TFPT scientific source excerpt Original repository path: experiments/theory-contracts/elliptic-moebius-source-bridge-20261026/boundary_origin_20261027/uniqueness/INDEX-ONE-VERSUS-Q-DIVISOR-REVIEW.txt Original file SHA-256: f4eaddfa0b1e47aae3238db1e930270f2712d2ff934d2c3520fa7ee3c7d0ae3c Inclusive original lines: 1-155 Line numbers below refer to the original source. This bounded excerpt is source evidence; it is not a new proof/checker run. Conversation metadata and personal execution details are omitted where indicated. 000001 | INDEPENDENT REVIEW: INDEX ONE VERSUS THE THREE-POINT q DIVISOR 000002 | 2026-10-27 000003 | 000004 | VERDICT 000005 | 000006 | The rank obstruction is correct for a claimed invertible Fredholm 000007 | identification of the old one-mode seam defect with the full D_q evaluation 000008 | defect. The full defect changes the index by -3, while the old directed 000009 | Fourier seam shift changes it by -1. An invertible Fredholm intertwiner 000010 | cannot identify those two operators. 000011 | 000012 | There is a precise one-dimensional escape: at a c-linearized fiber, the 000013 | three evaluation fibers form the regular C3 representation and its invariant 000014 | character line has dimension one. Imposing only that character condition 000015 | changes the index by -1. This removes the rank obstruction, but it does not 000016 | produce the already required inverse norm line. Its directed determinant 000017 | has the opposite beta phase. Thus the invariant-character route is allowed 000018 | as a new typed test, not yet an identification with N_q^(-1). 000019 | 000020 | 000021 | 1. THE FULL DOMAIN DEFECT HAS INDEX MINUS THREE 000022 | 000023 | Let 000024 | 000025 | E=S_0 tensor L, 000026 | D_q=q_0+q_1+q_2. 000027 | 000028 | For genus two, deg(E)=1 and therefore 000029 | 000030 | chi(E)=deg(E)+1-g=0. 000031 | 000032 | After imposing vanishing at all three q points, 000033 | 000034 | deg(E(-D_q))=-2, 000035 | chi(E(-D_q))=-3. (1) 000036 | 000037 | The evaluation sequence 000038 | 000039 | 0 -> E(-D_q) -> E -> E|D_q -> 0 (2) 000040 | 000041 | accordingly changes the Fredholm index by -3. Equivalently, 000042 | 000043 | lambda(E(-D_q)) tensor lambda(E)^(-1) 000044 | =det H^0(D_q,E|D_q)^(-1) (3) 000045 | 000046 | is the inverse determinant of a rank-three quotient. 000047 | 000048 | If the recovered Fourier seam operation is a directed index-minus-one 000049 | Fredholm defect, no bounded invertible intertwiner can turn it into (2): 000050 | Fredholm index is invariant under such an intertwining. This conclusion is 000051 | limited to an asserted identity of the two full operators. It does not 000052 | exclude a projection, an equivariant summand, a noninvertible pushforward, or 000053 | additional kernel and cokernel data. 000054 | 000055 | 000056 | 2. THE CUBIC CHARACTER DECOMPOSITION CONTAINS AN INDEX-ONE SUMMAND 000057 | 000058 | At a fiber carrying a c linearization, c cyclically permutes the three q 000059 | fibers. Hence 000060 | 000061 | H^0(D_q,E|D_q)=F_1 direct_sum F_omega direct_sum F_omega2, (4) 000062 | 000063 | with each character line one-dimensional. Composing evaluation with the 000064 | projection to F_1 and requiring only this invariant component to vanish gives 000065 | a codimension-one domain condition. Its index change is -1. 000066 | 000067 | This construction is not available fiberwise for generic L in Jac(C). 000068 | Generally c sends E_L to E_(cL), rather than acting on one fixed E_L. A 000069 | single-fiber invariant line requires a c-fixed L together with a 000070 | linearization, or an explicitly constructed equivariant family over a c 000071 | orbit. The source-selected L=O fiber satisfies this requirement, but it 000072 | does not justify the same reduction for the generic theta family. 