Download PDFOpen PDF in browser

Analysis of the causes and evolution of class solidification based on Q-learing and asymmetric repetitive eagle-pigeon game

EasyChair Preprint 8372

2 pagesDate: July 2, 2022

Abstract

The phenomenon of class solidification will lead to the lack of impetus for social production, which can be understood as a less liquid equilibrium formed by all strata of society in the interaction of radical and concessional strategies, so it can be seen as a groups’ repetitive eagle-dove game. Based on the practice of continuous selection strategy of each class group and the characteristics of eagle pigeon game, this paper establishes a class competition game model with parameters, including asymmetric factors µ=ka:kb and the payoff per unit is μ when two sides of the game are in conflict. Take advantage of the characteristics of agents in reinforcement learning to find excellence in trial and error, the Q-learing method is considered to simulate experiments.

In this paper, the repeated experiments of iteration 20000 are simulated under the settings of constant and variable values of parameters µ,considering the gap in strength of class groups, the µ value is divided into five cases: µ = 0.5: 0.5;0.4: 0.6;0.3: 0.7;0.2: 0.8;0.1: 0.9.Run the above experiments independently 100 times each to obtain:(1)as the µ or initial µ values become smaller, the probability of the group choosing the "pigeon-pigeon" strategy will gradually increase, that is, the larger the class gap, the easier it is for the class group to laid-back;(2)when µ = 0.1: 0.9, the low-class group when the µ value is unchanged will directly choose to laid-back, while about 26% of the low-class group is willing to choose a positive strategy when the µ value changes.

It can be seen that the asymmetric factors of class status and the payoff per unit has effectively stimulated the enthusiasm of different class groups.

Keyphrases: Reinforcement Learning, class mobility, class solidification, eagle-pigeon game

BibTeX entry
BibTeX does not have the right entry for preprints. This is a hack for producing the correct reference:
@booklet{EasyChair:8372,
  author    = {Xuan Zhou and Yicheng Gong},
  title     = {Analysis of the causes and evolution of class solidification based on Q-learing and asymmetric repetitive eagle-pigeon game},
  howpublished = {EasyChair Preprint 8372},
  year      = {EasyChair, 2022}}
Download PDFOpen PDF in browser