selector.generators.variable_graph_point_generator

This module contains the variable graph point generator.

Based on the variable tree in GGA [Ansótegui, C., Sellmann, M., Tierney, K.: A Gender-Based Genetic Algorithm for the Automatic Configuration of Algorithms. In: Principles and Practice of Constraint Programming - CP 2009. pp. 142–157 (09 2009).]

Functions

`check_valid`(config_label, C, N, cn)	Check if parameter is actually set in Configuration.
`choose_parents`(mode, data, lookback, results)	Pick configurations according to mode.
`decide_for_O`(config_label, C, N, cn)	Decide new label in case of label O.
`graph_crossover`(graph_structure, C, N, s)	Crossover according to variable graph.
`reset_no_goods`(s, config_setting, label, C, N)	Check if no goods are violated and change value if so.
`set_config_label`(paths, config_label, cn, C, N)	Set label.
`variable_graph_point`(s, identity, results[, ...])	Configuration is generated via variable graph method.
`variable_graph_structure`(s)	General variable graph structure is read from scenario.

Classes

`LabelType`(value)	Contains the types of labels a node can assume.
`Mode`(value)	Contains the types of modes for choosing parent configurations.

class selector.generators.variable_graph_point_generator.LabelType(value)[source]

Bases: IntEnum

Contains the types of labels a node can assume.

C = 1

N = 2

O = 3

class selector.generators.variable_graph_point_generator.Mode(value)[source]

Bases: Enum

Contains the types of modes for choosing parent configurations.

best_and_random = 3

only_best = 2

random = 1

selector.generators.variable_graph_point_generator.check_valid(config_label, C, N, cn)[source]

Check if parameter is actually set in Configuration.

Parameters:

config_label (dict) – Current labeling of parameters.
C (dict) – First configuration.
N (dict) – Second configuration.

Returns:

Updated configuration label.

Return type:

dict

selector.generators.variable_graph_point_generator.choose_parents(mode, data, lookback, results)[source]

Pick configurations according to mode.

Parameters:

mode (selector.generators.variable_graph_point_generator.Mode) – Mode of parent selection.
data (dict of selector.pool.Tournament) – Tournament data to select parents from.
lookback (int) – Number of past tournaments included.

Returns:

C : dict, Configuration C.
N : dict, Configuration N.

Return type:

tuple

selector.generators.variable_graph_point_generator.decide_for_O(config_label, C, N, cn)[source]

Decide new label in case of label O.

Parameters:

config_label (dict) – Current labeling of parameters.
C (dict) – First configuration.
N (dict) – Second configuration.
cn (str) – Child node in the graph.

Returns:

Updated configuration label.

Return type:

dict

selector.generators.variable_graph_point_generator.graph_crossover(graph_structure, C, N, s)[source]

Crossover according to variable graph.

Parameters:

graph_structure (dict of list) – General variable graph structure.
C (dict) – Configuration 1.
N (dict) – Configuration 2.
s (selector.scenario.Scenario) – AC scenario.

Returns:

New configuration setting.

Return type:

dict

selector.generators.variable_graph_point_generator.reset_no_goods(s, config_setting, label, C, N)[source]

Check if no goods are violated and change value if so.

Parameters:

s (selector.scenario.Scenario) – AC scenario.
config_setting (dict) – Parameter value setting.
label (str) – Label.
C (dict) – First configuration.
N (dict) – Second configuration.

Returns:

Updated config_setting.

Return type:

dict

selector.generators.variable_graph_point_generator.set_config_label(paths, config_label, cn, C, N)[source]

Set label.

Parameters:

paths (dict) – Nodes visited until the current node.
config_label (selector.generators.variable_graph_point_generator.LabelType) – Current labeling.
cn (str) – Current node.
C (dict) – First configuration.
N (dict) – Second configuration.

Returns:

Updated config_label.

Return type:

dict

selector.generators.variable_graph_point_generator.variable_graph_point(s, identity, results, mode=Mode.best_and_random, alldata=False, lookback=1, seed=False)[source]

Configuration is generated via variable graph method.

Parameters:

s (selector.scenario.Scenario) – AC scenario.
identity (uuid.UUID) – UUID to identify configuration.
mode (selector.generators.variable_graph_point_generator.Mode) – Mode of parent selection.
data (dict of selector.pool.Tournament) – Tournament data to select parents from.
lookback (int) – Number of past tournaments included.
seed (int) – Random seed.

Returns:

Configuration generated with GGA graph.

Return type:

selector.pool.Configuration

selector.generators.variable_graph_point_generator.variable_graph_structure(s)[source]

General variable graph structure is read from scenario.

Parameters:: s (selector.scenario.Scenario) – AC scenario.
Returns:: General variable graph structure.
Return type:: dict of lists