Computing the minimum or maximum probability of reaching some set of states in an SMG, i.e. PRISM properties of the form:
Quantitative properties only (i.e. the probability is not 0/1 for at least some states).
Model | Property | PRISM property & description |
Extra parameters |
Min/max | Notes |
avoid | exit |
<<p1>> Pmax=? [ F "at_exit" ] Maximise the probability of the intruder exiting |
- | max | |
avoid | find |
<<p1>> Pmax=? [ F "found_item" ] Maximise the probability of the intruder finding the item |
- | max | |
dice | p1wins |
<<p1>> Pmax=? [ F "p1win" ] Maximise the probability of player 1 winning |
- | max | |
investors | greater |
<<investor1>> Pmax=? [ F ("done1"&v>5) ] Maximum probability with which investor 1 can guarantee a share value greater than 5 |
- | max | |
safe_nav | reach |
<<1>> Pmax=? [ !"crash" U ("robot_goal" | "human_goal" | dl) ] Maximum probability of reaching the other side without crashing |
- | max |
Checking whether the minimum or maximum probability of reaching some set of states in an SMG is 0 or 1, e.g. PRISM properties of the form:
Model | Property | PRISM property & description |
Extra parameters |
Min/max | Notes |
hallway_human | save |
<<p1>> P>=1 [ F "saved" ] Maximum probability of the agent saving the human is 1 |
- | max |
Computing the minimum or maximum expected accumulated reward/cost to reach some set of states in an SMG, i.e. PRISM properties of the form:
Model | Property | PRISM property & description |
Extra parameters |
Min/max | Notes |
task_graph | time |
<<sched>> R{"time"}min=?[ F "tasks_complete" ] Minimum expected time to complete all tasks |
- | min |