matrix_q_permanent_strict_mono_of_order_le_three
If A is a non-diagonal real positive definite matrix of order n ≤ 3, then the function mapping q to the q-permanent of A is strictly increasing on all of ℝ.
Statement
theorem matrix_q_permanent_strict_mono_of_order_le_three (n : ℕ) (hn : n ≤ 3) (A : Matrix (Fin n) (Fin n) ℝ) (hpos : A.PosDef) (hnondiag : ¬ A.IsDiag) : StrictMono (fun q : ℝ => matrix_q_permanent q A)
- Formalizes
- arXiv:2611.14694 Proposition 5.1 -- Real positive definite counterexamples to a monotonicity conjecture for the q-permanent
- Verified
- 16 Nov 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 5
- matrix_q_permanentThe q-permanent of an n-by-n matrix M over a commutative semiring R at parameter q in R is the sum over all permutations sigma of Fin n of…
- matrix_q_permanent_diagonal_conjFor any natural number n, any n-by-n matrix A over a commutative ring R, any diagonal matrix D = Matrix.diagonal d with d : Fin n → R, and…
- matrix_q_permanent_two_of_is_symmFor any symmetric 2-by-2 matrix A over a commutative ring R and any parameter q in R, the q-permanent of A equals A 0 0 * A 1 1 + q * (A 0…
- matrix_q_permanent_three_of_diag_oneFor any real symmetric 3-by-3 matrix A with all diagonal entries equal to 1, writing x = A 0 1, y = A 1 2, and z = A 0 2, the q-permanent…
- cubic_poly_strict_mono_of_minors_lt_oneFor any real numbers x, y, z satisfying x^2 < 1, y^2 < 1, and (x, y, z) ≠ (0, 0, 0), the cubic polynomial function mapping q to 1 + (x^2 +…
Read back from the Lean
For any natural number n ≤ 3 and any n × n real matrix A, if A is positive definite and A is not a diagonal matrix, then the function mapping q ∈ ℝ to the q-permanent of A is strictly increasing on all of ℝ.
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 proceeds by case analysis on the dimension n ≤ 3. The cases n = 0 and n = 1 are ruled out because any matrix of order 0 or 1 is trivially diagonal, contradicting the non-diagonality hypothesis. For n = 2, symmetry implies that the q-permanent is an affine function in q whose slope is the square of the non-zero off-diagonal entry, hence strictly positive. For n = 3, symmetrically scaling A by the inverse square roots of its diagonal entries produces a symmetric matrix B with ones on the diagonal, whose q-permanent is a cubic polynomial. Strict positivity of 2 × 2 principal minors implies that the polynomial's coefficients satisfy the hypotheses of a monotonicity lemma, and strict monotonicity transfers back from B to A via the positive conjugation factor (det D)^2.
The matrix A is symmetric, and its q-permanent is the linear function q ↦ A 0 0 * A 1 1 + q * (A 0 1)^2.
Since A is positive definite over the reals, it is Hermitian and therefore symmetric. The explicit formula for 2 × 2 symmetric matrices then gives the q-permanent as an affine function in q.
uses
Matrix.isHermitian_iff_isSymm,matrix_q_permanent_two_of_is_symmhave hsymm : A.IsSymm := by rw [← Matrix.isHermitian_iff_isSymm] exact hpos.isHermitian have hperm : ∀ q, matrix_q_permanent q A = A 0 0 * A 1 1 + q * (A 0 1) ^ 2 := fun q => matrix_q_permanent_two_of_is_symm q A hsymm
The off-diagonal entry A 0 1 is non-zero, so its square (A 0 1)^2 is strictly positive.
If A 0 1 were zero, symmetry would imply A 1 0 = 0, which would make A diagonal, contradicting the assumption ¬ A.IsDiag. Since A 0 1 ≠ 0, its square is strictly positive.
uses
sq_pos_of_ne_zerohave hne : A 0 1 ≠ 0 := by intro hzero apply hnondiag intro i j hij fin_cases i <;> fin_cases j · contradiction · exact hzero · rw [← hsymm.apply]; exact hzero · contradiction have hsq_pos : 0 < (A 0 1) ^ 2 := sq_pos_of_ne_zero hne
The function q ↦ matrix_q_permanent q A is strictly increasing.
For x < y, comparing A 0 0 * A 1 1 + x * (A 0 1)^2 with A 0 0 * A 1 1 + y * (A 0 1)^2 yields a strict inequality because the slope (A 0 1)^2 is strictly positive.
intro x y hxy dsimp only rw [hperm x, hperm y] nlinarith
A is symmetric, and symmetrically scaling A by D = diag((A k k)^(-1/2)) yields a symmetric matrix B = D A D with all diagonal entries equal to 1.
