numsteps <- 100

A <- rbind(c(.9,.07),
           c(.1,.93))

#list of states on this realization. xlist(k)=1 means in state 1 at
#timestep k, etc
states <- rep(0,numsteps)

# init
states[1] <- 1

for (k in 1:(numsteps-1)) {
  #uniformly distributed random number - will use for transitions from
  #timestep k to current timestep k+1
  rd <- runif(1)

  if (rd < A[1,states[k]])
    states[k+1] <- 1
  else
    states[k+1] <- 2
}

plot(1:numsteps, states, xlab = "timestep", ylab= "state")