docs: reordered docs for solvers and kernels #759
Conversation
Performance reviewNo significant changes to performance. |
|
@db091756 merge conflict to resolve. I haven't started reviewing yet. |
Performance reviewNo significant changes to performance. |
Performance reviewNo significant changes to performance. |
Performance reviewNo significant changes to performance. |
Performance reviewStatistically significant changes
Normalisation values for new data: |
Performance reviewNo significant changes to performance. |
|
Merged with main, fixed those docs problems and a some small missed issues with the jaxtyping PR. Also fixed a |
Performance reviewNo significant changes to performance. |
2 similar comments
Performance reviewNo significant changes to performance. |
Performance reviewNo significant changes to performance. |
| @@ -162,25 +163,25 @@ def __check_init__(self): | |||
| raise ValueError("'output_scale' must be positive") | |||
|
|
|||
| @override | |||
| def compute_elementwise(self, x: ArrayLike, y: ArrayLike) -> Array: | |||
| def compute_elementwise(self, x, y): | |||
There was a problem hiding this comment.
The override inherits the type hints, and these ArrayLike type hints are now incorrect. If I wanted to include the correct ones here I would need to include every single override for every single kernel function which would add 100s of lines unnecessarily
There was a problem hiding this comment.
Unfortunately, this doesn't work with Sphinx for me. Take a look at the compiled documentation. Does it work for you?
coreax/kernels/scalar_valued.py
Outdated
| return self.output_scale * jnp.exp( | ||
| -jnp.linalg.norm(jnp.subtract(x, y)) / (2 * self.length_scale**2) | ||
| ) | ||
|
|
||
| @override | ||
| def grad_x_elementwise(self, x: ArrayLike, y: ArrayLike) -> Array: | ||
| def grad_x_elementwise(self, x, y): |
| return -self.grad_y_elementwise(x, y) | ||
|
|
||
| @override | ||
| def grad_y_elementwise(self, x: ArrayLike, y: ArrayLike) -> Array: | ||
| def grad_y_elementwise(self, x, y): |
| @@ -664,19 +670,19 @@ def __check_init__(self): | |||
| raise ValueError("'output_scale' must be positive") | |||
|
|
|||
| @override | |||
| def compute_elementwise(self, x: ArrayLike, y: ArrayLike) -> Array: | |||
| def compute_elementwise(self, x, y): | |||
| return self.output_scale / ( | ||
| 1 | ||
| - 2 * self.index * jnp.cos(jnp.linalg.norm(jnp.subtract(x, y))) | ||
| + self.index**2 | ||
| ) | ||
|
|
||
| @override | ||
| def grad_x_elementwise(self, x: ArrayLike, y: ArrayLike) -> Array: | ||
| def grad_x_elementwise(self, x, y): |
| return -self.grad_y_elementwise(x, y) | ||
|
|
||
| @override | ||
| def grad_y_elementwise(self, x: ArrayLike, y: ArrayLike) -> Array: | ||
| def grad_y_elementwise(self, x, y): |
| @@ -686,7 +692,7 @@ def grad_y_elementwise(self, x: ArrayLike, y: ArrayLike) -> Array: | |||
| ) ** 2 | |||
|
|
|||
| @override | |||
| def divergence_x_grad_y_elementwise(self, x: ArrayLike, y: ArrayLike) -> Array: | |||
| def divergence_x_grad_y_elementwise(self, x, y): | |||
PR Type
Description
Reordered docs for solvers and kernels
How Has This Been Tested?
Does this PR introduce a breaking change?
(Write your answer here.)
Screenshots
(Write your answer here.)
Checklist before requesting a review