13.22 Tensor Contraction Boundary

Tensor Contraction Boundary is the limiting condition that marks the point beyond which no further contraction can be applied to a given tensor or sequence of tensors, determined by the exhaustion of available upper or lower indices, and serving as the natural stopping point for any process of repeated index reduction.

Definition

For a tensor of type $$(p,q)$$, the contraction boundary is reached once further contraction becomes impossible, which occurs precisely when:

At this point, no valid upper-lower index pair remains available, so the pair validity check cannot be satisfied by any choice of indices from the tensor alone.

Reaching the Boundary

Through Repeated Contraction

Starting from an initial type $$(p_0,q_0)$$ and applying the type pair update rule repeatedly, the boundary is reached after exactly $$\text{min}(p_0,q_0)$$ contractions, at which point the type has become:

Two Kinds of Boundary

When $$p_0=q_0$$, the boundary is reached at type $$(0,0)$$, a scalar, and no indices of either variance remain. When $$p_0\ne q_0$$, the boundary is reached at a type with one entry zero and the other equal to $$|p_0-q_0|$$, leaving a tensor of purely one variance.

Crossing the Boundary With External Structure

Metric-Assisted Continuation

Although the boundary marks the limit of contraction using only the tensor's own indices, further reduction of a purely upper or purely lower tensor becomes possible if a metric is introduced, since the metric simplification pattern allows an index to be converted in variance, effectively creating a new pairing opportunity not present in the original tensor alone.

Boundary Relative to a Fixed Tensor Set

In a larger network involving several tensors, the effective boundary is reached only when no valid upper-lower pairing remains across any of the tensors in the network, not merely within a single tensor, so the boundary for a network can occur only after all cross-tensor pairing opportunities have also been exhausted.

Diagram

Significance

The contraction boundary provides the termination criterion for the general contraction sequence procedure, guaranteeing that any process of repeated contraction on a fixed tensor terminates after a finite, precisely determined number of steps, and marking the point at which any further reduction requires additional mathematical structure, such as a metric, beyond what the original tensor alone provides.