def get_speculative_worlds(self, takenMoves):
potentialMovesByOthers = {}
tempKb = self.copy()
for i in takenMoves.keys():
if i == self._player:
tempKb += Term('does', i, takenMoves[i].args[1])
potentialMovesByOthers[i] = {
Term('thinks_move', self._player,
Term('does', i, takenMoves[i].args[1])): 1
}
else:
potentialMovesByOthers[i] = self.query(\
[Term('thinks_move', self._player, \
Term('does', i, Var('_')))], kb=tempKb, prob_override=1)
potentialMoveSequences = \
self._processMoveSequences(\
itertools.product(*[[(key, val) for (key, val) in potentialMovesByOthers[key].items()] \
for key in potentialMovesByOthers.keys()]))
successorWorlds = []
for moves, seqProb in potentialMoveSequences:
copyKb = tempKb.extend()
for m in moves:
copyKb += m
nextState = map(lambda a: Term('ptrue', a.args[0]) if a.args[0].functor != "thinks" else a.args[0],\
self.query([Term('next', Var('_'))], kb=copyKb).keys())
potentialWorld = World(self._engine, self._baseModel, self._step + 1,\
seqProb*self._prob, set(nextState), self._player,None,)
successorWorlds.append(potentialWorld)
return set(successorWorlds)
def foreign_keys(a, b):
args = list(a.args)
bargs = list(b.args)
lgg_args = [None] * len(args)
lgg_bargs = [None] * len(bargs)
x = 0
for i in range(len(args)):
match = False
for j in range(len(bargs)):
if not isinstance(bargs[j], Var):
if args[i] == bargs[j]:
v = Var("X" + str(x))
x += 1
lgg_args[i] = v
lgg_bargs[j] = v
match = True
break
if not match:
lgg_args[i] = Var("X" + str(x))
x += 1
for i in range(len(lgg_bargs)):
if lgg_bargs[i] is None:
lgg_bargs[i] = Var("X" + str(x))
x += 1
return Term("join", a.with_args(*lgg_args), b.with_args(*lgg_bargs))
def main():
mod = GDLIIIEngine('./examples/guess.gdliii', File_Format.PREFIX)
res = {}
for i in range(1, 11):
res[i] = 0
res['lost'] = 0
#Choose a move
for i in range(0, 1000):
guesses = 0
mov = mod.get_legal_moves()
mod.set_actions({'player': mov['player'][0]})
mod.update_step()
while not mod.is_terminal():
mov = mod.get_legal_moves()
pmoves = sorted(mov[PLAYER_NAME],
key=lambda a: int(str(a.args[0])))
possible_nums = sorted(
[a[1] for a in mod.query('player', Term('secret', Var('_')))],
key=lambda a: int(str(a.args[0])))
num = int(str(possible_nums[round(
len(possible_nums) / 2)].args[0])) - 1
mod.set_actions({'player': pmoves[num]})
mod.update_step()
guesses += 1
result = mod.query('player', Term('secret', Var('_')))
if len(result) == 1:
res[guesses] += 1
else:
res['lost'] += 1
mod.undo_step(guesses + 1)
print(f'Game {i} Complete')
print(res)
def __getitem__(self, name):
if name == '#':
name = self._get_name(self.count)
self.count += 1
return Var(name)
else:
return Var(name)
def getSuccessorWorlds(self, takenMoves):
potentialMovesByOthers = {}
tempKb = self.copy()
for i in takenMoves.keys():
tempKb += Term('does', i, takenMoves[i].args[1])
if i == self._player:
potentialMovesByOthers[i] = {
Term('thinks_move', self._player,
Term('does', i, takenMoves[i].args[1])): 1
}
else:
potentialMovesByOthers[i] = self.query(\
[Term('thinks_move', self._player, \
Term('does', i, Var('_')))], kb=tempKb)
potentialMoveSequences = \
self._processMoveSequences(\
itertools.product(*[[(key, val) for (key, val) in potentialMovesByOthers[key].items()] \
for key in potentialMovesByOthers.keys()]))
tkquery = self.query([Term('sees', self._player, Var('_'))], kb=tempKb).keys()
tokens = list(map(lambda a: str(a.args[1]),tkquery))
tokens.sort()
tkkey = "".join(tokens)
worlds = {}
for moves, seqProb in potentialMoveSequences:
copyKb = tempKb.extend()
for m in moves:
copyKb += m
potentialTokens = set(map(lambda a: a.args[1], self.query([Term('thinks', self._player, \
