stacking

Convert a Python tree of separate objects <-> one stacked object.

stack/unstack, tree_zip, apply_func_to_leaf_subtrees, and the basic ragged/uniform twins. This is one of three distinct meanings of “stacking” in the library – distinct from leading stack_shape batch axes on one object’s cores, and from the uniform supercore representation (docs/batching_and_stacking.md).

Functions

tree_depth(t)
get_first_leaf(xx)
trees_have_same_structure(tree1, tree2)	Checks if two trees (nested sequences) have the same structure.
apply_func_to_leaf_subtrees(tree, func, leaf_structure)	Apply a function to all "leafs" in a tree.
stack(T, axes)	Stack array-like nested tree structure.
unstack(S, axes)	Unstack nested sequence of arrays along specificed array axes.
sum_leafs_along_axes(S, axes)	Sum leafs of a tree of NDArrays along specified axes.
basic_ragged_unstack(x, first_leaf_num_nonstacking_axes)	Unstack stacked ragged array tuple into array-like tree
basic_ragged_stack(xx)	Stack array-like tree of ragged array tuples into single ragged array tuple.
basic_uniform_unstack(x, first_leaf_num_nonstacking_axes)	Unstack stacked uniform array tuple into array-like tree
basic_uniform_stack(xx)	Stack array-like tree of uniform array tuples into single ragged array tuple.
tree_zip(T1, T2)	Zips two trees with the same structure.