def test_scalar_cast_grad():
""" test_scalar_cast_grad """
input_x = 255.5
input_t = get_py_obj_dtype(ms.int8)
def fx_cast(x):
output = F.scalar_cast(x, input_t)
return output
gfn = C.grad(fx_cast)(input_x)
expect_dx = 1
assert gfn == expect_dx
def first_derivative_if(x):
""" first_derivative_if """
return grad(if_test)(x)
def third_derivative(x):
""" third_derivative """
return grad(second_derivative)(x)
def second_derivative(x):
""" second_derivative """
return grad(first_derivative)(x)
def first_derivative(x):
""" first_derivative """
return grad(single)(x)
def second_derivative_if(x):
""" second_derivative_if """
return grad(first_derivative_if)(x)
def grad_fx_cast(input_x):
return C.grad(fx_cast)(input_x)
def invoke_while(x, y):
return C.grad(dont_unroll_while)(x, y)
def df_factorial(x):
return C.grad(factorial)(x)
def grad_rec(input_x):
return C.grad(rec)(input_x)
def df_t2_while(input_x, input_y):
return C.grad(t2_while)(input_x, input_y)
def test_grad_factorial():
res = C.grad(factorial)(3)
assert res == 11
def test_grad_while2():
res = C.grad(t2_while)(2, 3)
assert res == 3
def test_grad_rec():
""" test_grad_rec """
res = C.grad(rec)(10)
assert res == 1
def construct(self, x, y):
return C.grad(self.network)(x, y)
def mainf(x, y):
return C.grad(mul)(x, y)
def construct(self, input_):
return C.grad(self.network)(input_)
def force_fn(R: Tensor):
return -C.grad(energy_fn)(R)