Term('sees', self._player, Var('_')))],kb=copyKb).keys()))
move_id = list(map(lambda a: str(a.args[1]),moves))
move_id.sort()
move_id = "".join(move_id)
ptklist = list(map(lambda a: str(a.args[1]),potentialTokens))
ptklist.sort()
ptkkey = "".join(ptklist)
nextState = map(lambda a: Term('ptrue', a.args[0]) if a.args[0].functor != "thinks" else a.args[0],\
self.query([Term('next', Var('_'))], kb=copyKb).keys())
move_id = list(map(lambda a: str(a.args[1]),moves))
move_id.sort()
move_id = "".join(move_id)
ptklist = list(map(lambda a: str(a.args[1]),potentialTokens))
ptklist.sort()
ptkkey = "".join(ptklist)
potentialWorld = World(self._engine, self._baseModel, self._step + 1,\
seqProb*self._prob, set(nextState), self._player, ptkkey, move_id)
if ptkkey not in worlds.keys():
worlds[ptkkey] = set()
worlds[ptkkey].add(potentialWorld)
return (worlds, tkkey)
def load(self, data):
"""Load the settings from a data file.
Initializes language, target and examples.
:param data: data file
:type data: DataFile
"""
self.language.load(data) # for types and modes
if self._target is None:
try:
target = data.query('learn', 1)[0]
target_functor, target_arity = target[0].args
except IndexError:
raise KnownError('No target specification found!')
else:
target_functor, target_arity = self._target.functor, self._target.args[
0]
target_arguments = [
Var(chr(65 + i)) for i in range(0, int(target_arity))
]
self._target = Term(str(target_functor), *target_arguments)
# Find examples:
# if example_mode is closed, we will only use examples that are defined in the data
# this includes facts labeled with probability 0.0 (i.e. negative example)
# otherwise, the examples will consist of all combinations of values appearing in the data
# (taking into account type information)
example_mode = data.query(Term('example_mode'), 1)
if example_mode and str(example_mode[0][0]) == 'auto':
types = self.language.get_argument_types(self._target.functor,
self._target.arity)
values = [self.language.get_type_values(t) for t in types]
self._examples = list(product(*values))
elif example_mode and str(example_mode[0][0]) == 'balance':
# Balancing based on count only
pos_examples = [
r for r in data.query(self._target.functor, self._target.arity)
]
pos_count = len(pos_examples) # get no. of positive examples
types = self.language.get_argument_types(self._target.functor,
self._target.arity)
values = [self.language.get_type_values(t) for t in types]
from random import shuffle
neg_examples = list(product(*values))
shuffle(neg_examples)
logger = logging.getLogger(self._logger)
logger.debug('Generated negative examples:')
for ex in neg_examples[:pos_count]:
logger.debug(Term(self._target(*ex).with_probability(0.0)))
self._examples = pos_examples + neg_examples[:pos_count]
else:
self._examples = [
r for r in data.query(self._target.functor, self._target.arity)
]
with Timer('Computing scores', logger=self._logger):
self._scores_correct = self._compute_scores_correct()
def get_array_term(name, index, array):
return Term(
name, Constant(index), *[
Constant(array[j]) if array[j] is not None else Var("X" +
str(j))
for j in range(len(array))
])
def _generate_legal_moves(self):
legalMoves = {}
for player in self.worlds.keys():
legalMoves[player] = {}
for world in self.worlds[player]:
legalMoves[player] = world.query([Term('legal', player, Var('_'))])
break
return legalMoves
def __get_possible_term_arguments_for(self, functor, argument_mode_indicators, variables_in_query_by_type: Dict[TypeName, List[Term]]):
"""
Returns the possible arguments of the term with the given functor and as arity the length of the given list of
mode indicators.
