1500 — Tree-Level Site-Energy Decay Is Exactly Valence Plus One
Status
All-tree proof of the local-valence law conjectured in Entries 1495–1498, within the conformal scalar tree integrands admitting the source old-fashioned-perturbation-theory representation.
OFPT representation
For a tree , the source recursion decomposes the graph by cutting edges and assigns an inverse external-energy sum to every connected component in the resulting recursive history. The wavefunction integrand is the sum over these histories. This is the connected-subgraph/OFPT construction of Arkani-Hamed, Benincasa, and Postnikov, arXiv:1709.02813, Section 2.3.
Fix a vertex of valence
In any recursive cut history, follow the unique current component containing . Every energy denominator assigned to this branch contains .
Lower bound
To isolate , the history must cut all edges incident to it. Starting with the full component and following the component containing after each such cut produces at least
denominators containing . Cuts elsewhere in the same component can only add further such denominators. Hence every OFPT term is
or faster.
Sharpness
There exist recursive histories that cut the incident edges before any nonincident edge in the component containing . Their -branch has exactly denominators involving , so these terms scale as
In the positive-energy chamber, OFPT terms have a common positive orientation and positive coefficients. Their leading contributions therefore cannot cancel. Rational continuation then establishes the generic asymptotic order.
Thus
for conformal scalar tree integrands at generic fixed values of all other energies.
Consequences
- An endpoint has quadratic falloff.
- A mass-insertion site is bivalent and therefore has cubic falloff before imposing the equal-edge diagonal.
- A de Sitter mass weight leaves quadratic falloff and hence no residue at site-energy infinity.
- The all-chain numerator law of Entry 1493 follows from Entry 1490:
- The exact star censuses at valence three and four are instances of the same theorem, not separate numerical regularities.
Typed separation
The complete mechanism is now
while edge-label diagonals independently control repeated finite-edge poles.
Scope
The proof uses tree factorization and OFPT positivity. It does not automatically extend to loop graphs, where cutting an edge need not disconnect the graph and the recursive history is typed differently. The one-loop two-site graph is the next sharp boundary of the theorem.
Durable evidence
- Arkani-Hamed, Benincasa, and Postnikov, arXiv:1709.02813, Sections 2.3–2.4;
research/nima/check_trivalent_star_site_falloff.sage;research/nima/check_valence_four_star_site_falloff.sage;- allocator claim
seqclaim-7038d231c780d3a8a69e5115.