mgrad 0.1.4

A minimal automatic differentiation library.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

A minimal automatic differentiation library. Just for fun...

All code are in a single file (`src/mgrad.rs`) to easily copy-paste into other projects.  
```rust
use mgrad::nn;

fn main() {
    let x = nn::variable(1);
    let y = x.sin() + nn::constant(1);
    let y = (x.pow(2) * y).ln();
    y.backward(1);

    // y = ln(x^2 * (sin(x) + 1))
    // dy/dx should be ~ 2.29341
    println!("The gradient of y=ln(x^2 * (sin(x) + 1)) at x=1 is: {:?}", x.grad);
}
```

Run with `cargo doc --open` to see the documentation.  
More examples can be found in the `examples` folder.