Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

The exact fourth-vector feasibility wall of the canonical MUB triple in dimension six

Noah Erlwein

Published Jun 12, 2026• 0 citations
Publisher

Abstract

For the canonical product MUB triple {Z₂ ⊗ Z₃, X₂ ⊗ Q₀, Y₂ ⊗ Q₁} in dimension six, known to admit no vector unbiased to all three bases, we prove an exact two-sided theorem for the projector-space feasibility functional f, where f(v) = 1 would characterize a candidate fourth unbiased vector. We show that the global maximum of f is W = (88 + 3√6) / 100 ≈ 0.9534846923. The bound is attained by an explicit algebraic unit vector, verified entirely in exact arithmetic over the field Q(ζ₂₄, √5). Its symmetry orbit contains 72 distinct maximizing rays. The matching global upper bound f ≤ W is established via an exact positive-semidefinite certificate at the level-2 Veronese lift. The certificate consists of a symmetric matrix over Q(√2, √3), whose polynomial identity and positivity are machine-verified in exact arithmetic and closed deterministically under the triple's 432-element symmetry group. No randomized computation or floating-point arithmetic appears anywhere in the proof chain. We further prove, via exact separating functionals, that degree-4 and degree-6 sum-of-squares certificates do not exist. In this precise sense, the lift level employed is minimal and necessary. Companion results include: • an exact product-vector ceiling for aligned product triples; • the exact wall value for the second known triple; • a complete Weyl-board census in dimension six, together with a Lean 4 kernel-verified combinatorial layer. These results completely characterize the pinned canonical triple. They do not resolve the MUB(6) problem, which remains open. The deposit contains two files: (1) the manuscript (PDF); (2) a complete verification bundle (ZIP; 91 files) containing all computational modules, tests, evidence receipts, execution logs, a SHA-256 manifest, and a minimal independent verifier (verify_global_certificate.py). The verifier is implemented in standard-library-only Python, shares no code with the certificate-construction pipeline, loads the stored exact certificate matrix, and independently re-establishes both directions of the global bound in approximately 150 seconds. Reproduction is available via: make verify-global Code is released under the Apache License 2.0 (with SPDX headers throughout the bundle). The manuscript text is released under CC BY 4.0. This research made extensive use of AI systems for conjecture generation, computation, proof drafting, and adversarial review. However, no mathematical claim depends on the correctness of any AI system. Every reported result is verified by exact rational arithmetic, exact interval arithmetic, or the Lean 4 kernel, and all verification artifacts are included in the deposit.

DOI
10.5281/zenodo.20670933

Related papers

An Independent Reproduction and Four Extensions of the MUB(6) Fourth-Vector Feasibility Wall

JEREMY H. CARROLL

Title. An Independent Reproduction and Four Extensions of the MUB(6) Fourth-Vector Feasibility Wall Abstract. We present a companion note that cites N. Erlwein’s exact fourth-vector feasibility wall for a canonical mutually unbiased product triple in dimension six (Zenodo 20682233; theorem W = (88 + 3√6)/100). We do not redistribute the upstream software bundle. Independently, we (A) maintain a pure-Rust exact verification surface; (B) raise a d=10 closed-form hunt to a 460-decimal-place stationarity anchor with a widened PSLQ menu (no relation found); (C) run a full Cl(2)×Cl(5) single-qudit Clifford census at d=10, finding a single antiunitary Z/2 real structure; and (D) issue exact rational finite-cover exclusion certificates on a specified cell of the gauge-fixed candidate 5-torus. MUB(6)=3 and MUB(10) remain open. Local certificates mint no authority. Why a researcher should care. The note strengthens a rare exact, machine-checkable obstruction artifact in d=6 quantum information with a second language (Rust), higher-precision d=10 evidence, and independently checkable finite covers—while keeping claim boundaries explicit. What this does not claim. Not MUB(6)=3; not MUB(10); not a closed form at d=10; not joint authorship of Erlwein’s wall; not a global moduli theorem. Reproduction. cd crate && cargo test -j 1cargo run --release --bin mub_cover -- --region quadrant DOI. Pending human mint. Upstream wall: 10.5281/zenodo.20682233.

