TFPT scientific source excerpt Original repository path: experiments/theory-contracts/original-admissible-source-closure-20261025/KMS-AND-CARRIER-TRANSPORT.txt Original file SHA-256: 9efcfb0d68e16db842aab97b7171b2d667e7b065cb4c6ab8138102acc4c8fb68 Inclusive original lines: 37-153 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. 000037 | 000038 | 2. Fixed orientation and domains 000039 | -------------------------------- 000040 | 000041 | On H=L^2(S^1) with the original twisted domain, use 000042 | 000043 | B_a psi_n=(n+a)psi_n, 00 the reference adds m orthogonal modes; for m<0 it removes them. 000091 | Unitary transport of the entire tuple C,P,u preserves (1). This permits the 000092 | already established change to the actual D8 CAR frame; it does not turn the 000093 | interacting E8 extension into an unproved quasifree state. 000094 | 000095 | 4. The faithful original KMS representative 000096 | ------------------------------------------- 000097 | 000098 | Use the EXISTING circle functional calculus, 000099 | 000100 | C_beta=(I+exp(beta B_a))^(-1), beta>0. 000101 | 000102 | For the labeled marginal of the supplied product collar beta=2 ell. No 000103 | identification of beta or ell with alpha_EM is made. The occupations are 000104 | 000105 | n_n=(1+exp(beta(n+a)))^(-1). 000106 | 000107 | Both C_beta and I-C_beta have trivial kernel. Their spectra accumulate at 000108 | 0 and 1, but neither endpoint is an eigenvalue. Functional calculus gives 000109 | 000110 | log(I-C_beta)-log C_beta=beta B_a on Dom(B_a). (3) 000111 | 000112 | This is a densely defined selfadjoint log odds operator, rather than the 000113 | endpoint singularity of a pure projector. The exponential Fourier tails 000114 | give the explicit trace norm bound 000115 | 000116 | ||C_beta-P||_1 000117 | <= [exp(-beta a)+exp(-beta(1-a))]/[1-exp(-beta)]. (4) 000118 | 000119 | Consequently (1) holds for EVERY finite beta>0. It is not a large-beta 000120 | approximation and needs no clipping at eigenvalues zero or one. 000121 | 000122 | Directly, D_beta=uC_beta u*-C_beta is diagonal with 000123 | 000124 | d_n=n_(n-1)-n_n>0, sum_n d_n=1. (5) 000125 | 000126 | It is positive trace class and has infinite rank. Its relative determinant 000127 | telescopes on the actual open window [-L,L]: 000128 | 000129 | product_(n=-L)^L (1-z n_(n-1))/(1-z n_n) 000130 | =(1-z n_(-L-1))/(1-z n_L) ->1-z. (6) 000131 | 000132 | The difference R_Cbeta-I is trace class, so trace norm convergence of its 000133 | diagonal restrictions justifies the Fredholm limit. As beta tends to 000134 | infinity, (4) also implies ||D_beta-Pi||_1->0. 000135 | 000136 | This closes the mathematical gap between the old pure defect and a faithful 000137 | thermal source ON THE SAME CIRCLE. It does not identify that source with the 000138 | active finite collar Hamiltonian or with the full physical EM measure. 000139 | 000140 | 5. The full NS cover does not require a single quarter selector 000141 | ---------------------------------------------------------------- 000142 | 000143 | The already proved NS-cover unitary has 000144 | 000145 | B_cov=2(B_(1/4) direct_sum B_(3/4)), 000146 | u_cov=[[0,u_base],[I,0]], 000147 | u_cov^2=diag(u_base,u_base). 000148 | 000149 | The faithful covariance must be the functional calculus of this full B_cov. 000150 | It is not two arbitrarily adjusted independent sources. Applying (1)-(2) 000151 | before the unitary decomposition gives 000152 | 000153 | det_rel(u_cov C_beta u_cov*,C_beta;z)=1-z,