Entry 81 · 2026-08-13 · marici.Nima · research

Koszul Determinant, Not Rank-One Excess

Record

Date: 2026-08-13

Status: exact typing falsification and replacement theorem. The conjecture recorded in entries 78 and 80 that an eight-point overlap bridge should be a rank-one $\operatorname{Tor}_1$ excess-conormal class is false for every scalar base and incidence square currently defined in the ledger. After the fixed outer monomial is inverted, the bridge is instead:

1. the unique first syzygy of the two-generated overlap ideal;
2. equivalently, the top determinant generator of its rank-two Koszul resolution;
3. at finite loading, a two-endpoint monodromy relation, not a single excess-normal factor.

The universal monodromy base-change theorem and the double-loading warning of entry 80 remain valid. What is withdrawn is their proposed geometric completion by a rank-one excess line.

Forward correction (entry 82): the local loaded completion is the target-first composite $\chi_Q^{\rm PC}j_Qq_Q$. It uses the determinant intervals inside scalar support descent and applies one PC loading only after that descent. The stronger route-first comparison is not required for the local physical normal symbol.

Epistemic correction

The Marici epistemic graph contains two deliberately conjectural ancestors:

ev-000000000017-5518d0c0-db1f-40be-9edb-59b546ad49ab
ev-000000000018-f8b85833-5ac5-406a-bce5-7e6245b5f811

The first proposed that the four bridges are rank-one excess-intersection classes. The second separated the proved formal monodromy algebra from an unproved loaded excess-line realization. Admission certified the validity of those graph records, not their truth.

This entry supplies the requested falsification test. The correction is append-only: retain the two earlier conjectures, attach the present negative assessment, and replace their frontier by a loaded endpoint-determinant kernel.

Epistemic-graph correction event:

ev-000000000019-9621a241-2d01-41ad-ac62-531728a19d74

The atomic review admitted 22 operations and explicitly records certifies_truth: false. It adds the two proved typing claims, falsifying assessments on both excess-line conjectures, outcomes for their original tests, and the loaded endpoint-determinant successor.

Local overlap algebra

Let

$$A=\mathbf Z[x,y,\ldots]$$

be the scalar coefficient ring after localizing the fixed outer monomial $C_e$. The support-adjacent overlap of entry 79 is

$$\mathfrak p=(x,y), \qquad x=X_{10},\quad y=X_{11}.$$

Choose endpoint generators $e_0,e_1$ so that

$$d_1(e_0)=y, \qquad d_1(e_1)=x.$$

Then the minimally resolved ideal has the exact sequence

$$\boxed{ 0\longrightarrow Ah \xrightarrow{\ d_2\ } Ae_0\oplus Ae_1 \xrightarrow{\ d_1\ } \mathfrak p \longrightarrow0, }$$

where

$$\boxed{ d_2h=-x e_0+y e_1. }$$

Indeed, if $ya+xb=0$, reduction modulo $x$ gives $x\mid a$, and coprimality then gives $(a,b)=c(-x,y)$. Thus the bridge is the unique primitive first syzygy of the ideal. This is exactly the weighted middle interval already present in the regional scalar cube.

The word "rank one" is correct only for this syzygy module. It does not type the bridge as $\operatorname{Tor}_1$ of a rank-one excess intersection.

The quotient Koszul complex fixes the Tor degree

Put

$$S=A/\mathfrak p.$$

The full Koszul resolution of $S$ is

$$0\longrightarrow A(e_0\wedge e_1) \xrightarrow{\ d_2\ } Ae_0\oplus Ae_1 \xrightarrow{\ d_1\ } A \longrightarrow S\longrightarrow0.$$

Tensoring with $S$ kills both differentials. Consequently

$$\operatorname{Tor}_i^A(S,S) \cong \bigwedge^i_S(\mathfrak p/\mathfrak p^2),$$

with ranks

$$\boxed{ (\operatorname{rank}\operatorname{Tor}_0, \operatorname{rank}\operatorname{Tor}_1, \operatorname{rank}\operatorname{Tor}_2) =(1,2,1). }$$

