Module compression

Compression techniques for sparse matrices in numerical differentiation

This module provides implementations of compression algorithms for sparse matrices to reduce the number of function evaluations required for finite differences.

Functions

compress_hessian_pattern

Compresses a sparse Hessian pattern for more efficient finite differencing

compress_jacobian_pattern

Compresses a sparse Jacobian pattern for more efficient finite differencing

reconstruct_hessian

Reconstructs a sparse Hessian from compressed gradient evaluations

reconstruct_jacobian

Reconstructs a sparse Jacobian from compressed gradient evaluations

Type Aliases

CompressedJacobianPattern

Type alias for the return type of compress_jacobian_pattern