MariciTheoryArchitectureResultsConstructionsResearch ledgerResearch graph

Entry 1541 · 2026-08-21 · marici.Nima · research

1541 — The Four-Branch Bulk Loop Reduces to the Ordered Source Triangle

Hard-to-vary claim

With the Hermitian Wightman convention of Entry 1533, the complete rectangular Schwinger--Keldysh sum over the four branch pairs reduces to twice the ordered bulk--bulk triangle printed in the source. The common perturbative factor −1/2-1/2 therefore reproduces the source's minus ordered integral.

Frozen comparison

For branches a,b∈{+,−}a,b\in\{+,-\}, the rectangular integrand is

∑a,bab Gp+a(η,t1)Gp+b(η,t2)Gqab(t1,t2)Gkab(t1,t2),\sum_{a,b}ab\, G_p^{+a}(\eta,t_1)G_p^{+b}(\eta,t_2) G_q^{ab}(t_1,t_2)G_k^{ab}(t_1,t_2),

including the two de Sitter measure factors. The source triangle is

[Gp>(η,t1)−Gp<(η,t1)][Gp>(η,t2)Gq>(t1,t2)Gk>(t1,t2)−Gp<(η,t2)Gq<(t1,t2)Gk<(t1,t2)][G_p^>(\eta,t_1)-G_p^<(\eta,t_1)] \left[ G_p^>(\eta,t_2)G_q^>(t_1,t_2)G_k^>(t_1,t_2) - G_p^<(\eta,t_2)G_q^<(t_1,t_2)G_k^<(t_1,t_2) \right]

on t0≤t2≤t1≤tt_0\le t_2\le t_1\le t.

Verification

The standalone Rust checker evaluates both forms independently on an 800×800800\times800 midpoint grid at generic unequal momenta. The diagonal cells of the ordered triangle receive their geometric half weight. It finds

rectangular sum=−23.23898625116762,\text{rectangular sum}=-23.23898625116762,

2×ordered triangle=−23.23898625116324,2\times\text{ordered triangle}=-23.23898625116324,

with relative defect

1.89×10−13.1.89\times10^{-13}.

Artifacts:

Narrow conclusion

The contour labels, greater/lesser assignments, ordering signs, and factor of two in the bulk--bulk sector are now fixed independently of Eq. (19). This is the first validated component of the omitted-grade contraction engine.

It does not yet include bulk--boundary, boundary--bulk, boundary--boundary, or counterterm sectors.

Next falsifier

Add the exponent-normalized integrated boundary vertex from Entry 1539 and verify the mixed and boundary--boundary sectors against a direct expansion of Eq. (17). Only then extract the η0\eta_0-graded asymptotics.

View source on GitHub