warp.autograd

Utility functions for debugging automatic differentiation.

This module provides functions to evaluate and verify gradients computed by Warp’s automatic differentiation engine. Typical workflows involve computing Jacobian matrices using both automatic differentiation and finite differences, then comparing them to verify gradient accuracy.

Usage:

This module must be explicitly imported:

import warp.autograd

API

gradcheck	Checks whether the autodiff gradient of a Warp kernel matches finite differences.
gradcheck_tape	Checks whether the autodiff gradients for kernels recorded on the Warp tape match finite differences.
jacobian	Computes the Jacobians of a function or Warp kernel for the provided selection of differentiable inputs to differentiable outputs.
jacobian_fd	Computes the finite-difference Jacobian of a function or Warp kernel for the provided selection of differentiable inputs to differentiable outputs.
jacobian_plot	Visualizes the Jacobians computed by jacobian() or jacobian_fd() in a combined image plot.