Two difference operators #
Theorem 5.1 (thm:twofactor) of paper/nivat.tex. A primitive lattice basis
puts the first direction on the horizontal axis. Lemma 5.5 selects and
normalizes a low-cost boundary in the preimage of the original rectangle.
The positive-height argument uses Lemmas 5.6–5.8; a one-row window uses the
factorial common period from Corollary 5.3. Directional periods are transported
back through both coordinate changes. The parallel case uses bounded finite
differences. The final corollary treats sums of two periodic configurations.
The word complexity of any row is bounded by the configuration's horizontal-edge complexity,
which counts translations on all rows. Theorem 5.1 (thm:twofactor), one-row case using
Corollary 5.3 (cor:morse).
In the one-row case of Theorem 5.1 (thm:twofactor), Corollary 5.3
bounds every row period by the edge length, so its factorial is a period of
all rows simultaneously.
Reversing the transverse direction preserves the mixed-difference identity, by commutation
and negation of a period. Theorem 5.1 (thm:twofactor).
The directional conclusion of Theorem 5.1 (thm:twofactor): for
nonparallel directions, a nonzero integer multiple of one of them is a period.
The complexity bound is on the original axis-aligned rectangle.
Theorem 5.1 (thm:twofactor). A finite-range rational configuration with
at most m * n patterns on the original positive axis-aligned rectangle and
annihilated by the product of two nonzero directional differences has a
nonzero global period.
Difference operators distribute over sums, the algebraic observation after
Theorem 5.1 (thm:twofactor) that gives the two-periodic-summand corollary.
The consequence of Theorem 5.1 stated in the opening discussion of
Section 5 (sec:twofactor): a low-complexity finite-range sum of two periodic
rational configurations is periodic. The individual summands need not have
finite range.