Rfftn with real output values

When I tried the rfft function at Pytorch a few years ago, instead of getting a complex tensor, I could get a real tensor with an addition dimension that had the real and complex values as two real value channels in the outputs. I’m curious as to how I can get the output of the rfftn in this manner. I know I can concatenate the torch.real and torch.imag into a new dimension, but I was hoping there was a faster method to achieve it.