entries

The entries sampling type: evaluate individual T3 entries (one-hot probes) -> scalars.

Holds the t3 + tv ops, the ambient/tangent/corewise transposes, and the frame sweeps for entry sampling. The most-special case of probe ⊃ apply ⊃ entries; imports the general machinery from probing, never the reverse. See docs/entries_apply_probe.md.

Index semantics are numpy’s throughout (forward and all transposes): a negative index wraps; an out-of-range index raises under numpy but clamps silently under jax (gather semantics) – validate indices yourself on the jax path.

Functions

t3_entries(x, index)	Compute entries of a Tucker tensor train.
t3_entries_ambient_transpose(c, index, shape[, ...])	Ambient transpose of t3_entries(): scatter c at index into CP factors.
tv_entries(index, variation, frame)	Extract entries of the dense tangent at index (= apply with unit vectors, by slicing).
tv_entries_transpose(c, index, frame[, sum_over_probes])	Apply the transpose of tv_entries() -- scatter a residual c at index into a tangent.
tv_precompute_entries_frame_sweep(frame, index)	The all-modes entries frame sweep (lean): identical to tv_precompute_apply_frame_sweep()
tv_entries_jacobian_from_sweep(variation, index, ...)	Forward all-modes entries of a tangent vector reusing a precomputed frame sweep -- the bare 𝒥
tv_entries_transpose_from_sweep(c, index, frame, ...)	Transpose of the all-modes entries reusing a precomputed frame sweep -- the bare 𝒥ᵀ (entries),
t3_entries_corewise_transpose(c, index, core_pair[, ...])	Corewise (non-manifold) transpose of t3_entries(): gradient of the