AFTD Auto-Formalizing Theoretical Domains

open_problem_693_three_dim_fewer_than_nine

theoremverifiedformalization

There exists a natural number k strictly less than 9 such that has_subspace_compatible_incompatible_family 3 k holds, meaning there is a family of k binary measurements on C^3 that is incompatible on C^3 yet jointly measurable on every proper subspace.

Statement

theorem open_problem_693_three_dim_fewer_than_nine : ∃ k < 9, has_subspace_compatible_incompatible_family 3 k
Formalizes
arXiv:2610.37347 (Measurement incompatibility and noiseless quantum codes), after Corollary 1
Verified
12 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 19

Read back from the Lean

There exists a natural number $k < 9$ such that there exists a family of $k$ quantum effects in dimension 3 (i.e. $3 \times 3$ complex matrices $M_j$ with $0 \le M_j \le I$) that are jointly incompatible on $\mathbb{C}^3$, but whose compressions $P M_j P$ to any proper subspace (given by an orthogonal projection $P$ with $\operatorname{rank}(P) < 3$) are jointly compatible.

Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.

The proof, explained

The theorem is proved by constructing a family of k = 3 binary measurements on C^3 obtained by steering a two-qutrit positive partial transpose (PPT) Bell state using three measurement settings on Alice's side. The master reduction lemma open_problem_693_three_dim_fewer_than_nine_if_lemmas establishes that these 3 measurements are globally incompatible on C^3 and compatible on every 2-dimensional subspace via Gram matrix normalization, simultaneous diagonalization, and hidden-variable model completions. The remaining goal—joint measurability on subspaces of dimension at most 1—holds automatically because compressions of valid binary effects to rank-one subspaces trivially commute and admit a parent POVM.

  1. Every rank-2 orthogonal projection on C^3 admits an isometry factor producing a full-rank (rank 3) steered vector Gram matrix.

    Because any rank-2 orthogonal projection P on C^3 factors as U U† for a 3×2 isometry U (by rank_two_qutrit_projection_isometry), ppt_qutrit_source_projection_basis_from_isometry_factor establishes that U can be chosen such that the steered vectors with respect to the PPT Bell state have maximal rank 3.

    lines 3–4

    uses ppt_qutrit_source_projection_basis_from_isometry_factor, rank_two_qutrit_projection_isometry

      have hselect := ppt_qutrit_source_projection_basis_from_isometry_factor
        rank_two_qutrit_projection_isometry
  2. For every rank-2 orthogonal projection, the steered vectors admit a normalized Gram decomposition satisfying row-isometry and partial-transpose symmetry.

    Using the full-rank isometry choice (hselect), the partial transpose invariance of the PPT Bell state (ppt_qutrit_bell_state_partial_transpose_invariant), and the existence of an orthogonal normalizer for any positive definite real symmetric matrix (real_positive_qutrit_gram_normalizer), ppt_qutrit_source_normalized_gram_if_three_lemmas yields a factorized representation (F, T) satisfying the row-isometry condition V V† = 1 and block symmetry (F F†)_{ip,jq} = (F F†)_{jp,iq}.

    lines 5–6

    uses ppt_qutrit_source_normalized_gram_if_three_lemmas, ppt_qutrit_bell_state_partial_transpose_invariant, real_positive_qutrit_gram_normalizer

      have hgram := ppt_qutrit_source_normalized_gram_if_three_lemmas hselect
        ppt_qutrit_bell_state_partial_transpose_invariant real_positive_qutrit_gram_normalizer
  3. The problem reduces to verifying that the three steered effects are jointly measurable on subspaces of dimension at most 1.

    The master reduction lemma open_problem_693_three_dim_fewer_than_nine_if_lemmas proves that k = 3 binary measurements (which is strictly fewer than 9) witness the subspace-compatible incompatible family property, provided that the rank-2 Gram decomposition holds (hgram), Alice's measurement effects are valid (ppt_qutrit_bell_alice_binary_effects_valid), Bob's marginal is normalized (ppt_qutrit_steered_normalizer_marginal_identity), the codimension-one complement property holds (three_orthonormal_rows_codimension_one), the rank-four phase completion holds (rank_four_ppt_normal_completion_phase), commuting symmetric matrices simultaneously diagonalize (real_symmetric_commuting_pair_orthogonal_diagonalization), the Schur decomposition holds (qutrit_fixed_phase_schur_four_product_decomposition), the weighted hidden-variable model yields a parent POVM (finite_weighted_three_binary_hidden_model_has_parent), Bob's steered effects are valid (ppt_qutrit_steered_effect_is_valid), and joint measurability holds on subspaces of dimension at most 1. Applying this lemma discharges all requirements except subspace compatibility for rank at most 1.

    lines 7–15

    uses open_problem_693_three_dim_fewer_than_nine_if_lemmas, ppt_qutrit_bell_alice_binary_effects_valid, ppt_qutrit_steered_normalizer_marginal_identity, three_orthonormal_rows_codimension_one, rank_four_ppt_normal_completion_phase, real_symmetric_commuting_pair_orthogonal_diagonalization, qutrit_fixed_phase_schur_four_product_decomposition, finite_weighted_three_binary_hidden_model_has_parent, ppt_qutrit_steered_effect_is_valid

      apply open_problem_693_three_dim_fewer_than_nine_if_lemmas hgram
        ppt_qutrit_bell_alice_binary_effects_valid
        ppt_qutrit_steered_normalizer_marginal_identity
        three_orthonormal_rows_codimension_one
        rank_four_ppt_normal_completion_phase
        real_symmetric_commuting_pair_orthogonal_diagonalization
        qutrit_fixed_phase_schur_four_product_decomposition
        (finite_weighted_three_binary_hidden_model_has_parent 2 4)
        ppt_qutrit_steered_effect_is_valid
  4. On any subspace of dimension at most 1, the compressed measurements are jointly measurable.