•Aug 3, 2026•0 citations

Mutually Unbiased Bases in Dimension Twelve: Tensor Saturation, Strong Unextendibility, and the Flat-Sum Trichotomy

Han-Jun Lim

Abstract Setting. The maximal number N(12) of mutually unbiased bases (MUBs) of C12 is unknown; no composite non-prime-power dimension has a known exact value. The quantum MacNeish (tensor) bound gives N(12) ≥ min{N(3),N(4)} = 4, and within Array Cosmology (AC) the integer 4 = k is the ququart rank of the rendering algebra R12 = M3(C)⊗ M4(C). This paper maps the d = 12 MUB landscape and states the tensor saturation conjecture N(12) = 4. Canonical sector. (i) Strong unextendibility [A]: no vector of C12 is unbiased to all four bases of the canonical product quadruple; the proof closes through an entanglement-forcing theorem, a flatness–circle-structure reduction, a monomial-rigidity step, and an atomic lemma whose ninebranches collapse to f = U+UV+V withexactminimum5/4 = (k+1)/k > 1. (ii)Localrigidity[A]: the canonical quadruple is rigid modulo its 191-dimensional gauge orbit; the Jacobian has exact rank 385 and exact nullity 191, certified by an explicit integer basis of the kernel verified in exact arithmetic, by modular arithmetic at p = 1000033, and by an independent gauge-dimension proof

•Jun 13, 2026•0 citations

An exact MUB triple in dimension six with two Fourier edges

Adian Jaupi

In this research, I constructed an exact algebraic triple of mutually unbiased bases (MUBs) in ℂ⁶ consisting of the computational basis, a member D(ρ) of the Diţă slice from Karlsson's three-parameter family, and an explicit third basis B. Through exact symbolic computation over a degree-96 number field extension, I proved that both remaining transition matrices in the triple belong precisely to the two-parameter Fourier family F(a,b) up to equivalence and transposition, revealing that the Diţă-slice matrix factors directly into a product of two Fourier-type Hadamard matrices. Applying the theorem of Jaming et al., I established that both Fourier edges are non-extendable, proving the triple cannot be extended to any MUB quartet. Additionally, I derived a closed-form expression D(z) for the entire Diţă slice, resolved degeneracies in Karlsson's Möbius parametrization, and computed its 2×2 Hadamard submatrices exactly (51 generically and 75 at z=1), while identifying a second exact Fourier-Fourier triangle at z=1. Finally, using homotopy continuation to track witness systems across seven slice parameters, I recovered 40 third bases that all preserved this Fourier-Fourier pattern to residuals within 5×10⁻¹⁵, providing strong numerical backing for a global structural conjecture along the full Diţă slice.

•Oct 3, 2026•0 citations

No Szöllősi matrix lies in a quadruplet of mutually unbiased bases in dimension six

Anonymous

Unrefereed candidate excluding the specified closed two-circulant Szollosi family from MUB quadruplets in complex dimension six. The Fourier-family result re-proves prior work. The unrestricted dimension-six maximum remains conditional on MMVW Conjecture 2. The package separates an untrusted hull generator from an interval and graph checker; canonical half-open ownership repairs boundary coverage, and input-bound clean replay prevents stale cache acceptance. No end-to-end formal verification, unaffiliated reproduction or external human peer review is claimed. See the replay receipt for exact measured scope. Original prose and data CC0; original code MIT; third-party and review material retain their terms.

•Sep 30, 2026•0 citations

Certified recovery of trace-determinant structure under low-dimensional corruption

Daniel Eric Fredriksen

First preprint and reproducibility companion for bounded exact recovery of a nonzero trace-determinant tensor over an unknown binary extension field under corruption supported on a small common covector space. Includes the manuscript PDF and source, complete prose proofs, executable recovery and independent certificate checks, pinned partial Lean proofs, validation receipts, and three AI-model review reports with feedback dispositions. For s >= 2R+5, at most 8*2^(4R)*h trials return a verified compatible representation with probability at least 1-2^(-h). Accepted outputs are checked exactly; unique identification of a planted field is not claimed. The cubic-only bound does not establish the broader quantum application or a BPP/BQP result. Lean verification is partial. The complete recovery theorem, its probability and resource bounds, and the Python implementation are not fully formally verified. The reviews are AI-assisted internal reviews, not independent human peer review. Licenses: manuscript text and PDF are CC-BY-4.0; software and Lean sources are Apache-2.0. See LICENSE, LICENSES/CC-BY-4.0.txt and PROVENANCE.md. The package-level Other (Open) designation preserves these per-component licenses.

