Differentiable Floating-point Operations

The DFO crate aims at making your already existing code differentiable, leveraging automatic differentiation.

Currently, my goal is to implement the Float trait on DFloat for forward autodiff.

TODOs:

• [ ] Implement missing traits from std::ops
• [ ] Remove Copy trait when possible, especially for arrays
• [ ] Create utils functions to create constants, variables, etc.
• [ ] Create tests and benchmarks
• [ ] Implement other traits, until Float is implemented
• [ ] Reorganize and clean crate
• [ ] Create some GitHub actions
• [ ] Publish crate on https://crates.io
• [ ] Create examples for multiple inputs, multiple outputs, ndarrays, and more
• [ ] Go for backward autodiff

Similar crates:

• https://github.com/elrnv/autodiff