000073 | 000074 | The nontrivial character lines are exchanged by Q because Q conjugates c to 000075 | c^(-1). The invariant line is the only one-dimensional character summand 000076 | stable under that relation. Thus it is the correct smallest equivariant 000077 | rank test, but equivariance alone does not determine its determinant 000078 | orientation. 000079 | 000080 | 000081 | 3. THE BETA PHASE EXCLUDES THE NAIVE IDENTIFICATION 000082 | 000083 | On the marked q fiber, beta acts by +i on every evaluation fiber. It 000084 | therefore acts as follows: 000085 | 000086 | beta|F_1=+i, 000087 | beta|F_1^(-1)=-i, (5) 000088 | 000089 | whereas on the full permutation-free norm, 000090 | 000091 | beta|N_q=i^3=-i, 000092 | beta|N_q^(-1)=+i. (6) 000093 | 000094 | The directed index-minus-one relative determinant obtained by imposing the 000095 | invariant condition is F_1^(-1), by the same convention as (3). Its beta 000096 | phase is -i, opposite to the +i phase of N_q^(-1). In spin notation, its 000097 | alpha square is likewise -i rather than +i. 000098 | 000099 | The undualized invariant quotient F_1 has the desired +i scalar, but it is 000100 | the opposite determinant orientation. Identifying it with the old directed 000101 | defect would reverse the relative line or the index convention. Such a 000102 | reversal must come from the original operator orientation; it cannot be made 000103 | solely to match the phase. 000104 | 000105 | The full norm product succeeds because it multiplies all three character 000106 | lines before dualizing: 000107 | 000108 | det(F_1 direct_sum F_omega direct_sum F_omega2)^(-1) 000109 | has phase (i^3)^(-1)=+i. (7) 000110 | 000111 | One character line and the product of all three are therefore not 000112 | interchangeable, even though both can be one-dimensional determinant lines. 000113 | 000114 | 000115 | 4. EXACT SCOPE 000116 | 000117 | The test proves: 000118 | 000119 | * the old one-mode index-minus-one shift is not invertibly identical to the 000120 | full three-point D_q domain restriction; 000121 | 000122 | * a c-invariant rank-one restriction is possible at a c-linearized fiber; 000123 | 000124 | * its naturally directed inverse determinant has the wrong beta and alpha 000125 | square for N_q^(-1); 000126 | 000127 | * the full product of three character lines has the correct phase but the 000128 | index-minus-three domain defect. 000129 | 000130 | It does not prove that no original boundary construction can produce 000131 | N_q^(-1). A valid construction may retain the full three-point determinant, 000132 | use a noninvertible index pushforward, or contain a separately justified 000133 | orientation or duality. What is excluded is the simple claim that the old 000134 | single Fourier defect is already the full q evaluation defect, or that its 000135 | c-invariant part is automatically the inverse q norm. 000136 | 000137 | The next required identity remains operator-typed: the original source must 000138 | specify which evaluation character or full determinant enters, with which 000139 | relative orientation, on a c and Q compatible domain. No phase-only 000140 | identification supplies that information. 000141 | 000142 | 000143 | FILES CHECKED 000144 | 000145 | * research-20261027-boundary-origin/source/ 000146 | ORIGINAL-BOUNDARY-NORM-ORIGIN.txt 000147 | 000148 | * research-20261027-boundary-origin/originals/ 000149 | INVERSE-Q-POINT-NORM-ORIGIN.txt 000150 | 000151 | * research-20261026-real-boundary/ 000152 | MARKED-Q-DIVISOR-CENTRAL-TWIST.txt 000153 | 000154 | * research-20261026-real-boundary/ 000155 | SPIN-RIGIDIFIED-THETA-MODULE.txt