def parity(p):
if dlvhex.learnSupportSets():
pos = ()
for x in dlvhex.getInputAtoms():
pos = pos + (False, )
# special case: no input
if pos == ():
# always true
dlvhex.learn(dlvhex.storeOutputAtom(()).negate(), )
else:
pos = pos[:-1]
pos = list(pos)
overflow = False
while not overflow:
ng = ()
# enumerate all combinations except for the last element (which is then definite)
last = False
for i in range(0, len(pos)):
if pos[i] == True:
ng = ng + (dlvhex.getInputAtoms()[i], )
last = not last
else:
ng = ng + (dlvhex.getInputAtoms()[i].negate(), )
# add last element with a sign such that the partiy is even
if last:
ng = ng + (dlvhex.getInputAtoms()[-1], )
else:
ng = ng + (dlvhex.getInputAtoms()[-1].negate(), )
# generate nogood which implies that the external atom is true
supset = ng + (dlvhex.storeOutputAtom(()).negate(), )
dlvhex.learn(supset)
# go to next combination and check if we have an overflow, i.e., all combinations have been enumerated
inc=0
pos[inc] = not pos[inc]
while not overflow and not pos[inc]:
inc = inc + 1
if inc >= len(pos):
overflow = True
else:
pos[inc] = not pos[inc]
even = True
for atom in dlvhex.getInputAtoms():
if atom.isTrue():
even = not even
if even:
dlvhex.output(())
def parity(p):
if dlvhex.learnSupportSets():
pos = ()
for x in dlvhex.getInputAtoms():
pos = pos + (False, )
# special case: no input
if pos == ():
# always true
dlvhex.learn(dlvhex.storeOutputAtom(()).negate(), )
else:
pos = pos[:-1]
pos = list(pos)
overflow = False
while not overflow:
ng = ()
# enumerate all combinations except for the last element (which is then definite)
last = False
for i in range(0, len(pos)):
if pos[i] == True:
ng = ng + (dlvhex.getInputAtoms()[i], )
last = not last
else:
ng = ng + (dlvhex.getInputAtoms()[i].negate(), )
# add last element with a sign such that the partiy is even
if last:
ng = ng + (dlvhex.getInputAtoms()[-1], )
else:
ng = ng + (dlvhex.getInputAtoms()[-1].negate(), )
# generate nogood which implies that the external atom is true
supset = ng + (dlvhex.storeOutputAtom(()).negate(), )
dlvhex.learn(supset)
# go to next combination and check if we have an overflow, i.e., all combinations have been enumerated
inc=0
pos[inc] = not pos[inc]
while not overflow and not pos[inc]:
inc = inc + 1
if inc >= len(pos):
overflow = True
else:
pos[inc] = not pos[inc]
even = True
for atom in dlvhex.getInputAtoms():
if atom.isTrue():
even = not even
if even:
dlvhex.output(())
def testSetMinus(p, q): premisse = () outputatoms = () input = dlvhex.getInputAtoms() for x in input: tup = x.tuple() if tup[0].value() == p.value(): # keep true monotonic input atoms if dlvhex.isTrue(x): premisse = (x, ) + premisse if x.isTrue() and not dlvhex.isTrue(dlvhex.storeAtom((q, tup[1]))): outputatoms = (dlvhex.storeOutputAtom((tup[1], )), ) + outputatoms dlvhex.output((tup[1], )) if tup[0].value() == q.value(): # keep false antimonotonic input atoms if not dlvhex.isTrue(x): premisse = (x.negate(), ) + premisse # learn one nogood for each output atom for x in outputatoms: dlvhex.learn((x.negate(), ) + premisse)
def aOrNotB(a,b): if dlvhex.learnSupportSets(): dlvhex.learn(( dlvhex.storeAtom((a, )), dlvhex.storeOutputAtom(()).negate() # then () is in the output )); dlvhex.learn(( dlvhex.storeAtom((b, )).negate(), dlvhex.storeOutputAtom(()).negate() # then () is in the output )); else: aIsTrue = dlvhex.isTrue(dlvhex.storeAtom((a, ))) bIsFalse = dlvhex.isFalse(dlvhex.storeAtom((b, ))) if aIsTrue or bIsFalse: dlvhex.output(())
