Get.Game.Object.PD: Get Game Object which fully defines Prisoners Dilemma.

Description Usage Arguments

View source: R/PrisonersDilemmaStratTourn.R

Description

Get Game Object which fully defines Prisoners Dilemma.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
Get.Game.Object.PD(
  encoding.state = "Main.real",
  encoding.action = NULL,
  encoding.params = NULL,
  eval.strategy = Model.strat.Main.real.Exp.Path,
  basic.name.eval.model = "model.strat.global",
  basic.name.eval.model.par = "model.par.strat.global",
  game.setting = "BattleOfStrategiesThesis.Baseline",
  strats = "tit.for.tat"
)

Arguments

encoding.state

Which feature selection should be used to encode the game? Default case is the main encoding of the thesis of Martin Kies, Main.real

encoding.action

Which method should be used to encode the action? Currently supported:

  • main - [C,D]

encoding.params

Insofar the encoding.state is parametrizable, e.g. "Model.strat.RNN.TimeSeries.flexible" or "Encoding.last.X.rounds.PD", here the encoding specific parameters are designed.

eval.strategy

The name of the strategy used if one wants to evaluate the strategy with the package StratTourn or if self-play is used (i.e. if set to play a tournament with the strategy itself being a viable participant). By default "Model.strat.Main.real.Exp.Path" is used, which is designed to work with the encoding "Main.real".

basic.name.eval.model

Global name under which the name of the model is to be saved.

basic.name.eval.model.par

Global name under which the name of the parameter list of the model is to be saved.

game.setting

Default settings of the game. By default the setting "BattleOfStrategiesThesis.Baseline" is used, which is the relevant one for the thesis of Martin Kies. For more information regarding the implemented settings see link{Get.Game.Param.PD}.

strats

The strategies against which one wants to play. If several strategy names are given, each episode a random strategy is chosen. The code name "self" implements self play.


MartinKies/RLR documentation built on Dec. 24, 2019, 10:02 p.m.