These arguments are generated using the modes of the term, unifying if necessary with the variables of
the same type already in the query.
:param functor:
:param argument_mode_indicators:
:param variables_in_query_by_type:
:return:
"""
# INPUT e.g. 'parent', ['+', '+'], {person: A}
# we will collect the possible arguments for the current functor in a list
arguments = []
arity = len(argument_mode_indicators)
# get the types of the arguments of the given predicate
argument_types = self.get_argument_types(functor, arity) # type: TypeArguments
# e.g parent/2 has argument types ['person', 'person']
# argument types are necessary for unification
# a functor has multiple variables,
# each with their own modes.
# We have to consider each variable with its mode separately
# There are three different modes for an argument:
# '+': the variable must be unified with an existing variable
# --> use an existing variable
# '-': create a new variable, do not reuse an old one
# 'c': insert a constant
# NEW: '+-': the variable can be a new variable, OR it can be unified with an existing on
# for each argument of functor:
# check its mode
# '+' --> add to arguments
# a list of the variables in the rule with the same type as the current argument
# e.g. arguments = [ ..., [X,Y], ...]
# '-' --> add to arguments
# a list containing 1 new Var #
# 'c' --> add to arguments
# a list of all possible constants for the type of the argument
for index, (argmode, argtype) in enumerate(zip(argument_mode_indicators, argument_types)):
if argmode == '+':
# All possible variables of the given type
#TODO: this is were key output file needs to be adapted
arguments.append(variables_in_query_by_type.get(argtype, []))
elif argmode == '-':
# A new variable
arguments.append([Var('#')]) # what about adding a term a(X,X) where X is new?
elif argmode == 'c':
# Add a constant
type = functor + '_' + str(index)
arguments.append(self.get_type_values(type))
# arguments.append(self.get_type_values(argtype))
# pass
else:
raise ValueError("Unknown mode specifier '%s'" % argmode)
return arguments
def test_test(self):
model_string = "nn(fc,[X,Y],Z,[0,1]) :: a(X,Y,Z).\nb(X,Y,Z) :- a(X,Y,Z)."
fc = FC(10, 2)
net = Network(fc, 'fc',
lambda a, b, c: Variable(torch.FloatTensor([0.2, 0.8])))
model = Model(model_string, [net])
query = Term('b', Constant("string with 's"), Constant(3), Var('X'))
solution = model.solve(query, test=True)
print(solution)
def getSuccessorWorlds(self, takenMoves):
copyKb = self.copy()
for i in takenMoves.keys():
copyKb += Term('does', i, takenMoves[i].args[1])
#Assumption: random player never has any thinks predicates
nextState = map(lambda a: Term('ptrue', a.args[0]),\
self.query([Term('next', Var('_'))], kb=copyKb).keys())
return set([RandomWorld(self._engine, self._baseModel, self._step + 1,\
1, set(nextState), self._player, [])])
def get_prediction_goal(self) -> Term:
prediction_goal_term = Term(self.functor)
nb_of_args = len(self.modes)
arguments = [Var('#')] * nb_of_args
prediction_goal_term = prediction_goal_term(*arguments)
prediction_goal_term = prediction_goal_term.apply(TypeModeLanguage.ReplaceNew(0))
return prediction_goal_term
def main_loop(model):
while(not model.terminal):
playerMoves = tuple(model.getLegalMovesForPlayer(Constant(1)))
