Complex Neural Networks

The purpose of this repo is build neural networks that take complex valued data, e.g. RF signal/digital communications, synthetic aperture radar (SAR) and process the data using complex operations.

This code uses the complex tensor built William Falcon. Be careful to read his description of the tensor and what operations are supported. The readme is under pytorch_complex_tensor.

To Run

This notebook is a tutorial on how to build complex layers with PyTorch.

cd notebooks
jupyter lab complexLayer.ipynb

We need to write a complex relu before we can build models.

Notes:

• I included the Falcon's complex tensor repo in ours because I thought we may need to modify it. If we don't need to change it, he has a pip install that may be better to use.