•Oct 28, 2026•0 citations

Dimensions and generic recovery of matching tensors

Ignoramus

Give each vertex i of a graph a complex vector space V_i, place a tensor W_ij in V_i ⊗ V_j on each edge, and sum the tensor products of these edge tensors over all perfect matchings. We study the image of this matching-tensor map and the recovery of the edge tensors from its output. For every support graph with a perfect matching and all local dimensions at least three, the image closure has dimension N_* - |E_*| + dim P(G) + 1, where E_* is the set of edges lying in some perfect matching, N_* is the number of parameters on those edges and P(G) is the perfect matching polytope; on those edges the generic differential kernel consists of the edge rescalings that preserve every matching term. For complete support at every even order at least four and all local dimensions at least three, the entire fibre over a general output is a single vertex-rescaling orbit, and a normalized representative is a rational function of the output. The proof recovers cofactor lines from tangent spaces and glues lower-order representations, allowing vanishing cofactors. For complete support at every even order at least six, the dimension formula and the vertex-rescaling kernel remain valid when the local dimensions are only two. At four vertices we determine the dimension in every format with local dimensions at least two. Files. paper.pdf is the manuscript, revision 4 (25 pages). paper.tex is its LaTeX source. The other ten files are the contents of the directory verification/ that the paper refers to: five standard-library Python programs (verify_qubit_certificates.py, verify_qubit_from_matchings.py, check_graph.py, independent_rank_check.py, verify_four.py) and five JSON files holding their certificates and recorded results. The SHA-256 values of the five programs are printed in the final section of the paper. To run the programs, keep the ten files together in one folder. Prior work. Earlier results are credited in Section 1 and in the references of the paper. Disclosure. AI systems from OpenAI, working under the direction of the person who publishes as Ignoramus, developed the statements and proofs, performed the computations and wrote the first three versions; each version was reviewed by Claude (Anthropic), also an AI system, and the edits of this fourth version were written and checked by Claude agents. The final section of the paper sets this out in full. No AI system is an author; the author publishes under a pseudonym. The author has not checked the proofs, and no human mathematician has checked the manuscript. Two determinant computations in one proof can be replayed with the accompanying programs; the remaining arguments are analytic. The AI reviews are not human refereeing, peer review or a certification of originality, and novelty and significance are not established. The manuscript is a draft for mathematical review. Corrections are welcome at ignoramus110926@gmail.com.

