Using quimb within torch on GPU

[yuxuanzhang1995]

What is your issue?

Hi quimb team,

In example 12 you showed how to perform optimizations in torch. What do I need to change in order to use cuda as the device for torch optimization?

Best, Yuxuan

[jcmgray]

It should just be as simple as calling: apply_to_arrays(lambda x: torch.tensor(x, dtype=torch.float32).cuda()), or whichever way is your favorite to get the initial torch data onto the gpu:

1. torch.set_default_device(device) to set a default permanently
2. with torch.device(device): ... to set a default temporarily

[yuxuanzhang1995]

That works. Thank you!