Package 'RSGSolve'

Loads a json file rsg game specification and returns the rsg object as R list You then can call solveSG on the result to get the approximations to the equilibrium payoff sets

Description

Loads a json file rsg game specification and returns the rsg object as R list You then can call solveSG on the result to get the approximations to the equilibrium payoff sets

Usage

loadJsonSG(file)

Arguments

file	the file name of the json file

---

Hello, Rcpp!

Description

Returns an R list containing the character vector c("foo", "bar") and the numeric vector c(0, 1).

Usage

rcpp_hello()

Examples

rcpp_hello()

---

Solve a two-player stochastic game of perfect monitoring

Description

Uses Ben Brooks implementation SGSolve of the pencil sharpening algorithm of Abreu, Brooks and Sanikov (2016) See http://www.benjaminbrooks.net/sgsolvedoc/html/index.html

Usage

solveSG(delta = rsg$delta, states = rsg$states, rsg = NULL,
  duplicate.first.point = TRUE, all.iter = TRUE, tol = 1e-12,
  normtol = tol, directiontol = tol, leveltol = tol, improvetol = tol)

Arguments

delta	the discount factor between 0 and 1
all.iter	shall return value contain the field ipoints that contains the pivots of all iterations?
stages	a list of stages. Each stage is a list with the following elements: - numActions: a size 2 vector that contains the number of actions for player 1 and 2. Its product is numActionProfiles. - payoffs: a matrix of dimension numActionProfiles x 2. The payoffs as function of the action profile. The actions are mapped to action profiles as described in the documentation of Brooks C++ library: A pair (a1,a2) is mapped into an action profile index using the formula a=a1+a2*numActions[s][a1]. - transition: a matrix of dimension numActionProfiles x numStates. The transition probabilities. Each row has to sum up to 1. @param duplicate.first.point if TRUE (default) the first row o the point matrices will be added again to the end. This facilitates plotting of the payoff set using the lines command.

Value

a list with the following elements solved: TRUE if the game could be solved

points: a list with a matrix for every state that contains the the extreme points of the final payoff set approximation. Using the terminology of ABS (2016): the pivots from the last revolution.

ipoints: a list with a matrix for every state, containing the pivots (extreme points) from every iteration.

revolution: a vector that denotes the revolution of each point in the ipoints matrices.

---