Experiment Module

Experiment is a convenient way to configure and run your computations.

Experiment

class pmp.experiments.Experiment(config=None)

Experiment to run

add_election(rule, k, id)
Parameters

  • rule (Rule) – Scoring rule for added election

  • k (int) – Size of the winning committee for added election

  • id (str) – Text id for the election. Builds up first part of output filenames

For multi-rule experiments. It’s preferred way of defining elections in experiments.

get_generated_dir_path()
Returns

Path to the root directory where files are generated

Return type

str

run(visualization=False, n=1, save_win=False, save_in=False, save_out=False, log_on=True, elect_configs=None, split_dirs=True)
Parameters

  • visualization (bool) –

  • n (int) –

  • save_win (bool) –

  • save_in (bool) –

  • save_out (bool) –

  • log_on (bool) –

  • elect_configs (List[ElectionConfig]) – Election configs. If given, experiment ignores it’s one-rule configuration

  • split_dirs (bool) – When True create separate directory for each election related files

Run experiment. Experiment runs elections configured in following precedence: * from elect_configs parameter, if present * set up from election_configurations added by add_election, if added * set up from set_election

set_election(rule, k)
Parameters

  • rule (Rule) – Scoring rule being set

  • k (int) – Size of the committee being set

[Deprecated] Set election parameters the rule and the size of the committee. Overrides previous experiment setup. Only for setting up one-rule experiments.

set_generated_dir_path(dir_path)
Parameters

dir_path (str) – Path to the root directory where files are generated

set_inout_filename(name)
Parameters

name (str) – Inout filename

Set filename of files containing generated candidates and voters. Overrides previous value. Only for setting up one-rule experiments.

set_result_filename(name)
Parameters

name (str) – Result filename

[Deprecated] Set result name. It builds up first part of names of all files being generated during experiment. Overrides previous value. Only for setting up one-rule experiments.

ExperimentConfig

class pmp.experiments.ExperimentConfig(id='')

Store candidates and voters configuration. Above properties can be stored as static values or generating functions, so different setup is generated each time.

get_candidates()
Returns

Copy of candidates

Return type

List

impartial(m, n)

Candidates in impartial only as a list of consecutive integers starting from 0

set_candidates(list_of_candidates)
Parameters

list_of_candidates (List) – List of candidates

ElectionConfig

class pmp.experiments.ElectionConfig(rule, k, id)

Election configuration used by an Experiment

Histogram

class pmp.experiments.Histogram(minx, maxx, miny, maxy, opacity_mask=(1.0, 1.0, 1.0), ppu=None, W=256, H=256)

Two-dimensional histogram.

accumulate(points)
Parameters

points (List[Tuple[float, float]]) – List of points to be accumulated into histogram.

draw_fixed_points(points, color, size=0)
Parameters

  • points (List[Tuple[float, float]]) – List of points to be draw.

  • color (Tuple[int, int, int]) – RGB color of points.

  • size (int) – Size of points. When equal 0 draw single pixel.

Draw fixed points over the main histogram layer.

save_image(path)

Saves the histogram image. Fixed points are drawn above the histogram itself.

Generaring Functions

pmp.experiments.generating_functions.random() → x in the interval [0, 1).

Helpers

pmp.experiments.helpers.process_win_dir(path, strategy)
Parameters

  • path (str) – path of processed directory

  • strategy (Callable[List, List, List, List, str]) – Run with {candidates, voters, preferences, winners, election}

Helper for processing experiment-generated directories. Visits all election directories stored in path directory. After loading .win file runs strategy with candidates, voters, preferences, winners, election args. Election is a string id.