AFTD Auto-Formalizing Theoretical Domains

rank_two_frame_compression_no_parent

theoremverified

For r greater than one, no positive 2-by-2 parent has the displayed total matrix and three displayed marginals.

Statement

theorem rank_two_frame_compression_no_parent (r : ℝ) (hr : 1 < r) :
  ¬ ∃ G : (Fin 3 → Bool) → Matrix (Fin 2) (Fin 2) ℂ,
    (∀ s, (G s).PosSemidef) ∧
    (∑ s, G s = !![2, 1; 1, 2]) ∧
    (∑ s, if s 0 then G s else 0) = !![1, 0; 0, 0] ∧
    (∑ s, if s 1 then G s else 0) = 1 ∧
    (∑ s, if s 2 then G s else 0) =
      (1/2 : ℝ) • (!![2, 1; 1, 2] - r • !![1, 1; 1, 1])
Source
Helper excluding a four-outcome rank-one-frame construction for OP-693; exact compressed matrices in experiment 22918.
Verified
12 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 2
  • subspace_incompatible_family_of_mic_dataFor all d k : ℕ, let g : Fin k → Fin d → ℂ, write G i = Matrix.vecMulVec (g i) (star (g i)) (the rank-one positive matrix g_i g_i^*), let…
  • FThe prime field $\mathrm{GF}(p)$, realized as $\mathbb{Z}/p\mathbb{Z}$ via ZMod.

The proof, explained

The proof proceeds by contradiction: assuming a positive semidefinite parent family G exists, each matrix G(s) is bounded above in the Loewner order by rank-one marginals or their complements. Because any positive semidefinite matrix dominated by a rank-one matrix v v* is a non-negative scalar multiple of v v*, each G(s) must be proportional to F_3, F_2, or F_0, and must vanish when s 0 = true and s 1 = false. This forces the real part of the off-diagonal entry (G(s)) 0 1 to be non-negative for every s, which implies that the third marginal sum must also have a non-negative (0, 1)-entry. Evaluating this entry directly from the given marginal gives (1 - r) / 2 ≥ 0, which contradicts r > 1.

  1. The outer products v_0 v_0*, v_2 v_2*, and v_3 v_3* equal the rank-one matrices F_0, F_2, and F_3 respectively. automation

    By extensionality on 2-by-2 matrices, each entry is checked across the finite index cases {0, 1} × {0, 1} and evaluated by numeric computation.

    lines 13–18

    uses vecMulVec_apply

    private lemma in767_outer0 : vecMulVec in767_v0 (star in767_v0) = in767_F0 := by
      ext i j; fin_cases i <;> fin_cases j <;> norm_num [vecMulVec_apply,in767_v0,in767_F0]
    private lemma in767_outer2 : vecMulVec in767_v2 (star in767_v2) = in767_F2 := by
      ext i j; fin_cases i <;> fin_cases j <;> norm_num [vecMulVec_apply,in767_v2,in767_F2]
    private lemma in767_outer3 : vecMulVec in767_v3 (star in767_v3) = in767_F3 := by
      ext i j; fin_cases i <;> fin_cases j <;> norm_num [vecMulVec_apply,in767_v3,in767_F3]
  2. The matrix algebraic relations F_S - I = F_0 and I - F_3 = F_2 hold. automation

    By checking each matrix entry over the four index combinations using the definition of the identity matrix.

    lines 20–23

    uses Matrix.one_apply

    private lemma in767_Ssub : in767_FS - 1 = in767_F0 := by
      ext i j; fin_cases i <;> fin_cases j <;> norm_num [in767_FS,in767_F0,Matrix.one_apply]
    private lemma in767_Isub : (1 : Matrix (Fin 2) (Fin 2) ℂ) - in767_F3 = in767_F2 := by
      ext i j; fin_cases i <;> fin_cases j <;> norm_num [in767_F3,in767_F2,Matrix.one_apply]
  3. Assume for contradiction that a family G of positive semidefinite matrices exists satisfying the total sum and three marginals.

    We introduce the assumed witness G with its properties: each G(s) is positive semidefinite, their total sum is F_S, and their marginals along coordinate directions 0, 1, and 2 are F_3, I, and (1/2) • (F_S - r • F_0), respectively.

    lines 32–33

      classical
      rintro ⟨G, hG, hsum, h0, h1, h2⟩
  4. For any index s with s 0 = true, F_3 - G(s) is positive semidefinite.

