14.9.5 Tensor Kronecker Product Boundary

The Tensor Kronecker Product Boundary defines limits in tensor algebra, shaping how tensors interact through Kronecker products in multi-dimensional spaces.

Tensor Kronecker Product Boundary is the set of structural constraints that delimit where a Kronecker product between two tensors begins and ends: the index ranges it is permitted to span, the dimensional limits its resulting blocks may occupy, and the conditions under which the product degenerates or ceases to be well formed. The boundary is not a single rule but a family of edge conditions that together fix the outer envelope of validity for the Kronecker construction when it is used to combine tensor factors into a single composite object.

Index Range Boundary

Row and Column Extent

For tensors represented in matricized form, a Kronecker product of a tensor of shape (m₁, n₁) with a tensor of shape (m₂, n₂) produces a composite object whose index ranges are bounded above by the product of the respective extents. No index pair generated by the construction may exceed these bounds, and no index pair below the first position is admissible.

1 \leq i \leq m_{1} m_{2}

1 \leq j \leq n_{1} n_{2}

Boundary Index Decomposition

Every composite index at the boundary decomposes uniquely into an outer index drawn from the first factor and an inner index drawn from the second factor. This decomposition holds exactly at the first and last admissible positions, which anchors the boundary of the product to the boundary of each factor individually.

i = (i_{1} - 1) m_{2} + i_{2}

Block Boundary Structure

Block Partitioning

The Kronecker product arranges its result into contiguous blocks, each block being a scaled copy of the second factor. The boundary of each block is fixed by the extent of the second factor, so block boundaries repeat at regular intervals rather than at arbitrary positions.

Outer Boundary of the Block Grid

The outermost rows and columns of blocks correspond directly to the first and last rows and columns of the first factor. Any modification to the boundary entries of the first factor propagates only to the outermost blocks of the Kronecker product, leaving interior blocks unaffected.

Inner Boundary Within Each Block

Within a single block, the boundary is inherited entirely from the second factor. The first and last entries of every block coincide, up to the scalar multiplier supplied by the first factor, with the first and last entries of the second factor.

Degenerate Boundary Cases

Singleton Factor Boundary

When one factor reduces to a single entry, the Kronecker product boundary collapses onto the boundary of the remaining factor, and the block structure degenerates to a single block equal to a scalar multiple of that factor.

Zero Extent Boundary

If either factor has a zero extent along some axis, the corresponding axis of the product also has zero extent, and the boundary in that direction is empty; no valid index exists to anchor a first or last position.

Boundary Under Rank-One Factors

When both factors are rank-one, the boundary of the resulting product reduces to the outer product boundary, since the block structure collapses to a single row or column of scaled entries with no internal block subdivision.

Boundary Compatibility With Composition

Sequential Kronecker Products

When a Kronecker product is formed from three or more factors in sequence, the boundary of the composite is determined by processing factors in a fixed order. The boundary computed by grouping factors from the left must coincide with the boundary computed by grouping factors from the right, since the boundary depends only on the final extents and not on the grouping used to reach them.

(A \otimes B) \otimes C = A \otimes (B \otimes C)

Boundary Preservation Under Reordering

Reordering the factors of a Kronecker product changes the internal arrangement of blocks but does not change the total extent of the boundary, since the overall size is fixed by the product of all factor extents regardless of the order in which the factors are combined.

Boundary Conditions for Well-Formedness

Minimum Extent Requirement

Every factor entering a Kronecker product must have at least one row and one column for the boundary to be well defined; a factor with an undefined extent along any axis makes the boundary of the product undefined along that same axis.

Consistency With Declared Shape

The boundary of the Kronecker product must match the shape declared for the composite tensor at the point where the product is used. Any mismatch between the computed boundary and the declared shape indicates that the factors supplied to the product are not the ones intended for that position in the computation.