randomMoves = tuple(model.getLegalMovesForPlayer(Constant(0)))
model.submitAction(choice(playerMoves), Constant(1))
model.submitAction(choice(randomMoves), Constant(0))
model.applyActionsToModelAndUpdate()
#For Monty Hall
#print(model.query(Constant(1), Term('car', Var('_'))))
#For guessing game
print(model.query(Constant(1), Term('num', Var('_'))))
def apply_substitution_to_term(term: Term, substitution: Dict[str,
Term]) -> Term:
complete_substitution = {}
# NOTE: all variables in the term need to be defined in the substitution
for var in term.variables():
complete_substitution[var.name] = Var(var.name)
complete_substitution.update(substitution)
term_substitution = term.apply(complete_substitution)
return term_substitution
def get_knows_for_worlds(self, worlds):
preds = set()
for w in worlds:
dis_set = set()
for p in w.query([Term('thinks', w._player, Var('_'))],prob_override=1):
if p.functor == "thinks":
tp = p.args[1]
if tp.functor != "knows":
dis_set.add(tp)
if len(preds) == 0:
preds = dis_set
else:
preds &= dis_set
return preds
def generate_speculative_worlds(self, player, actions):
knowsSet = set()
total_new_worlds = {}
newWorlds = set()
for world in self.worlds[player]:
(wd, truekey) = world.getSuccessorWorlds(actions)
newWorlds = newWorlds.union(wd[truekey])
total_new_worlds[player] = self._normaliseProbability(newWorlds)
#Find knowledge that is known across every state, this will derive the knows predicate
res = [k for (k,v) in self.raw_query(player, Term('thinks', player, Var('_')), total_new_worlds).items() if v == 1]
for p in res:
t = Term('knows', p.args[0], p.args[1])
knowsSet.add(t)
knowsSet.add(Term('thinks', player, t))
for w in total_new_worlds[player]:
for p in knowsSet:
w._preds.add(p)
w._kb += p
return GDLNode(total_new_worlds, self.game_data, self)
def __init__(self, data, language, target=None, logger=None, **kwargs):
self._language = language
if target is not None:
try:
t_func, t_arity = target.split('/')
arguments = []
i = 0
while i < int(t_arity):
arguments.append(Var(chr(65+i)))
i += 1
target = Term(t_func, *arguments)
except Exception:
raise KnownError('Invalid target specification \'%s\'' % target)
self._target = target
self._examples = None
self._logger = logger
self._data = data
self._scores_correct = None
from problog.logic import Constant, Var, Term, AnnotatedDisjunction
from problog import get_evaluatable
true = Term('true')
ok = Term('ok')
query = Term('query')
p = PrologFile("Try.pl")
x = [p]
for i in range(0, 2):
s = SimpleProgram()
brand = Term('brand')
ind = Term('ind')
prop = Term('prop')
query = Term('query')
X = Var('X')
ford = Term('ford')
seat = Term('seat')
suzuki = Term('suzuki')
audi = Term('audi')
fiat = Term('fiat')
s = SimpleProgram()
s += AnnotatedDisjunction([
prop(ind, brand, ford, p=0.222222222222222),
prop(ind, brand, seat, p=0.222222222222222),
prop(ind, brand, suzuki, p=0.222222222222222),
prop(ind, brand, audi, p=0.222222222222222),
prop(ind, brand, fiat, p=0.111111111111111)
], true)
if i == 1:
q = PrologFile("TryQuery.pl")
def __getitem__(self, key):
return self.base.get(key, Var(key))
def __getitem__(self, key):
if key == '_':
self.cnt += 1
return Var('anon_%s' % self.cnt)
else:
return Var(key)
"""
Definition of Terms, Constants and Vars used in the machine parts examples.
