BanditInteraction#

class statrl.settings.bandits.stochastic.knownhorizon.interaction.BanditInteraction[source]#

Bases: Interaction

Interaction loop for the known-horizon stochastic bandit setting.

Identical to the anytime loop except that the horizon is passed to learner.reset(horizon) instead of being withheld.

See also

statrl.settings.bandits.stochastic.anytime.interaction.BanditInteraction

The anytime counterpart.

Examples

>>> from statrl.settings.bandits.stochastic.anytime.envs.parametric import BernoulliBandit
>>> from statrl.settings.bandits.stochastic.anytime.agents.IMED import IMED
>>> from statrl.settings.bandits.stochastic.knownhorizon.wrappers.wrapper_anytime_knownhorizon import (
...     AnytimeToKnownHorizonAgentWrapper)
>>> env = BernoulliBandit([0.2, 0.9, 0.5])
>>> agent = AnytimeToKnownHorizonAgentWrapper(IMED(env.number_arms))
>>> BanditInteraction().run(env, agent, horizon=50).shape
(50,)

Methods

__init__()

renderrun(env, learner, horizon)

Run one interaction, printing each round to stdout.

run(env, learner, horizon)

Run one interaction and return its cumulative expected score.

Attributes

plotlabels

Axis labels (x, y) for the regret plots.
property plotlabels#

Axis labels (x, y) for the regret plots.

Type:

tuple of (str, str)

renderrun(env, learner, horizon)[source]#

Run one interaction, printing each round to stdout.

Parameters:

  • env (StochasticBanditEnv) – The bandit instance. A Textrenderer is appended to its renderers, so calling this twice on the same environment prints every round twice.

  • learner (BanditAgent) – The agent.

  • horizon (int) – Number of rounds to play.

run(env, learner, horizon)[source]#

Run one interaction and return its cumulative expected score.

Parameters:

  • env (StochasticBanditEnv) – The bandit instance; the same class as in the anytime setting, since only the agent interface differs between the two.

  • learner (BanditAgent) – The agent, reset with horizon so it can plan against it.

  • horizon (int) – Number of rounds to play.

Returns:

Cumulative sum of the expected rewards of the arms played.

Return type:

ndarray of shape (horizon,)