def testSetMinus(p, q): premisse = () outputatoms = () input = dlvhex.getInputAtoms() for x in input: tup = x.tuple() if tup[0].value() == p.value(): # keep true monotonic input atoms if dlvhex.isTrue(x): premisse = (x, ) + premisse if not dlvhex.isTrue(dlvhex.storeAtom((q, tup[1]))): outputatoms = (dlvhex.storeOutputAtom((tup[1], )), ) + outputatoms dlvhex.output((tup[1], )) if tup[0].value() == q.value(): # keep false antimonotonic input atoms if not dlvhex.isTrue(x): premisse = (x.negate(), ) + premisse # learn one nogood for each output atom for x in outputatoms: dlvhex.learn((x.negate(), ) + premisse)
def testSetMinus2(p, q):
for x in p.extension():
if not x in q.extension():
dlvhex.learn((dlvhex.storeAtom((p, ) + x),
dlvhex.storeAtom((q, ) + x).negate(),
dlvhex.storeOutputAtom(x).negate()))
dlvhex.output(x)
def aOrNotB(a,b): if dlvhex.learnSupportSets(): dlvhex.learn(( dlvhex.storeAtom((a, )), dlvhex.storeOutputAtom(()).negate() # then () is in the output )); dlvhex.learn(( dlvhex.storeAtom((b, )).negate(), dlvhex.storeOutputAtom(()).negate() # then () is in the output )); else: aIsTrue = dlvhex.isTrue(dlvhex.storeAtom((a, ))) bIsFalse = dlvhex.isFalse(dlvhex.storeAtom((b, ))) if aIsTrue or bIsFalse: dlvhex.output(())
def testSetMinus2(p, q): for x in p.extension(): if not x in q.extension(): dlvhex.learn(( dlvhex.storeAtom((p, ) + x), dlvhex.storeAtom((q, ) + x).negate(), dlvhex.storeOutputAtom(x).negate() )) dlvhex.output(x)
def storeOutputAtom(self, otuple):
# all the otuple elements are ISymbol s
#logging.info("jsci.storeOutputAtom %s", otuple)
s = dlvhex.storeOutputAtom([x.hid for x in otuple])
#logging.info(" got symbol %s", s)
r = jpype.JObject(JavaSymbolImpl(s), ISymbol)
#logging.info("jsci.storeOutputAtom %s returns %s with type %s", otuple, repr(r), type(r))
return r
def neg(p):
if dlvhex.learnSupportSets():
dlvhex.learn((
dlvhex.storeAtom((p, )).negate(), # if p is true
dlvhex.storeOutputAtom(()).negate() # then () is in the output
));
else:
for x in dlvhex.getTrueInputAtoms():
return
dlvhex.output(())
def id(p): if dlvhex.learnSupportSets(): dlvhex.learn(( dlvhex.storeAtom((p, )), # if p is true for some X dlvhex.storeOutputAtom(()).negate() # then () is in the output )); else: for x in dlvhex.getTrueInputAtoms(): dlvhex.output(()) return
def testSetMinusLearn(p, q):
# is true for all constants in extension of p but not in extension of q
# (same as testSetMinus)
# uses learning
pe = p.extension()
#logging.error("ATOM got pe {}".format(repr(pe)))
#for y in pe:
# logging.error("ATOM y in pe {} {} {}".format(str(y), repr(y[0].symlit.lit), hash(y)))
qe = q.extension()
#logging.error("ATOM got qe {}".format(repr(qe)))
#for y in qe:
# logging.error("ATOM y in qe {} {} {}".format(str(y), repr(y[0].symlit.lit), hash(y)))
for x in pe:
#logging.error("ATOM x = {} {}".format(repr(x), x.__class__))
if x not in qe:
#logging.error("ATOM {} not in qe".format(x))
# learn that it is not allowed that p(x) and -q(x) and this atom is false for x
nogood = (dlvhex.storeAtom((p, ) + x),
dlvhex.storeAtom((q, ) + x).negate(),
dlvhex.storeOutputAtom(x).negate())
#logging.error("ATOM nogood {}".format(repr(nogood)))
dlvhex.learn(nogood)
#logging.error("ATOM output {}".format(repr(x)))
dlvhex.output(x)
def someSelectedLearning(selected):
for x in dlvhex.getInputAtoms():
if x.tuple()[0] == selected and x.isTrue():
dlvhex.output(())
nogood = [x, dlvhex.storeOutputAtom(()).negate()]
dlvhex.learn(nogood)