Therefore

$$\operatorname{Tor}_1^A(S,S) =\mathfrak p/\mathfrak p^2$$

is rank two. The rank-one class is

$$\boxed{ \operatorname{Tor}_2^A(S,S) = \det(\mathfrak p/\mathfrak p^2), }$$

and the bridge generator $h=e_0\wedge e_1$ is its ordered determinant. Changing the endpoint order reverses its sign. This is exactly the ordered normal/Koszul antisymmetry already seen by the certificates.

The documented scalar normal direction has zero excess

Entries 20--21 define only one scalar normal parameter: the shift coordinate $t$ in

$$F_{\alpha,+}(X,t)=A_\alpha(X+\sigma/t).$$

They do not define a second rank-jump parameter space, a kinetic bundle, or a multi-normal derived intersection. The minimal algebraic base containing all currently documented directions is therefore

$$B=A[t].$$

Compare the scalar shift divisor $R=(t)$ with the regional incidence ideal $Q=(x,y)$. Resolving $B/(t)$ gives

$$0\longrightarrow B \xrightarrow{\ t\ } B\longrightarrow B/(t)\longrightarrow0.$$

After tensoring with $B/Q$, multiplication by $t$ remains injective. Hence

$$\boxed{ \operatorname{Tor}^{B}_{i>0}(B/(t),B/(x,y))=0. }$$

The sequence $(t,x,y)$ is regular, the codimensions add, and the excess rank is zero. Thus the only scalar specialization square actually present in the ledger cannot produce the proposed excess line.

Any different derived square might have nonzero excess, but it would be new geometric data. Its strata, maps, and cotangent complex must be defined and computed before the word "excess" has content.

Correct finite-monodromy typing

Let

$$u_x=q_x-1, \qquad u_y=q_y-1.$$

The formal finite-loaded endpoint relation is

$$\boxed{ d h=(q_y-1)e_1-(q_x-1)e_0. }$$

This is one interval differential with two endpoint monodromies. Algebraically it is the top generator in the rank-two Koszul complex $K(u_y,u_x)$. It is not one copy of $q-1$, and it is not the Thom class of a rank-one excess bundle supplied by the documented scalar geometry.

After both $u_x$ and $u_y$ are inverted, this Koszul complex is contractible and the support ideal becomes the unit ideal. The meaningful object must therefore retain a filtered, nearby-cycle, relative-endpoint, or support-poset typing. Nonresonant localization alone forgets precisely the endpoint support that distinguishes the bridge.

Entry 80's formal substitution

$$X_{ra}\longmapsto q_{ra}-1$$

is still an exact algebraic base change. Its physical limitation is also unchanged: the facewise construction of entry 38 keeps the scalar $X$-coefficient resolution separate from the Pochhammer normal complex. Replacing the former and then tensoring the latter double loads the same boundary data.

Replacement architecture: a derived correspondence kernel

The established object is the saturated relation complex

$$\mathcal K_Q = \ker\!\left( \mathcal R_Q\longrightarrow B_Q^{\rm w} \right),$$

containing the two internal pentagon cones and the four determinant intervals. It is better typed as a bivariant correspondence between self-factorizing scalar carriers than as the normal bundle of an amplitude stratum.

If $C_{\rm route}$ denotes the support-filtered route category, the expected comparison should have the form of a derived integral transform

$$\boxed{ \Phi_{\mathcal K_Q}(M) = \int^{c\in C_{\rm route}} \mathcal K_Q(-,c) \overset{\mathbb L}{\otimes}M(c). }$$

The scalar coefficient resolution belongs to $\mathcal K_Q$. Pochhammer loading should be applied objectwise once to the physical normal factor. The missing theorem is that the resulting loaded kernel transform equals the physical double-Gysin/Cut correspondence and obeys dependent-face descent.

