Add complex autograd support for torch.Tensor.index_put_

[anjali411]

It also looks like autograd for torch.Tensor.index_put_ is not even being tested for floating point tensors correctly.

>>> mask = torch.zeros(size=(10, ), dtype=torch.bool)
>>> y = torch.ones(size=(10, 10), dtype=torch.cdouble, requires_grad=True)
>>> y.clone().index_put_((mask, ), y[mask])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
RuntimeError: index does not support automatic differentiation for outputs with complex dtype.

cc @ezyang @anjali411 @dylanbespalko @mruberry @aocsa

[sunilkpai]

This is particularly an important issue for me. I'd like to help address it if possible sooner rather than later, but I've never contributed to PyTorch. What's the best place for me to learn to contribute on this?

[anjali411]

Hi @sunilkpai! Glad you are interested in contributing to PyTorch. As a first step, check out our contribution guide. For this task, you could take a look at other similar PRs adding complex autograd support. For e.g., #53240

[kurtamohler]

This issues seems to be a duplicate of #53605, right?

[anjali411]

@kurtamohler yeah! let's close this issue then