mdp_computePpolicyPRpolicy: Computes the transition matrix and the reward matrix for a...

In MDPtoolbox: Markov Decision Processes Toolbox

Description

Computes the transition matrix and the reward matrix for a given policy.

Usage

mdp_computePpolicyPRpolicy(P, R, policy)

Arguments

P	transition probability array. P can be a 3 dimensions array [S,S,A] or a list [[A]], each element containing a sparse matrix [S,S].
R	reward array. R can be a 3 dimensions array [S,S,A] or a list [[A]], each element containing a sparse matrix [S,S] or a 2 dimensional matrix [S,A] possibly sparse.
policy	a policy. policy is a length S vector of integer representing actions.

Details

mdp_computePpolicyPRpolicy computes the state transition matrix and the reward matrix of a policy, given a probability matrix P and a reward matrix.

Value

Ppolicy	transition probability array of the policy. Ppolicy is a [S,S] matrix.
PRpolicy	reward matrix of the policy. PRpolicy is a vector of length S.

Examples

# With a non-sparse matrix
P <- array(0, c(2,2,2))
P[,,1] <- matrix(c(0.6116, 0.3884, 0, 1.0000), 2, 2, byrow=TRUE)
P[,,2] <- matrix(c(0.6674, 0.3326, 0, 1.0000), 2, 2, byrow=TRUE)
R <- array(0, c(2,2,2))
R[,,1] <- matrix(c(-0.2433, 0.7073, 0, 0.1871), 2, 2, byrow=TRUE)
R[,,2] <- matrix(c(-0.0069, 0.6433, 0, 0.2898), 2, 2, byrow=TRUE)
policy <- c(2,2)
mdp_computePpolicyPRpolicy(P, R, policy)

# With a sparse matrix (P)
P <- list()
P[[1]] <- Matrix(c(0.6116, 0.3884, 0, 1.0000), 2, 2, byrow=TRUE, sparse=TRUE)
P[[2]] <- Matrix(c(0.6674, 0.3326, 0, 1.0000), 2, 2, byrow=TRUE, sparse=TRUE)
mdp_computePpolicyPRpolicy(P, R, policy)

MDPtoolbox documentation built on May 2, 2019, 2:10 p.m.