Execute JAX-native discrete, stationary, and continuous wavelet transforms on input arrays.
npx skills add https://github.com/handley-lab/jaxwavelets --skill jaxwavelets
Please help me install this Agent Skill. Skill: jaxwavelets Source: https://github.com/handley-lab/jaxwavelets/tree/main Command: npx skills add https://github.com/handley-lab/jaxwavelets --skill jaxwavelets
This Skill helps you compute discrete, stationary, and continuous wavelet transforms inside JAX so you can differentiate through them, JIT-compile them, and run them efficiently on CPU/GPU.
Use the jaxwavelets Skill to compute a 2D wavelet decomposition for a JAX array x with wavelet 'db4' at level 3.
High-intent search queries and answers about installing and using this skill.
You can compute differentiable wavelet transforms in JAX using native discrete, stationary, and continuous wavelet algorithms that output JAX pytrees, enabling seamless gradient computation through jax.grad.
Yes, JAX wavelet transforms are JIT-compatible and run efficiently on GPU, with the continuous wavelet transform using a two-phase prepare and apply kernel design to maintain JIT-friendliness.
Yes, it provides a PyWavelets-matched API with single-example functional semantics, allowing users familiar with PyWavelets to perform 1D, nD, and 2D discrete wavelet transforms natively in JAX.
You can apply soft, hard, garrote, or firm thresholding directly to wavelet coefficients to perform signal denoising and multi-scale analysis within the JAX ecosystem.
Stationary wavelet transforms perform shift-invariant or undecimated decompositions in JAX, reconstructing the original signal while keeping outputs as pytree coefficients for batching and gradients.
Use discrete wavelet transforms for standard multi-scale decomposition and reconstruction, and continuous wavelet transforms when you need a two-phase prepare and apply workflow that remains JIT-friendly.
tondevrelSolve PDEs and run differentiable physics simulations with JAX.
firstsun-devDeconstructs problems to fundamental truths and rebuilds solutions from hard constraints.
tondevrelCompose JAX function transformations for automatic differentiation, JIT compilation, and vectorization.
NVIDIA-BioNeMoMap CPU-bound genomics workflow steps to NVIDIA Parabricks GPU equivalents.