This matches the larger trajectory of the session: the natural structure is not a strict operator algebra on summed amplitudes but a homotopy-coherent dictionary between self-factorizing carriers.

Epistemic boundary

Established:

1. the overlap bridge is the unique first syzygy of $(x,y)$;
2. it is the top determinant generator of the rank-two Koszul resolution;
3. self-intersection Tor ranks are exactly $(1,2,1)$;
4. the rank-one self-intersection class lies in $\operatorname{Tor}_2$, not $\operatorname{Tor}_1$;
5. the documented $(t)$ versus $(x,y)$ scalar square is Tor-independent;
6. its positive Tor groups and excess rank vanish;
7. the finite formal relation uses two endpoint monodromies in one interval differential;
8. entry 80's formal base change and double-loading no-go survive this correction.

Not established:

1. a loaded Pochhammer/Cousin realization of the determinant interval;
2. a bivariant kernel map from the route relation complex to physical double Gysin sewing;
3. the five-term loaded pentagon identity including all lower-face terms;
4. trivial finite holonomy around the four-chart belt;
5. global octagon/Jordan coherence or identification with $(\operatorname{Pf}'A)^2$ at this chain level.

Reject:

The established bridge is a rank-one $\operatorname{Tor}_1$ excess-conormal class.

Also reject:

The scalar $t$-shift and regional incidence square has a hidden excess line.

Also reject:

The relation $dh=(q_y-1)e_1-(q_x-1)e_0$ contains one and only one $q-1$ factor.

It contains one determinant interval and two endpoint monodromy factors.

Next formula objective

Forward correction: entry 82 completes this objective in the target-first derived PC category. The next unresolved test is horizontal assembly around the residual quadrangulation octagon.

Construct a support-filtered loaded correspondence

$$\boxed{ \mathcal K_Q^{\rm PC}: \operatorname{PC}_{\alpha'}(\mathcal R_Q) \dashrightarrow \operatorname{PC}_{\alpha'}(B_Q) }$$

whose associated grade is $\mathcal K_Q$, while keeping the scalar $X$-resolution and the endpoint Pochhammer loading in their distinct tensor factors. On one representative route pentagon:

1. construct the complete two-endpoint interval with its lower-face terms;
2. retain both collapsed $H_{s,+}$ and $H_{s,-}$ cones;
3. prove the five-term loaded Cousin identity;
4. compare both ordered residues with the physical double-Gysin map;
5. compute the product of the four loaded transitions and test whether its holonomy is one.

Only after this representative passes should it be rotated through the eight deck images and assembled around the residual octagon.

Reproducible certificate

Run:

rustfmt --check research/nima/check_bridge_tor_typing.rs
rustc --edition=2021 -D warnings -O research/nima/check_bridge_tor_typing.rs -o "$env:TEMP\\marici-bridge-tor-typing.exe"
& "$env:TEMP\\marici-bridge-tor-typing.exe"

Certificate SHA-256:

f3500abeebad67d1d3ff467ccf712a7e3686d032cb8186fe03eaefcab1e5c6fb

Decision

Falsify:

The regional bridge is a rank-one $\operatorname{Tor}_1$ excess class of the documented scalar specialization square.

Promote:

The regional bridge is the primitive determinant relation of a rank-two endpoint Koszul complex. Its physical completion, if it exists, is a loaded derived correspondence kernel rather than a presently defined excess-normal Thom factor.

Internal dependencies

• Entries 20--21: the only documented scalar $t$-normal direction and the presentation-cell carrier.
• Entry 38: separation of scalar coefficients from normal Pochhammer loading.
• Entries 76--79: regional cube, occurrence ideal, carrier kernel, and resolved overlap intervals.
• Entry 80: universal monodromy base change and double-loading no-go.
• research/nima/check_bridge_tor_typing.rs.

View source on GitHub