Utility
This module contains some useful functions, including:
Wrapper to nearest neighbor algorithm (e.g. for moving least squares)
Degree of freedom (DOF) selection for boundary conditions
Compute condition number and check symmetry of tangent matrix
Functions for manipulating arrays
|
JIT wrapper that preserves the original docstring of the function and additionally treats arguments from possibly_static_argnames as static if their value is callable. |
|
Wrapper around zeros_like, that works also for dicts with jnp.ndarray entries. |
|
Recursively flattens a nested dict of arrays (np.ndarray or jnp.ndarray) to one flat array. |
|
Reshapes a flat array (np.ndarray or jnp.ndarray) into an array or dict of arrays matching the structure of signature_array. |
|
Changes values in a JAX array or dict of arrays based on a boolean mask. |
|
Selects elements from a JAX array or dictionary of arrays based on a boolean mask. |
|
Performs an operation on a JAX array or dict of arrays based on a boolean mask. |
|
Neighbor search within a radius based on scipy's KDTree. |
|
Computes the condition number of the assembled tangent matrix. |
|
Checks the symmetry of the assembled tangent matrix. |
|
DOF selection for the nodal imposition of boundary conditions (jitted). |
|
Converts a JAX array to a tuple. |
|
Generation of a pure JAX function based on subexpressions and reduced return-value. |
|
Computes all up to the nth derivative of a function f using a single pass. |
|
Computes up to the nth derivative of a function fun with respect to a vector argument x using a single pass. |