binary_plot_convergence

poisson_approval.binary_plot_convergence(xyy_to_profile, xscale, yscale, n_max_episodes, init='sincere', samples_per_point=1, perception_update_ratio=<function one_over_log_t_plus_one>, ballot_update_ratio=<function one_over_log_t_plus_one>, title='Convergence frequency', meth='fictitious_play', reverse_right=False, **kwargs)[source]

Shortcut: binary plot for the convergence frequency in fictitious play / iterated voting.

Convergence frequency: out of samples_per_points trials, in which proportion of the cases did fictitious play or iterated voting converge within n_max_episodes iterations?

Parameters

  • xyy_to_profile (XyyToProfile) – This is responsible for generating the profiles.

  • xscale (Number) – Scale of the plot (resolution) on the x-axis.

  • yscale (Number) – Scale of the plot (resolution) on the y-axis.

  • n_max_episodes (int) – Maximum number of episodes for the fictitious play / iterated voting.

  • init (Strategy or TauVector or str) – Cf. fictitious_play() or iterated_voting().

  • samples_per_point (int) – How many trials are made for each point drawn. Useful only when initialization is random.

  • perception_update_ratio (callable or Number) – Cf. fictitious_play() or iterated_voting().

  • ballot_update_ratio (callable or Number) – Cf. fictitious_play() or iterated_voting().

  • title (str) – Title of the plot.

  • meth (str) – The name of the method ('fictitious_play' or 'iterated_voting').

  • reverse_right (bool) – If True, then the y-axis on the right goes decreasing from 1 to 0 (whereas the y-axis on the left goes increasing from 0 to 1).

  • kwargs – Other keyword arguments are passed to the function BinaryAxesSubplotPoisson.heatmap_intensity().

Examples

>>> xyy_to_profile = XyyToProfile(ProfileNoisyDiscrete, left_ranking='bca', right_ranking='cab', noise=0.01)
>>> figure, ax = binary_plot_convergence(xyy_to_profile, xscale=5, yscale=5, n_max_episodes=10)