v0.2.0

This release significantly cleans up the code base to be minimal. We removed many versions of the Zernike evaluation functions due to their lack of performance, redundant and mostly duplicate nature. This release only has 2 jax implementations for evaluating Zernike polynomials,

• zernike_radial_cpu
• zernike_radial_gpu

zernike_radial_gpu can run on both CPUs and GPUs but due to significant performance degradation, we force zernike_radial_cpu to only run on CPUs (even if user has a GPU).

With the codebase clean up, we removed backend functions like set_device. Only jax config change done automatically is the use of double-precision. Moreover, zernipax now only runs with jax, previously we offered pure numpy backend, but we remove it for differentiability priority.

This release also removes the upper cap from dependencies. Please report a bug if a jax version doesn't work!

Full Changelog: v0.1.1...v0.2.0

v0.1.1

Full Changelog: https://github.com/PlasmaControl/FastZernike/commits/v0.1.1