"""
from problog.logic import Term, Constant, Var
# defining the terms
sendback, fix, ok = \
Term('sendback'), Term('fix'), Term('ok')
worn, replaceable, not_replaceable = \
Term('worn'), Term('replaceable'), Term('not_replaceable')
p0, p1, class_ = \
Term('p0'), Term('p1'), Term('class')
# defining the constants
gear, engine, chain, wheel, control_unit = \
Constant('gear'), Constant('engine'), Constant('chain'), Constant('wheel'), Constant('control_unit')
# defining the vars
X = Var('X')
Y = Var('Y')
Z = Var('Z')
def query_transformation_digit(query):
args0, _ = query.args
return query(args0, Var('X'))
self._gdl_rep._randomIdentifier] is None:
self._gdl_rep.submitAction(
choice(
tuple(
self._gdl_rep.getLegalMovesForPlayer(
self._gdl_rep._randomIdentifier))),
self._gdl_rep._randomIdentifier)
self._gdl_rep.applyActionsToModelAndUpdate()
self._cur_step += 1
def is_terminal(self):
return self._gdl_rep.terminal
if __name__ == "__main__":
mod = GDLIIIEngine('./examples/montyhall.gdliii', File_Format.INFIX)
print(mod.player_names)
print(mod.get_legal_moves())
mod.set_actions({'candidate': mod.get_legal_moves('candidate')[0]})
mod.update_step()
print(mod.get_legal_moves())
mod.set_actions({'candidate': mod.get_legal_moves('candidate')[0]})
mod.update_step()
print(mod.is_terminal())
print(mod.query('candidate', Term('car', Var("_")), 0))
print(mod.query('candidate', Term('car', Var("_")), 1))
mod.set_actions({'candidate': mod.get_legal_moves('candidate')[0]})
mod.update_step()
print(mod.is_terminal())
print(mod.query('candidate', Term('car', Var("_"))))
def query_transformation_happens_at(query):
args0, _, args2 = query.args
return query(args0, Var('X'), args2)
def query_transformation_initiated_at(query):
args0, args1 = query.args
return query(args0(Var('X'), Var('Y')), args1)
def getLegalMoves(self, playerOfLegalMove):
return set(map(lambda a: a.args[2],
self.query([Term('thinks', self._player, Term('legal', playerOfLegalMove, Var('_')))]).keys()\
))
ancestor(X,Y) :- parent(X,Y).
ancestor(X,Y) :- ancestor(X,Z), parent(Z,Y).
sibling(X,Y) :- parent(Z,X),parent(Z,Y), not(X=Y).
cousin(X,Y) :- grandchild(X,Z), grandchild(Y,Z).
uncle(X,Y) :- sibling(X,Z),parent(Z,Y),male(X).
uncle(X,Y) :- married(X,Z),sibling(Z,P),parent(P,Y),male(X).
aunt(X,Y) :- sibling(X,Z),parent(Z,Y),female(X).
aunt(X,Y) :- married(X,Z),sibling(Z,P),parent(P,Y),female(X).
dad(X,Y) :- parent(X,Y), male(X).
mom(X,Y) :- parent(X,Y), female(X).
query(parent(X,giacomo)).
""")
engine = DefaultEngine()
db = engine.prepare(
p) # This compiles the Prolog model into an internal format.
