def parse(self):
tree = ET.parse(self.file)
root = tree.getroot()
actiondict = {}
statedict = {}
for st in root.iter('location'):
st_id = st.find('id')
state = State(st_id)
statedict[st_id] = state
for a in root.iter('branchpoint'):
a_id = a.find('id')
action = Action(a_id)
actiondict[a_id] = action
for t in root.iter('transition'):
pass
actionlist = list(actiondict.keys())
statelist = list(statedict.keys())
return statelist, actionlist
from MDP import State, Action, Transition
import ValueIter
y = 0.2
e = 0.01
reward = 15
#Add states
states = {
"A": State("A"),
"B": State("B"),
"C": State("C"),
"R": State("R", reward, 1)
}
#Add actions
states["A"].add_action(Action("A", "rt"))
states["A"].add_action(Action("A", "up"))
states["B"].add_action(Action("B", "lt"))
states["B"].add_action(Action("B", "up"))
states["C"].add_action(Action("C", "rt"))
states["C"].add_action(Action("C", "dn"))
#Add transitions A
states["A"].actions["rt"].add_transition("B", 0.8, -1)
states["A"].actions["rt"].add_transition("A", 0.2, -1)
states["A"].actions["up"].add_transition("C", 0.8, -1)
states["A"].actions["up"].add_transition("A", 0.2, -1)
#Add transitions B
states["B"].actions["lt"].add_transition("A", 0.8, -1)
states["B"].actions["lt"].add_transition("B", 0.2, -1)
ref[state] = idx
idx += 1
# Add actions
for p in POSITIONS:
for m in MATERIALS:
for a in ARROWS:
for mm in MM:
for h in HEALTH:
cur = (p, m, a, mm, h)
now = ref[cur]
if h == 0:
states[now].add_action(Action(now, "NONE"))
continue
states[now].add_action(Action(now, "STAY"))
if p == "W":
states[now].add_action(Action(now, "RIGHT"))
if a > 0:
states[now].add_action(Action(now, "SHOOT"))
elif p == "E":
states[now].add_action(Action(now, "LEFT"))
states[now].add_action(Action(now, "HIT"))
if a > 0:
states[now].add_action(Action(now, "SHOOT"))
ref[idx] = state
idx += 1
# Add actions
for p in POSITIONS:
for m in MATERIALS:
for a in ARROWS:
for mm in MM:
for h in HEALTH:
if h == 0:
continue
cur = (p, m, a, mm, h)
now = ref[cur]
states[now].add_action(Action(now, "STAY"))
if p == "W":
states[now].add_action(Action(now, "RIGHT"))
if a > 0:
states[now].add_action(Action(now, "SHOOT"))
elif p == "E":
states[now].add_action(Action(now, "LEFT"))
states[now].add_action(Action(now, "HIT"))
if a > 0:
states[now].add_action(Action(now, "SHOOT"))
elif p == "S":
states[now].add_action(Action(now, "UP"))
# ------------
# ---- States ----
states_list = [State(1, 5.), State(2, 10), State(3, -7.2)]
# ---- MDP ----
disc_fact = 0.5
chain = MDP(states_list, disc_fact)
# ---- Actions ----
# ----
transitions_a = np.array([[0.3, 0.6, 0.1], [0.1, 0.6, 0.3],
[0.5, 0.0, 0.5]])
reward_a = np.array([5.0, 5.0, 1.0])
a = Action("a", chain, transitions_a, reward_a)
# ----
# ----
transitions_b = np.array([[0.0, 0.3, 0.7], [1 / 3, 1 / 3, 1 / 3],
[0.0, 0.5, 0.5]])
reward_b = np.array([2.8, 0.0, 10.0])
b = Action("b", chain, transitions_b, reward_b)
# ----
# ----
transitions_c = np.array([[0.2, 0.4, 0.4], [0.2, 0.6, 0.2],
[1 / 3, 1 / 3, 1 / 3]])
reward_c = np.array([-7.2, -7.2, 0.0])
c = Action("c", chain, transitions_c, reward_c)
# ----