to return, by default None

Returns:

Sorted results.

Return type:

np.recarray

add_evaluation_to_results(results, score_function, penalty=0)[source]#

Add evaluation to results.

Example:

results = solver.solve()
add_evaluation_to_results(
    results.probabilities, solver.problem.get_score)
Parameters:

  • results (np.recarray) – Results to evaluate. It should contain variables and probability.

  • score_function (Callable[[np.record, float], float]) – Function to evaluate results. Most often it’s a problem’s get_score method.

  • penalty (float, optional) – Penalty for the constraint violation, by default 0

Returns:

Results with evaluation added. Can be found under ‘evaluation’ key.

Return type:

np.recarray

search_for(class_type, path)[source]#

This function searches for classes of a given type in a given path.

If class contains a name attribute, it will be used as a key in the returned dictionary. Otherwise, the class name will be used. Either way, the key will be lowercased.

Parameters:

  • class_type (type) – Type of the class to search for e.g. Problem, Solver.

  • path (str) – Path to the file or directory to search in.

Returns:

Dictionary of found classes with their names as keys and classes as values.

Return type:

dict[str, type]