    Isolating G(s) from the marginal sum over all s with s 0 = true, the difference F_3 - G(s) equals the sum of the remaining terms. Since each term G(a) is positive semidefinite (or 0 when a 0 is false), their sum is positive semidefinite.

    lines 34–42

    uses Finset.add_sum_erase, Finset.mem_univ, if_pos, add_sub_cancel_left, Matrix.posSemidef_sum, Matrix.PosSemidef.zero

      have h0le : ∀ s, s 0 = true → (in767_F3 - G s).PosSemidef := by
        intro s hs
        have he := h0
        rw [← Finset.add_sum_erase _ _ (Finset.mem_univ s), if_pos hs] at he
        rw [← he, add_sub_cancel_left]
        exact Matrix.posSemidef_sum _ fun a _ => by
          split_ifs
          · exact hG a
          · exact Matrix.PosSemidef.zero
  5. The sum of G(s) over all s with s 1 = false equals F_0.

    The sum over s with s 1 = false is obtained by subtracting the marginal sum where s 1 = true (which equals I) from the total sum (which equals F_S). By the algebraic relation F_S - I = F_0, this sum is F_0.

    lines 43–50

    uses Finset.sum_sub_distrib, Finset.sum_congr, in767_Ssub

      let F : (Fin 3 → Bool) → Matrix (Fin 2) (Fin 2) ℂ := fun s => if s 1 then 0 else G s
      have hFsum : ∑ s, F s = in767_F0 := by
        calc
          ∑ s, F s = (∑ s, G s) - ∑ s, (if s 1 then G s else 0) := by
            rw [← Finset.sum_sub_distrib]
            refine Finset.sum_congr rfl fun s _ => ?_
            cases hs : s 1 <;> simp [F,hs]
          _ = in767_F0 := by rw [hsum,h1,in767_Ssub]
  6. For any index s with s 1 = false, F_0 - G(s) is positive semidefinite.

    Isolating G(s) from the sum over all s with s 1 = false, the difference F_0 - G(s) is the sum over the remaining indices of terms that are each positive semidefinite (or 0), hence positive semidefinite.

    lines 51–60

    uses Finset.add_sum_erase, Finset.mem_univ, Bool.false_eq_true, if_false, add_sub_cancel_left, Matrix.posSemidef_sum, Matrix.PosSemidef.zero

      have h1le : ∀ s, s 1 = false → (in767_F0 - G s).PosSemidef := by
        intro s hs
        have he := hFsum
        rw [← Finset.add_sum_erase _ _ (Finset.mem_univ s)] at he
        simp only [F,hs,Bool.false_eq_true,if_false] at he
        rw [← he, add_sub_cancel_left]
        exact Matrix.posSemidef_sum _ fun a _ => by
          split_ifs
          · exact Matrix.PosSemidef.zero
          · exact hG a
  7. If s 0 = true and s 1 = false, then G(s) = 0.

    Such a G(s) is positive semidefinite and Loewner-dominated by both rank-one matrices F_3 = v_3 v_3* and F_0 = v_0 v_0*. Dominance by a rank-one matrix implies G(s) = a • F_3 and G(s) = b • F_0 for non-negative scalars a and b. Comparing the off-diagonal (0, 1)-entry shows (a • F_3) 0 1 = 0 while (b • F_0) 0 1 = b, forcing b = 0 and thus G(s) = 0.

    lines 61–73

    uses in767_outer3, in767_outer0, rank_one_psd_dominated_eq_smul, congrArg, zero_smul

      have hzero : ∀ s, s 0 = true → s 1 = false → G s = 0 := by
        intro s hs0 hs1
        have hA := h0le s hs0
        have hB := h1le s hs1
        rw [← in767_outer3] at hA
        rw [← in767_outer0] at hB
        obtain ⟨a, ha, ea⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v3 (hG s) hA
        obtain ⟨b, hb, eb⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v0 (hG s) hB
        rw [in767_outer3] at ea
        rw [in767_outer0] at eb
        have e := congrArg (fun X : Matrix (Fin 2) (Fin 2) ℂ => (X 0 1).re) (ea.symm.trans eb)
        have hb0 : b = 0 := by simpa [in767_F0,in767_F3] using e.symm
        rw [eb, hb0, zero_smul]
  8. The sum of G(s) over all s with s 0 = false and s 1 = true equals F_2.