•Nov 10, 2026•0 citations
For the canonical product MUB triple {Z₂ ⊗ Z₃, X₂ ⊗ Q₀, Y₂ ⊗ Q₁} in dimension six, known to admit no vector unbiased to all three bases, we prove an exact two-sided theorem for the projector-space feasibility functional f, where f(v) = 1 would characterize a candidate fourth unbiased vector. We show that the global maximum of f is W = (88 + 3√6) / 100 ≈ 0.9534846923. The bound is attained by an explicit algebraic unit vector, verified entirely in exact arithmetic over the field Q(ζ₂₄, √5). Its symmetry orbit contains 72 distinct maximizing rays. The matching global upper bound f ≤ W is established via an exact positive-semidefinite certificate at the level-2 Veronese lift. The certificate consists of a symmetric matrix over Q(√2, √3), whose polynomial identity and positivity are machine-verified in exact arithmetic and closed deterministically under the triple's 432-element symmetry group. No randomized computation or floating-point arithmetic appears anywhere in the proof chain. We further prove, via exact separating functionals, that degree-4 and degree-6 sum-of-squares certificates do not exist. In this precise sense, the lift level employed is minimal and necessary. Companion results include: • an exact product-vector ceiling for aligned product triples; • the exact wall value for the second known triple; • a complete Weyl-board census in dimension six, together with a Lean 4 kernel-verified combinatorial layer. These results completely characterize the pinned canonical triple. They do not resolve the MUB(6) problem, which remains open. The deposit contains two files: (1) the manuscript (PDF); (2) a complete verification bundle (ZIP; 91 files) containing all computational modules, tests, evidence receipts, execution logs, a SHA-256 manifest, and a minimal independent verifier (verify_global_certificate.py). The verifier is implemented in standard-library-only Python, shares no code with the certificate-construction pipeline, loads the stored exact certificate matrix, and independently re-establishes both directions of the global bound in approximately 150 seconds. Reproduction is available via: make verify-global Code is released under the Apache License 2.0 (with SPDX headers throughout the bundle). The manuscript text is released under CC BY 4.0. This research made extensive use of AI systems for conjecture generation, computation, proof drafting, and adversarial review. However, no mathematical claim depends on the correctness of any AI system. Every reported result is verified by exact rational arithmetic, exact interval arithmetic, or the Lean 4 kernel, and all verification artifacts are included in the deposit.
Title. An Independent Reproduction and Four Extensions of the MUB(6) Fourth-Vector Feasibility Wall Abstract. We present a companion note that cites N. Erlwein’s exact fourth-vector feasibility wall for a canonical mutually unbiased product triple in dimension six (Zenodo 20682233; theorem W = (88 + 3√6)/100). We do not redistribute the upstream software bundle. Independently, we (A) maintain a pure-Rust exact verification surface; (B) raise a d=10 closed-form hunt to a 460-decimal-place stationarity anchor with a widened PSLQ menu (no relation found); (C) run a full Cl(2)×Cl(5) single-qudit Clifford census at d=10, finding a single antiunitary Z/2 real structure; and (D) issue exact rational finite-cover exclusion certificates on a specified cell of the gauge-fixed candidate 5-torus. MUB(6)=3 and MUB(10) remain open. Local certificates mint no authority. Why a researcher should care. The note strengthens a rare exact, machine-checkable obstruction artifact in d=6 quantum information with a second language (Rust), higher-precision d=10 evidence, and independently checkable finite covers—while keeping claim boundaries explicit. What this does not claim. Not MUB(6)=3; not MUB(10); not a closed form at d=10; not joint authorship of Erlwein’s wall; not a global moduli theorem. Reproduction. cd crate && cargo test -j 1cargo run --release --bin mub_cover -- --region quadrant DOI. Pending human mint. Upstream wall: 10.5281/zenodo.20682233.
Abstract Setting. The maximal number N(12) of mutually unbiased bases (MUBs) of C12 is unknown; no composite non-prime-power dimension has a known exact value. The quantum MacNeish (tensor) bound gives N(12) ≥ min{N(3),N(4)} = 4, and within Array Cosmology (AC) the integer 4 = k is the ququart rank of the rendering algebra R12 = M3(C)⊗ M4(C). This paper maps the d = 12 MUB landscape and states the tensor saturation conjecture N(12) = 4. Canonical sector. (i) Strong unextendibility [A]: no vector of C12 is unbiased to all four bases of the canonical product quadruple; the proof closes through an entanglement-forcing theorem, a flatness–circle-structure reduction, a monomial-rigidity step, and an atomic lemma whose ninebranches collapse to f = U+UV+V withexactminimum5/4 = (k+1)/k > 1. (ii)Localrigidity[A]: the canonical quadruple is rigid modulo its 191-dimensional gauge orbit; the Jacobian has exact rank 385 and exact nullity 191, certified by an explicit integer basis of the kernel verified in exact arithmetic, by modular arithmetic at p = 1000033, and by an independent gauge-dimension proof
