def brute_force(window: int, data: str):
"""Brute force O(m * n) solution runs in plenty of time."""
digits = get_data(data)
max_product = product(digits[0:window])
for x in range(1, len(digits) - window + 1):
max_product = max(max_product, product(digits[x:x + window]))
return max_product
def fire(self, new_atom, cond_index, enqueue_func):
if self.empty_atom_list_no:
return
# Binding: a (var_no, object) pair
# Bindings: List-of(Binding)
# BindingsFactor: List-of(Bindings)
# BindingsFactors: List-of(BindingsFactor)
bindings_factors = []
for pos, cond in enumerate(self.conditions):
if pos == cond_index:
continue
atoms = self.atoms_by_index[pos]
assert atoms, "if we have no atoms, this should never be called"
factor = [self._get_bindings(atom, cond) for atom in atoms]
bindings_factors.append(factor)
eff_args = self.prepare_effect(new_atom, cond_index)
for bindings_list in tools.product(*bindings_factors):
bindings = itertools.chain(*bindings_list)
for var_no, obj in bindings:
eff_args[var_no] = obj
enqueue_func(self.effect.predicate, eff_args)
def fire(self, new_atom, cond_index, enqueue_func):
if self.empty_atom_list_no:
return
# Binding: a (var_no, object) pair
# Bindings: List-of(Binding)
# BindingsFactor: List-of(Bindings)
# BindingsFactors: List-of(BindingsFactor)
bindings_factors = []
for pos, cond in enumerate(self.conditions):
if pos == cond_index:
continue
atoms = self.atoms_by_index[pos]
assert atoms, "if we have no atoms, this should never be called"
factor = [self._get_bindings(atom, cond) for atom in atoms]
bindings_factors.append(factor)
eff_args = self.prepare_effect(new_atom, cond_index)
for bindings_list in tools.product(*bindings_factors):
bindings = itertools.chain(*bindings_list)
for var_no, obj in bindings:
eff_args[var_no] = obj
enqueue_func(self.effect.predicate, eff_args)
def restriction(shape, dense=False):
"""
Create a nonsquare restriction matrix by taking every second point along
each axis in a 1D, 2D, or 3D domain.
Parameters
----------
shape : tuple of ints
The (rectangular) geometry of the problem.
Optional Parameters
-------------------
dense=False : bool
Whether to use and return a sparse (CSR) restriction matrix.
Returns
-------
R : ndarray
Array of shape (N/(2**alpha), N), where the int N is the number of
unknowns (product of shape ints), and alpha=len(shape).
Requiring the shape of the domain is bad for the black-boxibility.
Implementing a separate Ruge Steuben restriction method would avoid this.
Returns a CSR matrix by default, but will return dense if dense=True is given.
It would be nice to make this rely only on N, not shape. But, really, the
restriction operator is problem-dependent. There are certainly problem
domains where "dimensionality" means something other than spatial dimensions.
"""
# alpha is the dimensionality of the problem.
alpha = len(shape)
NX = shape[0]
if alpha >= 2:
NY = shape[1]
#NZ = shape[3] # don't need this
N = tools.product(shape)
n = N / (2**alpha)
if n in (0, 1):
raise ValueError('New restriction matrix would have shape ' +
str((n, N)) + '. ' +
'Coarse set would have %d point(s)! ' % n +
'Try a larger problem or fewer gridLevels.')
if dense:
R = np.zeros((n, N))
else:
R = scipy.sparse.lil_matrix((n, N))
# columns in the restriction matrix:
if alpha == 1:
coarseColumns = np.array(range(N)).reshape(shape)[::2].ravel()
elif alpha == 2:
coarseColumns = np.array(range(N)).reshape(shape)[::2, ::2].ravel()
elif alpha == 3:
coarseColumns = np.array(
range(N)).reshape(shape)[::2, ::2, ::2].ravel()
else:
raise ValueError("restriction(): Greater than 3 dimensions is not"
"implemented. (shape was" + str(shape) + " .)")
each = 1.0 / (2**alpha)
for r, c in zip(range(n), coarseColumns):
R[r, c] = each
R[r, c + 1] = each
if alpha >= 2:
R[r, c + NX] = each
R[r, c + NX + 1] = each
if alpha == 3:
R[r, c + NX * NY] = each
R[r, c + NX * NY + 1] = each
R[r, c + NX * NY + NX] = each
R[r, c + NX * NY + NX + 1] = each
if dense:
return R
else:
return R.tocsr()
def restriction(shape, dense=False):
"""
Create a nonsquare restriction matrix by taking every second point along
each axis in a 1D, 2D, or 3D domain.
Parameters
----------
shape : tuple of ints
The (rectangular) geometry of the problem.
Optional Parameters
-------------------
dense=False : bool
Whether to use and return a sparse (CSR) restriction matrix.
Returns
-------
R : ndarray
Array of shape (N/(2**alpha), N), where the int N is the number of
unknowns (product of shape ints), and alpha=len(shape).
Requiring the shape of the domain is bad for the black-boxibility.
Implementing a separate Ruge Steuben restriction method would avoid this.
Returns a CSR matrix by default, but will return dense if dense=True is given.
It would be nice to make this rely only on N, not shape. But, really, the
restriction operator is problem-dependent. There are certainly problem
domains where "dimensionality" means something other than spatial dimensions.
"""
# alpha is the dimensionality of the problem.
alpha = len(shape)
NX = shape[0]
if alpha >= 2:
NY = shape[1]
#NZ = shape[3] # don't need this
N = tools.product(shape)
n = N / (2 ** alpha)
if n in (0, 1):
raise ValueError('New restriction matrix would have shape ' + str((n,N))
+ '. ' + 'Coarse set would have %d point(s)! ' % n+
'Try a larger problem or fewer gridLevels.')
if dense:
R = np.zeros((n, N))
else:
R = scipy.sparse.lil_matrix((n, N))
# columns in the restriction matrix:
if alpha == 1:
coarseColumns = np.array(range(N)).reshape(shape)[::2].ravel()
elif alpha == 2:
coarseColumns = np.array(range(N)).reshape(shape)[::2, ::2].ravel()
elif alpha == 3:
coarseColumns = np.array(range(N)).reshape(shape)[::2, ::2, ::2].ravel()
else:
raise ValueError("restriction(): Greater than 3 dimensions is not"
"implemented. (shape was" + str(shape) +" .)")
each = 1.0 / (2 ** alpha)
for r, c in zip(range(n), coarseColumns):
R[r, c] = each
R[r, c + 1] = each
if alpha >= 2:
R[r, c + NX] = each
R[r, c + NX + 1] = each
if alpha == 3:
R[r, c + NX * NY] = each
R[r, c + NX * NY + 1] = each
R[r, c + NX * NY + NX] = each
R[r, c + NX * NY + NX + 1] = each
if dense:
return R
else:
return R.tocsr()
def test_product_4():
gen = tools.product([3, 4], [0, 1, 2])
assert [next(gen) for _ in range(5)] == [(3, 0), (3, 1), (3, 2), (4, 0), (4, 1)]
def test_product_3():
gen = tools.product('123', [1, 2])
assert [next(gen) for _ in range(5)] == [('1', 1), ('1', 2), ('2', 1), ('2', 2), ('3', 1)]
def test_product_2():
assert next(tools.product('ABC', [1, 2])) == ('A', 1)
def test_product_1():
assert list(tools.product('AB', [1, 2])) == [
('A', 1), ('A', 2), ('B', 1), ('B', 2)]