    Subtracting the marginal sum where s 0 = true from the marginal sum where s 1 = true yields I - F_3 = F_2. In this difference, the terms where s 0 = true and s 1 = true cancel, the terms where s 0 = true and s 1 = false vanish because G(s) = 0, and the terms where both are false are zero, leaving precisely the sum over indices where s 0 = false and s 1 = true.

    lines 74–86

    uses Finset.sum_sub_distrib, Finset.sum_congr, in767_Isub

      let R : (Fin 3 → Bool) → Matrix (Fin 2) (Fin 2) ℂ := fun s => if s 1 then (if s 0 then 0 else G s) else 0
      have hRsum : ∑ s, R s = in767_F2 := by
        calc
          ∑ s, R s = (∑ s, if s 1 then G s else 0) - ∑ s, (if s 0 then G s else 0) := by
            rw [← Finset.sum_sub_distrib]
            refine Finset.sum_congr rfl fun s _ => ?_
            by_cases hs0 : s 0 = true <;> by_cases hs1 : s 1 = true
            · simp [R,hs0,hs1]
            · have hz := hzero s hs0 (by simpa using hs1)
              simp [R,hs0,hs1,hz]
            · simp [R,hs0,hs1]
            · simp [R,hs0,hs1]
          _ = in767_F2 := by rw [h1,h0,in767_Isub]
  9. For any index s with s 0 = false and s 1 = true, F_2 - G(s) is positive semidefinite.

    Isolating G(s) from the sum over all s with s 0 = false and s 1 = true, the difference F_2 - G(s) is a sum of positive semidefinite matrices (or zeros), hence positive semidefinite.

    lines 87–97

    uses Finset.add_sum_erase, Finset.mem_univ, if_true, Bool.false_eq_true, if_false, add_sub_cancel_left, Matrix.posSemidef_sum, Matrix.PosSemidef.zero

      have h2le : ∀ s, s 0 = false → s 1 = true → (in767_F2 - G s).PosSemidef := by
        intro s hs0 hs1
        have he := hRsum
        rw [← Finset.add_sum_erase _ _ (Finset.mem_univ s)] at he
        simp only [R,hs0,hs1,if_true,Bool.false_eq_true,if_false] at he
        rw [← he, add_sub_cancel_left]
        exact Matrix.posSemidef_sum _ fun a _ => by
          split_ifs
          · exact Matrix.PosSemidef.zero
          · exact hG a
          · exact Matrix.PosSemidef.zero
  10. For every index s, the real part of the (0, 1)-entry of G(s) is non-negative.

    We examine the three possible cases for the pair (s 0, s 1): if s 0 = true, G(s) is a multiple of F_3, whose (0, 1)-entry is 0; if s 0 = false and s 1 = true, G(s) is a multiple of F_2, whose (0, 1)-entry is 0; if both are false, G(s) is a non-negative multiple of F_0, whose (0, 1)-entry is 1 ≥ 0. In all cases, Re((G(s)) 0 1) ≥ 0.

    lines 98–121

    uses in767_outer3, rank_one_psd_dominated_eq_smul, in767_outer2, in767_outer0

      have hentry : ∀ s, 0 ≤ (G s 0 1).re := by
        intro s
        by_cases hs0 : s 0 = true
        · have hA := h0le s hs0
          rw [← in767_outer3] at hA
          obtain ⟨a,ha,ea⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v3 (hG s) hA
          rw [in767_outer3] at ea
          rw [ea]
          simp [in767_F3]
        · have hs0' : s 0 = false := by simpa using hs0
          by_cases hs1 : s 1 = true
          · have hA := h2le s hs0' hs1
            rw [← in767_outer2] at hA
            obtain ⟨a,ha,ea⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v2 (hG s) hA
            rw [in767_outer2] at ea
            rw [ea]
            simp [in767_F2]
          · have hs1' : s 1 = false := by simpa using hs1
            have hA := h1le s hs1'
            rw [← in767_outer0] at hA
            obtain ⟨a,ha,ea⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v0 (hG s) hA
            rw [in767_outer0] at ea
            rw [ea]
            simpa [in767_F0] using ha
  11. The marginal sum along coordinate 2 yields a contradiction r ≤ 1. automation

