def test_column_vector(self):
R_function = pnl.Reduce(operation=pnl.SUM)
R_mechanism = pnl.ProcessingMechanism(
function=pnl.Reduce(operation=pnl.SUM),
default_variable=[[1], [2], [3], [4], [5]],
name="R_mechanism")
assert np.allclose(R_function.execute([[1], [2], [3], [4], [5]]),
[1, 2, 3, 4, 5])
# assert np.allclose(R_function.execute([[1], [2], [3], [4], [5]]), [15.0])
assert np.allclose(R_function.execute([[[1], [2], [3], [4], [5]]]),
[15.0])
assert np.allclose(R_mechanism.execute([[1], [2], [3], [4], [5]]),
[1, 2, 3, 4, 5])
def test_single_array(self):
R_function = pnl.Reduce(operation=pnl.SUM)
R_mechanism = pnl.ProcessingMechanism(
function=pnl.Reduce(operation=pnl.SUM),
default_variable=[[1, 2, 3, 4, 5]],
name="R_mechanism")
assert np.allclose(R_function.execute([1, 2, 3, 4, 5]), [15.0])
assert np.allclose(R_function.execute([[1, 2, 3, 4, 5]]), [15.0])
assert np.allclose(R_function.execute([[[1, 2, 3, 4, 5]]]),
[1, 2, 3, 4, 5])
# assert np.allclose(R_function.execute([[[1, 2, 3, 4, 5]]]), [15.0])
assert np.allclose(R_mechanism.execute([1, 2, 3, 4, 5]), [[15.0]])
assert np.allclose(R_mechanism.execute([[1, 2, 3, 4, 5]]), [[15.0]])
assert np.allclose(R_mechanism.execute([1, 2, 3, 4, 5]), [15.0])
def test_matrix(self):
R_function = pnl.Reduce(operation=pnl.SUM)
R_mechanism = pnl.ProcessingMechanism(
function=pnl.Reduce(operation=pnl.SUM),
default_variable=[[1, 2, 3], [4, 5, 6], [7, 8, 9]],
name="R_mechanism")
assert np.allclose(
R_function.execute([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), [6, 15, 24])
assert np.allclose(
R_function.execute([[[1, 2, 3], [4, 5, 6], [7, 8, 9]]]),
[12, 15, 18])
assert np.allclose(
R_mechanism.execute([[1, 2, 3], [4, 5, 6], [7, 8, 9]]),
[6, 15, 24])
def test_reduce_function(variable, operation, exponents, weights, scale,
offset, mode, benchmark):
if weights == 'VC':
weights = [[(-1)**i] for i, v in enumerate(variable)]
if weights == 'VR':
weights = [(-1)**i for i, v in enumerate(variable[0])]
if exponents == 'V':
exponents = [[v[0]] for v in variable]
try:
f = pnl.Reduce(default_variable=variable,
operation=operation,
exponents=exponents,
weights=weights,
scale=scale,
offset=offset)
except ValueError as e:
if not np.isscalar(scale) and "The truth value of an array" in str(e):
pytest.xfail("vector scale is not supported")
if not np.isscalar(offset) and "The truth value of an array" in str(e):
pytest.xfail("vector offset is not supported")
raise e from None
if mode == 'Python':
EX = f.function
elif mode == 'LLVM':
e = pnlvm.execution.FuncExecution(f)
EX = e.execute
elif mode == 'PTX':
e = pnlvm.execution.FuncExecution(f)
EX = e.cuda_execute
res = benchmark(EX, variable)
scale = 1.0 if scale is None else scale
offset = 0.0 if offset is None else offset
exponent = 1.0 if exponents is None else exponents
weights = 1.0 if weights is None else weights
tmp = (variable**exponent) * weights
if operation == pnl.SUM:
expected = np.sum(tmp, axis=1) * scale + offset
if operation == pnl.PRODUCT:
expected = np.product(tmp, axis=1) * scale + offset
assert np.allclose(res, expected)
rate=0.5,
default_variable=[[0.0, 0.0]],
),
output_ports=["RESULTS"],
termination_comparison_op=">=",
default_variable=[[0.0, 0.0]],
)
DECISION = pnl.DDM(
name="DECISION",
function=pnl.DriftDiffusionAnalytical(default_variable=[[0.0]]),
input_ports=[{
pnl.NAME:
pnl.ARRAY,
pnl.VARIABLE: [[0.0, 0.0]],
pnl.FUNCTION:
pnl.Reduce(default_variable=[[0.0, 0.0]], weights=[1, -1]),
}],
)
Conflict_Monitor = pnl.ObjectiveMechanism(
name="Conflict_Monitor",
function=pnl.Energy(matrix=[[0, -2.5], [-2.5, 0]],
default_variable=[[0.0, 0.0]]),
monitor=[OUTPUT],
default_variable=[[0.0, 0.0]],
)
CONTROL = pnl.ControlMechanism(
name="CONTROL",
default_allocation=[0.5],
function=pnl.DefaultAllocationFunction(default_variable=[[1.0]]),
monitor_for_control=[],