def test_transpose2(x, y, axis1, axis2):
perm = (scalar_cast(axis1, u64), scalar_cast(axis2, u64))
xt = transpose(x, perm)
yt = transpose(y, perm)
d = dot(xt, yt)
sm = array_reduce(scalar_add, d, ())
return array_to_scalar(sm)
def test_prim_array_reduce():
def add(a, b):
return a + b
tests = [
(add, (2, 3, 7), (1, 3, 1), 14),
(add, (2, 3, 7), (1, 3, 8), ValueError),
(add, (2, 3, 7), (1, 2, 3, 7), ValueError),
(add, (2, 3, 7), (3, 1), 14),
(add, (2, 3, 7), (1, 1, 1), 42),
(add, (2, 3, 7), (), 42),
]
for f, inshp, outshp, value in tests:
v = np.ones(inshp)
try:
res = array_reduce(f, v, outshp)
except Exception as e:
if isinstance(value, type) and isinstance(e, value):
continue
else:
print(f'Expected {value}, but got {e}')
raise
assert res.shape == outshp
assert (res == value).all()
def test_transpose(x, y):
xt = transpose(x, (1, 0))
yt = transpose(y, (1, 0))
d = dot(xt, yt)
sm = array_reduce(scalar_add, d, ())
return array_to_scalar(sm)
def test_dot(x, y):
d = dot(x, y)
sm = array_reduce(scalar_add, d, ())
return array_to_scalar(sm)
def test_array_operations_std(xs, ys):
div = xs / ys
sm = array_reduce(scalar_add, div, ())
return array_to_scalar(sm)
def test_array_operations_reshape(xs, ys):
xs = reshape(xs, (6, ))
ys = reshape(ys, (6, ))
div = array_map(scalar_div, xs, ys)
sm = array_reduce(scalar_add, div, ())
return array_to_scalar(sm)
def test_array_operations_distribute(x, y):
xs = distribute(scalar_to_array(x, AA), (4, 3))
ys = distribute(scalar_to_array(y, AA), (4, 3))
div = array_map(scalar_div, xs, ys)
sm = array_reduce(scalar_add, div, ())
return array_to_scalar(sm)
def test_array_operations(xs, ys):
div = array_map(scalar_div, xs, ys)
sm = array_reduce(scalar_add, div, ())
return array_to_scalar(sm)
def f(x):
def add(x, y):
return x + y
return array_reduce(add, x, (1, 3))
def test_array_reduce2(x):
return array_reduce(scalar_add, x, (3, ))
def test_array_reduce(x):
return array_reduce(scalar_add, x, (1, 3))
def helper(fn):
return array_reduce(fn, xs, ()), array_reduce(fn, ys, ())
def before(x):
return array_reduce(scalar_add, x, (3, 1))
def cost(model, x, y):
yy = model.apply(x)
diff = (yy - y)
return (array_reduce(scalar_add, diff**2, ())).item()