    Since each G(s) has non-negative real part in its (0, 1)-entry, the (0, 1)-entry of the sum over s with s 2 = true has non-negative real part. But by assumption, this sum equals (1/2) • (F_S - r • F_0), whose (0, 1)-entry has real part (1 - r) / 2. Hence 0 ≤ (1 - r) / 2, which implies r ≤ 1, contradicting 1 < r.

    lines 122–127

    uses Matrix.sum_apply, Complex.re_sum, Finset.sum_nonneg

      have hnonneg : 0 ≤ ((∑ s : (Fin 3 → Bool), if s 2 then G s else 0) 0 1).re := by
        simp only [Matrix.sum_apply, Complex.re_sum]
        exact Finset.sum_nonneg fun s _ => by split_ifs <;> simp_all
      rw [h2] at hnonneg
      norm_num [in767_FS,in767_F0] at hnonneg
      linarith
  12. No such parent matrix family G exists for r > 1.

    The main theorem rank_two_frame_compression_no_parent is an immediate restatement of in767_frame_no_parent with explicit matrix literals.

    lines 129–137

    uses in767_frame_no_parent

    theorem rank_two_frame_compression_no_parent (r : ℝ) (hr : 1 < r) :
      ¬ ∃ G : (Fin 3 → Bool) → Matrix (Fin 2) (Fin 2) ℂ,
        (∀ s, (G s).PosSemidef) ∧
        (∑ s, G s = !![2, 1; 1, 2]) ∧
        (∑ s, if s 0 then G s else 0) = !![1, 0; 0, 0] ∧
        (∑ s, if s 1 then G s else 0) = 1 ∧
        (∑ s, if s 2 then G s else 0) =
          (1/2 : ℝ) • (!![2, 1; 1, 2] - r • !![1, 1; 1, 1]) := by
      exact in767_frame_no_parent r hr
vecMulVec_apply
The (i, j)-entry of the outer product of vectors w and v is w i * v j.
Matrix.one_apply
The (i, j)-entry of the identity matrix is 1 if i = j and 0 otherwise.
Finset.add_sum_erase
A finite sum over a set s can be decomposed into an element f a and the sum over s without a.
Finset.mem_univ
Every element of a finite type belongs to the universal finset.
if_pos
If a condition c holds, the conditional expression (if c then t else e) equals t.
add_sub_cancel_left
In an additive group, subtracting a from a + b leaves b.
Matrix.posSemidef_sum
A finite sum of positive semidefinite matrices is positive semidefinite.
Matrix.PosSemidef.zero
The zero matrix is positive semidefinite.
Finset.sum_sub_distrib
A finite sum of differences equals the difference of the finite sums.
Finset.sum_congr
Two finite sums are equal if their index sets are equal and their summands agree at every index.
in767_Ssub
The matrix difference F_S - I equals F_0.
Bool.false_eq_true
The boolean value false does not equal true.
if_false
A conditional expression with a false condition evaluates to its second branch.
if_true
A conditional expression with a true condition evaluates to its first branch.
in767_outer3
The outer product of v_3 = (1, 0) with its conjugate transpose equals the matrix F_3 = !![1, 0; 0, 0].
in767_outer0
The outer product of v_0 = (1, 1) with its conjugate transpose equals the all-ones matrix F_0 = !![1, 1; 1, 1].
in767_outer2
The outer product of v_2 = (0, 1) with its conjugate transpose equals the matrix F_2 = !![0, 0; 0, 1].
rank_one_psd_dominated_eq_smul
If a positive semidefinite matrix A is dominated by the rank-one matrix v v*, then A is a non-negative scalar multiple of v v*.
congrArg
Applying a function to equal arguments produces equal results.
zero_smul
Scalar multiplication of any element by zero gives zero.
in767_Isub
The matrix difference I - F_3 equals F_2.
Matrix.sum_apply
The (i, j)-entry of a finite sum of matrices equals the sum of their (i, j)-entries.
Complex.re_sum
The real part of a finite sum of complex numbers is the sum of their real parts.
Finset.sum_nonneg
A finite sum of non-negative real numbers is non-negative.
in767_frame_no_parent
For r > 1, no family of positive semidefinite matrices matches the total matrix F_S and the three prescribed marginals.

