corewise

Corewise (per-core) coordinate operations on trees of arrays.

The Euclidean coordinate layer: add/sub/scale/map/sum/dot/norm applied leaf-by-leaf to nested tuples of cores (NDArrayTree), with no dense-tensor semantics – “corewise” is the semantic marker for exactly this. Gotcha: corewise_dot/corewise_norm collapse EVERY axis (stacks included) to a scalar; the corewise_stack_* variants keep a leading stack.

Attributes

NDArrayTree

Functions

corewise_add(X, Y)	Add nested objects, X,Y -> X+Y.
corewise_sub(X, Y)	Subtract nested objects, X,Y -> X-Y.
corewise_scale(X, s)	Scale nested objects, X,s -> s*X.
corewise_zeros_like(X)	Tree of zeros matching X's structure and leaf shapes/backend (= corewise_scale(X, 0)).
corewise_map(f, *Xs)	Apply f elementwise over the leaves of one or more identically-structured trees (the general
corewise_neg(X)	Negate nested objects, X -> -X.
corewise_sum(X[, axis])	Sum each array in a nested object along the given axis or axes, X -> sum(X, axis).
corewise_dot(X, Y)	Dot product of nested objects, X,Y -> X.Y.
corewise_stack_dot(X, Y, n_stack)	Like corewise_dot, but vectorized over the leading n_stack (stack) axes.
corewise_stack_scale(X, s)	Scale each leaf by a per-stack-slice factor s, broadcasting s over each leaf's trailing
corewise_stack_sum(X, axis, n_stack)	Sum each leaf over stack axes, vectorized.
corewise_norm(X)	Norm of nested objects, X -> ||X||
corewise_stack_norm(X, n_stack)	Like corewise_norm(), but vectorized over the leading n_stack (stack) axes.
corewise_err(X_true, X)
corewise_relerr(X_true, X)
corewise_logical_not(X)	Perform logical not operation on nested objects

Module Contents

NDArrayTree