In this research, I constructed an exact algebraic triple of mutually unbiased bases (MUBs) in ℂ⁶ consisting of the computational basis, a member D(ρ) of the Diţă slice from Karlsson's three-parameter family, and an explicit third basis B. Through exact symbolic computation over a degree-96 number field extension, I proved that both remaining transition matrices in the triple belong precisely to the two-parameter Fourier family F(a,b) up to equivalence and transposition, revealing that the Diţă-slice matrix factors directly into a product of two Fourier-type Hadamard matrices. Applying the theorem of Jaming et al., I established that both Fourier edges are non-extendable, proving the triple cannot be extended to any MUB quartet. Additionally, I derived a closed-form expression D(z) for the entire Diţă slice, resolved degeneracies in Karlsson's Möbius parametrization, and computed its 2×2 Hadamard submatrices exactly (51 generically and 75 at z=1), while identifying a second exact Fourier-Fourier triangle at z=1. Finally, using homotopy continuation to track witness systems across seven slice parameters, I recovered 40 third bases that all preserved this Fourier-Fourier pattern to residuals within 5×10⁻¹⁵, providing strong numerical backing for a global structural conjecture along the full Diţă slice.
Unrefereed candidate excluding the specified closed two-circulant Szollosi family from MUB quadruplets in complex dimension six. The Fourier-family result re-proves prior work. The unrestricted dimension-six maximum remains conditional on MMVW Conjecture 2. The package separates an untrusted hull generator from an interval and graph checker; canonical half-open ownership repairs boundary coverage, and input-bound clean replay prevents stale cache acceptance. No end-to-end formal verification, unaffiliated reproduction or external human peer review is claimed. See the replay receipt for exact measured scope. Original prose and data CC0; original code MIT; third-party and review material retain their terms.
First preprint and reproducibility companion for bounded exact recovery of a nonzero trace-determinant tensor over an unknown binary extension field under corruption supported on a small common covector space. Includes the manuscript PDF and source, complete prose proofs, executable recovery and independent certificate checks, pinned partial Lean proofs, validation receipts, and three AI-model review reports with feedback dispositions. For s >= 2R+5, at most 8*2^(4R)*h trials return a verified compatible representation with probability at least 1-2^(-h). Accepted outputs are checked exactly; unique identification of a planted field is not claimed. The cubic-only bound does not establish the broader quantum application or a BPP/BQP result. Lean verification is partial. The complete recovery theorem, its probability and resource bounds, and the Python implementation are not fully formally verified. The reviews are AI-assisted internal reviews, not independent human peer review. Licenses: manuscript text and PDF are CC-BY-4.0; software and Lean sources are Apache-2.0. See LICENSE, LICENSES/CC-BY-4.0.txt and PROVENANCE.md. The package-level Other (Open) designation preserves these per-component licenses.
Give each vertex i of a graph a complex vector space V_i, place a tensor W_ij in V_i ⊗ V_j on each edge, and sum the tensor products of these edge tensors over all perfect matchings. We study the image of this matching-tensor map and the recovery of the edge tensors from its output. For every support graph with a perfect matching and all local dimensions at least three, the image closure has dimension N_* - |E_*| + dim P(G) + 1, where E_* is the set of edges lying in some perfect matching, N_* is the number of parameters on those edges and P(G) is the perfect matching polytope; on those edges the generic differential kernel consists of the edge rescalings that preserve every matching term. For complete support at every even order at least four and all local dimensions at least three, the entire fibre over a general output is a single vertex-rescaling orbit, and a normalized representative is a rational function of the output. The proof recovers cofactor lines from tangent spaces and glues lower-order representations, allowing vanishing cofactors. For complete support at every even order at least six, the dimension formula and the vertex-rescaling kernel remain valid when the local dimensions are only two. At four vertices we determine the dimension in every format with local dimensions at least two. Files. paper.pdf is the manuscript, revision 4 (25 pages). paper.tex is its LaTeX source. The other ten files are the contents of the directory verification/ that the paper refers to: five standard-library Python programs (verify_qubit_certificates.py, verify_qubit_from_matchings.py, check_graph.py, independent_rank_check.py, verify_four.py) and five JSON files holding their certificates and recorded results. The SHA-256 values of the five programs are printed in the final section of the paper. To run the programs, keep the ten files together in one folder. Prior work. Earlier results are credited in Section 1 and in the references of the paper. Disclosure. AI systems from OpenAI, working under the direction of the person who publishes as Ignoramus, developed the statements and proofs, performed the computations and wrote the first three versions; each version was reviewed by Claude (Anthropic), also an AI system, and the edits of this fourth version were written and checked by Claude agents. The final section of the paper sets this out in full. No AI system is an author; the author publishes under a pseudonym. The author has not checked the proofs, and no human mathematician has checked the manuscript. Two determinant computations in one proof can be replayed with the accompanying programs; the remaining arguments are analytic. The AI reviews are not human refereeing, peer review or a certification of originality, and novelty and significance are not established. The manuscript is a draft for mathematical review. Corrections are welcome at ignoramus110926@gmail.com.