Manuscript XII-A@2 · XII-A-v2.tex · theorem · 955d4b22

Theorem: Polynomial product

The product and one contraction convention · lines 654–701 · proof included or linked

XII-A@2#thm:product

Back to manuscript

Every contraction is defined. Its support is compact, and its positive-degree kernel satisfies (eq:microcausal). A complete contraction is a scalar. The product is associative, unital, even, and ghost-number preserving. It preserves the regular subalgebra and extends coefficientwise to formal series.

Proof. A forbidden product cancellation would require zero covectors in all unpaired variables and , in the paired variables. The corresponding nonzero input wavefront tuple would lie wholly in a forbidden causal cone, contrary to (eq:microcausal). Hence the distribution product exists.

The projection is proper on the compact support of . Suppose a nonzero output tuple were wholly future causal. If , the second input tuple, consisting of its future external entries and , would be wholly future causal. It must therefore be zero. This forces and the second external tuple to vanish. The first tuple is now wholly future causal and must also vanish, a contradiction. For , start with the first input. Past-causal output tuples are excluded by the reversed argument. Symmetrization preserves the exclusion. When no variables remain, proper pushforward gives a scalar; there is no degree-zero wavefront claim.

On three factors the even contraction operators commute: their coefficients are independent of the configuration variables and the graded signs are included. Both bracketings equal

The required multi-contractions satisfy the same wavefront cancellation argument, so this identity involves defined distributions, not merely formal symbols. Constants give the unit. Finite-rank kernels contract to finite-rank kernels. The grading assertions follow from Definition def:admissible. For a fixed formal coefficient only finitely many orders of contraction and input coefficients contribute.

Displayed formulas in this passage

  • unlabeled · line 689

Nearby in this section