# This step is optional, but it might be worthwhile if you
# want to query the same model multiple times.
query1 = Term('parent', [Var('X'), Constant('giacomo')])
results = engine.query(db, query1)
r = get_evaluatable().create_from(p).evaluate()
print(results)
print(r)
# -*- coding: utf-8 -*-
"""
Created on Fri Jan 31 14:04:54 2020
@author: PasqualeDeMarinis
"""
from problog.program import SimpleProgram
from problog.logic import Constant, Var, Term, AnnotatedDisjunction
from problog import get_evaluatable
coin, heads, tails, win, query = Term('coin'), Term('heads'), Term(
'tails'), Term('win'), Term('query')
C = Var('C')
p = SimpleProgram()
p += coin(Constant('c1'))
p += coin(Constant('c2'))
p += AnnotatedDisjunction([heads(C, p=0.4), tails(C, p=0.6)], coin(C))
p += (win << heads(C))
p += query(coin(C))
r = get_evaluatable().create_from(p).evaluate()
print(r)
def _generate_successor_worlds(self, actions):
total_new_worlds = {}
players_less_random = [_ for _ in self.game_data.players if _ != self.game_data.random_id]
all_worlds = {}
knowsSet = set()
for i in players_less_random:
all_worlds[i] = {}
newWorlds = set()
truekey = None
for world in self.worlds[i]:
(wd,truekey) = world.getSuccessorWorlds(actions)
try:
newWorlds = newWorlds.union(wd[truekey])
except:
pass
for m in wd.keys():
if m not in all_worlds[i].keys():
all_worlds[i][m] = set()
all_worlds[i][m] = all_worlds[i][m].union(wd[m])
total_new_worlds[i] = self._normaliseProbability(newWorlds)
#Find knowledge that is known across every state, this will derive the knows predicate
res = [k for (k,v) in self.raw_query(i, Term('thinks', i, Var('_')), total_new_worlds).items() if v == 1 and k.args[1].functor != "legal"]
for p in res:
t = Term('knows', p.args[0], p.args[1])
knowsSet.add(t)
for w in total_new_worlds[i]:
for p in knowsSet:
w._preds.add(Term('thinks', i, p))
w._kb += Term('thinks', i, p)
for i in players_less_random:
for w in total_new_worlds[i]:
for p in knowsSet:
w._preds.add(p)
w._kb += p
for i in players_less_random:
#Get knowledge in every other state
for m in all_worlds[i].keys():
if m != truekey:
knows = self.get_knows_for_worlds(all_worlds[i][m])
for p in knows:
for wm in all_worlds[i][m]:
wm._preds.add(Term('knows', i,p))
#Generate Inferred knowledge
kvworlds = {}
for i in players_less_random:
for w in total_new_worlds[i]:
kvworlds[w] = [Term('thinks',i,p) for p in self.generate_inferred_knowledge(w,all_worlds)]
for (w,k) in kvworlds.items():
for p in k:
w._preds.add(p)
w._kb += p
knowsSet2 = set()
#Generate new legal moves
for i in players_less_random:
res = [k for (k,v) in self.raw_query(i, Term('thinks', i, Var('_')), total_new_worlds).items() if v == 1 and k.args[1].functor == "legal"]
for q in res:
knowsSet2.add(Term('knows', i, q.args[1]))
for i in players_less_random:
for w in total_new_worlds[i]:
for p in knowsSet2:
w._preds.add(p)
w._kb += Term(p)
#Extra case, should only be one world in the random players collection
if self.game_data.random_id in self.worlds.keys():
for w in self.worlds[self.game_data.random_id]:
total_new_worlds[self.game_data.random_id] = set(w.getSuccessorWorlds(actions))
for w1 in total_new_worlds[self.game_data.random_id]:
for p1 in knowsSet:
w1._preds.add(p1)
w1._kb += p1
for p2 in knowsSet2:
w1._preds.add(p2)
w1._kb += p2
break
return GDLNode(total_new_worlds, self.game_data, self)
def _initialiseKB(self):
self._kb = self._engine.prepare(self._baseModelFile)
initialState = \
set(map(lambda a: Term('ptrue', a[0].args[0]), self._engine.ground_all(self._kb, queries=[Term('init',Var('_'))]).get_names()))
players = \
set(map(lambda a: a[0], self._engine.ground_all(self._kb, queries=[Term('role',Var('_'))]).get_names()))
#Not needed, but dont care to remove right now
self._step = 0
playerWorldState = {}
for playerNum in map(lambda a: a.args[0], players):
knowledge = map(lambda a: Term('thinks', playerNum, a.args[0]), initialState)
playerPreds = initialState.union(set(knowledge))
self._playerList.append(playerNum)
#Each player starts with a single initial world
if playerNum == self._randomIdentifier:
#Random Specific world has no thinks predicates
playerWorldState[playerNum] = [RandomWorld(self._engine, self._baseModelFile, self._step, 1, initialState, playerNum)]
else:
playerWorldState[playerNum] = [World(self._engine, self._baseModelFile, self._step, 1, playerPreds, playerNum)]
return playerWorldState
