>>> torch.linspace(-10j, 10j, steps=5)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
RuntimeError: value cannot be converted to type float without overflow: (-0,-10)
>>> torch.from_numpy(numpy.linspace(-10j,10j,4))
tensor([(0.0000-10.0000j), (0.0000-3.3333j), (0.0000+3.3333j), (0.0000+10.0000j)],
dtype=torch.complex128)
Linspace behavior for complex dtypes seems buggy (atleast on the cpu side)
Additionally the document says start and end should be
floatwhich should be changed tofloat or complex(similar comment for logspace as well)cc @ezyang @anjali411 @dylanbespalko