    Let P be an orthogonal projection of rank at most 1. By rank_one_qutrit_compression_three_binary_parent, compressing any three valid binary quantum effects (here ppt_qutrit_steered_effect, whose validity is certified by ppt_qutrit_steered_effect_is_valid) to a subspace of dimension ≤ 1 always admits a joint parent POVM. This completes the remaining hypothesis.

    lines 16–18

    uses rank_one_qutrit_compression_three_binary_parent, ppt_qutrit_steered_effect_is_valid

      intro P hP hPP hrank
      exact rank_one_qutrit_compression_three_binary_parent
        ppt_qutrit_steered_effect P ppt_qutrit_steered_effect_is_valid hP hPP hrank
ppt_qutrit_source_projection_basis_from_isometry_factor
If every rank-2 orthogonal projection on C^3 factors through a 2D isometry, then for each such projection there exists an isometry whose steered vectors under the PPT Bell state have full rank 3.
rank_two_qutrit_projection_isometry
Every rank-2 orthogonal projection P on C^3 factors as P = U U† for a 3×2 isometry U satisfying U† U = I_2.
ppt_qutrit_source_normalized_gram_if_three_lemmas
Combining the full-rank steered isometry selection, PPT state partial transpose invariance, and real Gram normalizability produces a normalized Gram decomposition (F, T) for any rank-2 projection.
ppt_qutrit_bell_state_partial_transpose_invariant
The two-qutrit PPT Bell state density matrix is invariant under partial transposition on the first system.
real_positive_qutrit_gram_normalizer
Every positive definite real symmetric 3×3 matrix A can be normalized to the identity by an invertible symmetric matrix H such that H A H = I_3.
open_problem_693_three_dim_fewer_than_nine_if_lemmas
Reduces the existence of a family of strictly fewer than 9 subspace-compatible but globally incompatible binary measurements on C^3 to ten structural conditions: rank-2 Gram decomposition, validity of Alice's effects, marginal normalization, codimension-one completion, phase completion, commuting matrix diagonalization, Schur decomposition, hidden-variable parent POVM existence, validity of Bob's effects, and compatibility on subspaces of rank at most 1.
ppt_qutrit_bell_alice_binary_effects_valid
Each of Alice's three measurement settings defines a valid quantum effect (a positive semidefinite matrix bounded above by the identity).
ppt_qutrit_steered_normalizer_marginal_identity
The marginal state of the two-qutrit PPT Bell state, when conjugated by the steered normalizer matrix, equals the identity matrix.
three_orthonormal_rows_codimension_one
For any 3×4 matrix with orthonormal rows, the orthogonal projection onto the 1-dimensional orthogonal complement of its row space is given by a rank-one outer product w w†.
rank_four_ppt_normal_completion_phase
For any two 3×3 real matrices and any two linearly independent vectors, there exists a 2D phase angle rotation and real scalars satisfying the rank-four normal completion equation.
real_symmetric_commuting_pair_orthogonal_diagonalization
Any pair of commuting real symmetric n×n matrices can be simultaneously diagonalized by a single orthogonal matrix.
qutrit_fixed_phase_schur_four_product_decomposition
The identity, an orthogonally diagonalized matrix B, and B† B + u u† can be decomposed simultaneously into a sum of four real rank-one projections weighted by factorized coefficients.
finite_weighted_three_binary_hidden_model_has_parent
Any family of three binary measurements generated by a finite weighted hidden-variable model with response probabilities bounded by weights admits a joint parent POVM.
ppt_qutrit_steered_effect_is_valid
Each of the three steered effects on Bob's system is a valid quantum effect, satisfying 0 ≤ M_j ≤ I_3.
rank_one_qutrit_compression_three_binary_parent
On any subspace of dimension at most 1, the compression of any three valid binary quantum effects automatically admits a joint parent POVM.

Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.

Lean source view module on GitHub

/-- Open Problem OP-693: existence of a family of strictly fewer than 9 binary measurements on C^3 incompatible globally but compatible on all proper subspaces. -/theorem open_problem_693_three_dim_fewer_than_nine : ∃ k < 9, has_subspace_compatible_incompatible_family 3 k := by  have hselect := ppt_qutrit_source_projection_basis_from_isometry_factor    rank_two_qutrit_projection_isometry  have hgram := ppt_qutrit_source_normalized_gram_if_three_lemmas hselect    ppt_qutrit_bell_state_partial_transpose_invariant real_positive_qutrit_gram_normalizer  apply open_problem_693_three_dim_fewer_than_nine_if_lemmas hgram    ppt_qutrit_bell_alice_binary_effects_valid    ppt_qutrit_steered_normalizer_marginal_identity    three_orthonormal_rows_codimension_one    rank_four_ppt_normal_completion_phase    real_symmetric_commuting_pair_orthogonal_diagonalization    qutrit_fixed_phase_schur_four_product_decomposition    (finite_weighted_three_binary_hidden_model_has_parent 2 4)    ppt_qutrit_steered_effect_is_valid  intro P hP hPP hrank  exact rank_one_qutrit_compression_three_binary_parent    ppt_qutrit_steered_effect P ppt_qutrit_steered_effect_is_valid hP hPP hrank
See it in the graphDiscuss this resultChallenge it