AnnealedPRSConfigsGen
This class is a child of ConfigsGenAlg. It contains the procedure for the formation of group organizations which optimizes the state of the players, inspired by the Simulated Annealing local search. Sometimes, it acts like a PureRandomSearchConfigsGen, and sometimes it delivers a random configuration without performing any search. The frequency of solution randomness is decreased through time, through a temperature that starts with the value 1.0 and is to be decreased to 0.0 (the decay of the temperature per iteration is defined by the programmer).
Just like in PureRandomSearchConfigsGen, the optimization is naively performed per-iteration and minimal data is stored among iterations. This may entail sub-optimal convergence but it is a simple way to assure adaptation to dynamic environments. This algorithm may be more computationally efficient than the base PRS approach, as initially it may frequently return without performing any optimization.
AnnealedPRSConfigsGen(playerModelBridge, interactionsProfileTemplate, regAlg, persEstAlg, temperatureDecay, numberOfConfigChoices, preferredNumberOfPlayersPerGroup, minNumberOfPlayersPerGroup, maxNumberOfPlayersPerGroup, qualityWeights)| Name: expected type | Default value | Description |
|---|---|---|
| regAlg: RegressionAlg | - | The regression algorithm to be used in the adaptation iterations |
| persEstAlg: PersonalityEstAlg | - | The algorithm used to build players' personality estimates |
| temperatureDecay: int | - | The decay of the temperature in one iteration |
| numberOfConfigChoices: int | 100 | The number of configurations to generate, from where to choose the best one |
| qualityWeights: PlayerCharacteristics | (ability: 0.5, engagement: 0.5) | The importance given to each player characteristic when computing the quality of a configuration |
(No more additional attributes besides the ones from the base class ConfigsGenAlg)
calcQuality(state: [PlayerState](https://github.com/SamGomes/GIMME/wiki/PlayerState)): float(Auxiliary) Computes the quality of a player state, between 0 and 1. A higher value means better quality.
init(): voidInitializes the temperature to 1.0.
reset(temperature): voidResets the temperature to a value between 0.0 and 1.0. If temperature is not in this range, it is clamped.
organize(): int[][]