mdp

module M
  s: [0..2] init 0;

  [a] s=0 -> 1/3:(s'=0) + 1/3:(s'=1) + 1/3:(s'=2);
  [b] s=0 -> 1:(s'=2);

  [a] s=1 -> 1:(s'=1);
  [a] s=2 -> 1:(s'=2);
endmodule

rewards "steps"
  true : 1;
endrewards

rewards "s0"
  s=0 : 1;
endrewards

rewards "s0_a"
  [a] s=0 : 2;
endrewards

rewards "s0_b"
  [b] s=0 : 1/100000;
endrewards

rewards "s1"
  s=1 : 1;
endrewards