Nathan Simpson

XLA translation rule for primitive 'two_phase_op_lin' not found

1

Hi fax team! We're still making use of the `two_phase` method to implicitly differentiate through a maximum likelihood fit. In this context, I recently tried to add an additional root-finding...

Possible standalone distribution of `xarray_jax`?

2

Hi! I'm a big fan of the boilerplate code used to wrap `xarray` into a JAX-compatible entity -- I think this could have wider usage were it more well-known, especially...

Document behavior of generating destination path when copying

2

Documents #1497.

[Question] Using eqx.Linear with inferred input size?

1

Hi Patrick! I'm currently converting a Haiku codebase to use equinox, and there's a _lot_ of boilerplate that tries to construct MLPs before knowing the full size of the input...

feat: Add Cramer-rao uncertainties + covariance using autodiff to non-minuit fits by default

15

# Description One of the main reasons minuit is the default optimizer is that it packages up fit results with uncertainties. We can do that for any fitting procedure using...

### Outline As I understand, `cabinetry` has within its scope to enable and drive an analysis that is fully differentiable. In order for this to work in practice, one should...

Error in OpenGL SurfaceExample: `'Sphere' object has no attribute 'uv_func'`

4

Hi! I'm running [this scene (`SurfaceExample`)](https://github.com/ManimCommunity/manim/blob/04bfa221075166a00eaa11fc9e9f0757f56625d2/example_scenes/opengl.py#L447) in the OpenGL examples, and I'm unable to create the `OpenGLTexturedSurface` because the `Sphere` object is not of type `OpenGLSurface` and has no `uv_func`...