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 Matrixopen scoped ComplexOrder Matrix BigOperatorsset_option maxHeartbeats 4000000private noncomputable def in767_F0 : Matrix (Fin 2) (Fin 2) ℂ := !![1, 1; 1, 1]private noncomputable def in767_F2 : Matrix (Fin 2) (Fin 2) ℂ := !![0, 0; 0, 1]private noncomputable def in767_F3 : Matrix (Fin 2) (Fin 2) ℂ := !![1, 0; 0, 0]private noncomputable def in767_FS : Matrix (Fin 2) (Fin 2) ℂ := !![2, 1; 1, 2]private noncomputable def in767_v0 : Fin 2 → ℂ := ![1,1]private noncomputable def in767_v2 : Fin 2 → ℂ := ![0,1]private noncomputable def in767_v3 : Fin 2 → ℂ := ![1,0]private lemma in767_outer0 : vecMulVec in767_v0 (star in767_v0) = in767_F0 := by  ext i j; fin_cases i <;> fin_cases j <;> norm_num [vecMulVec_apply,in767_v0,in767_F0]private lemma in767_outer2 : vecMulVec in767_v2 (star in767_v2) = in767_F2 := by  ext i j; fin_cases i <;> fin_cases j <;> norm_num [vecMulVec_apply,in767_v2,in767_F2]private lemma in767_outer3 : vecMulVec in767_v3 (star in767_v3) = in767_F3 := by  ext i j; fin_cases i <;> fin_cases j <;> norm_num [vecMulVec_apply,in767_v3,in767_F3]private lemma in767_Ssub : in767_FS - 1 = in767_F0 := by  ext i j; fin_cases i <;> fin_cases j <;> norm_num [in767_FS,in767_F0,Matrix.one_apply]private lemma in767_Isub : (1 : Matrix (Fin 2) (Fin 2) ℂ) - in767_F3 = in767_F2 := by  ext i j; fin_cases i <;> fin_cases j <;> norm_num [in767_F3,in767_F2,Matrix.one_apply]private theorem in767_frame_no_parent (r : ℝ) (hr : 1 < r) :  ¬ ∃ G : (Fin 3 → Bool) → Matrix (Fin 2) (Fin 2) ℂ,    (∀ s, (G s).PosSemidef) ∧    (∑ s, G s = in767_FS) ∧    (∑ s, if s 0 then G s else 0) = in767_F3 ∧    (∑ s, if s 1 then G s else 0) = 1 ∧    (∑ s, if s 2 then G s else 0) = (1/2 : ℝ) • (in767_FS - r • in767_F0) := by  classical  rintro ⟨G, hG, hsum, h0, h1, h2⟩  have h0le : ∀ s, s 0 = true → (in767_F3 - G s).PosSemidef := by    intro s hs    have he := h0    rw [← Finset.add_sum_erase _ _ (Finset.mem_univ s), if_pos hs] at he    rw [← he, add_sub_cancel_left]    exact Matrix.posSemidef_sum _ fun a _ => by      split_ifs      · exact hG a      · exact Matrix.PosSemidef.zero  let F : (Fin 3 → Bool) → Matrix (Fin 2) (Fin 2) ℂ := fun s => if s 1 then 0 else G s  have hFsum : ∑ s, F s = in767_F0 := by    calc      ∑ s, F s = (∑ s, G s) - ∑ s, (if s 1 then G s else 0) := by        rw [← Finset.sum_sub_distrib]        refine Finset.sum_congr rfl fun s _ => ?_        cases hs : s 1 <;> simp [F,hs]      _ = in767_F0 := by rw [hsum,h1,in767_Ssub]  have h1le : ∀ s, s 1 = false → (in767_F0 - G s).PosSemidef := by    intro s hs    have he := hFsum    rw [← Finset.add_sum_erase _ _ (Finset.mem_univ s)] at he    simp only [F,hs,Bool.false_eq_true,if_false] at he    rw [← he, add_sub_cancel_left]    exact Matrix.posSemidef_sum _ fun a _ => by      split_ifs      · exact Matrix.PosSemidef.zero      · exact hG a  have hzero : ∀ s, s 0 = true → s 1 = false → G s = 0 := by    intro s hs0 hs1    have hA := h0le s hs0    have hB := h1le s hs1    rw [← in767_outer3] at hA    rw [← in767_outer0] at hB    obtain ⟨a, ha, ea⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v3 (hG s) hA    obtain ⟨b, hb, eb⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v0 (hG s) hB    rw [in767_outer3] at ea    rw [in767_outer0] at eb    have e := congrArg (fun X : Matrix (Fin 2) (Fin 2) ℂ => (X 0 1).re) (ea.symm.trans eb)    have hb0 : b = 0 := by simpa [in767_F0,in767_F3] using e.symm    rw [eb, hb0, zero_smul]  let R : (Fin 3 → Bool) → Matrix (Fin 2) (Fin 2) ℂ := fun s => if s 1 then (if s 0 then 0 else G s) else 0  have hRsum : ∑ s, R s = in767_F2 := by    calc      ∑ s, R s = (∑ s, if s 1 then G s else 0) - ∑ s, (if s 0 then G s else 0) := by        rw [← Finset.sum_sub_distrib]        refine Finset.sum_congr rfl fun s _ => ?_        by_cases hs0 : s 0 = true <;> by_cases hs1 : s 1 = true        · simp [R,hs0,hs1]        · have hz := hzero s hs0 (by simpa using hs1)          simp [R,hs0,hs1,hz]        · simp [R,hs0,hs1]        · simp [R,hs0,hs1]      _ = in767_F2 := by rw [h1,h0,in767_Isub]  have h2le : ∀ s, s 0 = false → s 1 = true → (in767_F2 - G s).PosSemidef := by    intro s hs0 hs1    have he := hRsum    rw [← Finset.add_sum_erase _ _ (Finset.mem_univ s)] at he    simp only [R,hs0,hs1,if_true,Bool.false_eq_true,if_false] at he    rw [← he, add_sub_cancel_left]    exact Matrix.posSemidef_sum _ fun a _ => by      split_ifs      · exact Matrix.PosSemidef.zero      · exact hG a      · exact Matrix.PosSemidef.zero  have hentry : ∀ s, 0 ≤ (G s 0 1).re := by    intro s    by_cases hs0 : s 0 = true    · have hA := h0le s hs0      rw [← in767_outer3] at hA      obtain ⟨a,ha,ea⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v3 (hG s) hA      rw [in767_outer3] at ea      rw [ea]      simp [in767_F3]    · have hs0' : s 0 = false := by simpa using hs0      by_cases hs1 : s 1 = true      · have hA := h2le s hs0' hs1        rw [← in767_outer2] at hA        obtain ⟨a,ha,ea⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v2 (hG s) hA        rw [in767_outer2] at ea        rw [ea]        simp [in767_F2]      · have hs1' : s 1 = false := by simpa using hs1        have hA := h1le s hs1'        rw [← in767_outer0] at hA        obtain ⟨a,ha,ea⟩ := rank_one_psd_dominated_eq_smul (G s) in767_v0 (hG s) hA        rw [in767_outer0] at ea        rw [ea]        simpa [in767_F0] using ha  have hnonneg : 0 ≤ ((∑ s : (Fin 3 → Bool), if s 2 then G s else 0) 0 1).re := by    simp only [Matrix.sum_apply, Complex.re_sum]    exact Finset.sum_nonneg fun s _ => by split_ifs <;> simp_all  rw [h2] at hnonneg  norm_num [in767_FS,in767_F0] at hnonneg  linariththeorem rank_two_frame_compression_no_parent (r : ℝ) (hr : 1 < r) :  ¬ ∃ G : (Fin 3 → Bool) → Matrix (Fin 2) (Fin 2) ℂ,    (∀ s, (G s).PosSemidef) ∧    (∑ s, G s = !![2, 1; 1, 2]) ∧    (∑ s, if s 0 then G s else 0) = !![1, 0; 0, 0] ∧    (∑ s, if s 1 then G s else 0) = 1 ∧    (∑ s, if s 2 then G s else 0) =      (1/2 : ℝ) • (!![2, 1; 1, 2] - r • !![1, 1; 1, 1]) := by  exact in767_frame_no_parent r hr
See it in the graphDiscuss this resultChallenge it