# Independent exact audit The narrow claim holds: the displayed symmetric and Duflo cubic lists conflict with the quantization maps defined in the same paper. The counts and discrepancy tensors in CLAIM.md are reproduced exactly. No conclusion about the rest of the paper follows from this audit. Primary sources, independently downloaded on 2026-11-05: [arXiv v2](https://arxiv.org/abs/1301.7750v2), [journal version of record](https://pure.mpg.de/rest/items/item_1671208/component/file_1859595/content). Visual transcription and page locations are in TRANSCRIPTION.md; the PDFs and page images are retained. The title, authors, normalization, operator-word interpretation, and both plane-wave definitions were checked. Direct publisher access returned 403. Latest arXiv version is v2 (2013). Reproduce from this directory, sequentially: ```sh nice -n 10 .venv/bin/python referee_pbw.py nice -n 10 .venv/bin/python referee_plane_waves.py ``` The first program constructs a 20-dimensional exact PBW basis through degree three and the three forward quantization matrices. It inverts these matrices over Q(i), checks every roundtrip in U(su(2)), and compares every quadratic/cubic word with our transcription. The Duflo quadratic factor is independently derived from tr(ad_X^2)/48. The forward FLM map is extracted from its defining exponential, rather than a proposed table. The second program imports none of the first program. It multiplies explicit 2x2 SU(2) group exponentials in an exact ring where each independent parameter squares to zero, recovers the local group coordinates, and differentiates the three inverse plane waves. X remains a free commuting polynomial variable; Pauli group coordinates do not quotient U(su(2)) by a spin-1/2 representation. All 108 map/word cases agree between programs. All quadratics agree (9/9 per map). Cubic agreements are symmetric 18/27, Duflo 12/27, FLM 27/27. The separately fitted discrepancy tensors are: S: -delta_ik X_j; D: -delta_ik X_j - i epsilon_ijk/3; FLM: zero. Detailed actual/printed/difference records: pbw_results.json and plane_wave_results.json. There is no floating-point tolerance. The decisive trivial control is a single repeated coordinate. S must give X1^3, but the displayed cubic gives X1^3+X1. The Duflo inverse factor 1-Delta/6 gives X1^3-X1, but the display gives X1^3. Direct one-variable plane-wave Taylor coefficients separately check both. There is also an associativity obstruction in the printed Duflo list: its printed triples 123 minus 213 give 2i X3^2, whereas its own quadratic and [X1,X2]=2i X3 require 2i(X3^2-1/3). This detects the missing scalar. Reversing generator order or epsilon orientation cannot fix the repeated coordinate witness. Rescaling structure constants cannot alter the symmetrization of a pure power. Reinterpreting RHS juxtaposition as star multiplication contradicts Eq. (28), Eq. (54), and the displayed quadratic. The projection of plane waves is inactive near the identity, explicitly noted after Eq. (54). These are unrestricted external indices, not a stated distinct-index case. No convention actually used in the paper rescues the two displayed lists. Correction history and prior related work: PRIOR_ART.md. Fresh arXiv v1 also has the defective lists; no published correction was located in the checked sources. No hold-out, fitted score, gain, or SBS applies. The strongest author response is that only the examples contain typos; the general construction may remain valid. This audit does not settle that. NEXT BAR: publish the two corrected cubic identities, checked against the defining maps for all 27 index triples.