Positive definiteness ensures that A is Hermitian (hence symmetric) and each diagonal entry A k k is strictly positive, making the scaling factors d k = (A k k)^(-1/2) well-defined and positive. The conjugated matrix B = D A D is symmetric by matrix algebra, and each diagonal entry simplifies to d i * A i i * d i = A i i / A i i = 1.
uses
Matrix.isHermitian_iff_isSymm,Matrix.mul_diagonal,Matrix.diagonal_mul,Real.sq_sqrt,mul_inv_cancel₀have hsymm : A.IsSymm := by rw [← Matrix.isHermitian_iff_isSymm] exact hpos.isHermitian let d : Fin 3 → ℝ := fun k => (Real.sqrt (A k k))⁻¹ have hd_pos : ∀ i, 0 < d i := fun i => inv_pos.mpr (Real.sqrt_pos.mpr hpos.diag_pos) have hd_ne : ∀ i, d i ≠ 0 := fun i => ne_of_gt (hd_pos i) let B := Matrix.diagonal d * A * Matrix.diagonal d have hBsymm : B.IsSymm := by ext i j dsimp [B] simp only [Matrix.mul_diagonal, Matrix.diagonal_mul] rw [hsymm.apply i j] ring have hBdiag : ∀ i, B i i = 1 := by intro i dsimp [B] simp only [Matrix.mul_diagonal, Matrix.diagonal_mul] have hposi : 0 < A i i := hpos.diag_pos have hsq : (Real.sqrt (A i i)) ^ 2 = A i i := Real.sq_sqrt (le_of_lt hposi) have : (Real.sqrt (A i i))⁻¹ * A i i * (Real.sqrt (A i i))⁻¹ = A i i * ((Real.sqrt (A i i))⁻¹ * (Real.sqrt (A i i))⁻¹) := by ring rw [this, ← mul_inv, ← sq, hsq, mul_inv_cancel₀ (ne_of_gt hposi)]
The off-diagonal entries x = B 0 1, y = B 1 2, and z = B 0 2 satisfy x^2 < 1 and y^2 < 1.
For any distinct indices i ≠ j, the 2 × 2 principal submatrix on {i, j} is positive definite, so its determinant A i i * A j j - (A i j)^2 is strictly positive. This implies (A i j)^2 < A i i * A j j, so (B i j)^2 = (A i j)^2 / (A i i * A j j) < 1. In particular, x^2 < 1 and y^2 < 1.
uses
Matrix.PosDef.det_pos,Matrix.det_fin_two,div_lt_onelet x := B 0 1 let y := B 1 2 let z := B 0 2 have hminor : ∀ i j : Fin 3, i ≠ j → (B i j) ^ 2 < 1 := by intro i j hij have hposi : 0 < A i i := hpos.diag_pos have hposj : 0 < A j j := hpos.diag_pos have hij_sq : A i j ^ 2 < A i i * A j j := by let e : Fin 2 → Fin 3 := ![i, j] have he : Function.Injective e := by intro a b hab fin_cases a <;> fin_cases b · rfl · exfalso; apply hij; exact hab · exfalso; apply hij; exact hab.symm · rfl have hsub := hpos.submatrix he have hdet := Matrix.PosDef.det_pos hsub rw [Matrix.det_fin_two] at hdet have h10 : A j i = A i j := hsymm.apply i j change (A i i) * (A j j) - (A i j) * (A j i) > 0 at hdet rw [h10] at hdet linarith dsimp [B] simp only [Matrix.mul_diagonal, Matrix.diagonal_mul] have hsqi : (Real.sqrt (A i i)) ^ 2 = A i i := Real.sq_sqrt (le_of_lt hposi) have hsqj : (Real.sqrt (A j j)) ^ 2 = A j j := Real.sq_sqrt (le_of_lt hposj) have : ((Real.sqrt (A i i))⁻¹ * A i j * (Real.sqrt (A j j))⁻¹) ^ 2 = A i j ^ 2 / (A i i * A j j) := by rw [mul_pow, mul_pow, inv_pow, inv_pow, hsqi, hsqj] ring rw [this] have hprod_pos : 0 < A i i * A j j := mul_pos hposi hposj exact (div_lt_one hprod_pos).mpr hij_sq have hx : x ^ 2 < 1 := hminor 0 1 (by decide) have hy : y ^ 2 < 1 := hminor 1 2 (by decide)
The vector of off-diagonal entries (x, y, z) is non-zero.
If (x, y, z) were (0, 0, 0), then because the scaling factors d k are non-zero, each off-diagonal entry A 0 1, A 1 2, and A 0 2 would vanish. By symmetry, every off-diagonal entry of A would be zero, making A diagonal and contradicting ¬ A.IsDiag.
uses
Prod.mk.inj,mul_eq_zero.mp,False.elimhave hne : (x, y, z) ≠ (0, 0, 0) := by intro hzero obtain ⟨hx0, hyz⟩ := Prod.mk.inj hzero obtain ⟨hy0, hz0⟩ := Prod.mk.inj hyz dsimp only [x, y, z, B] at hx0 hy0 hz0 rw [Matrix.mul_diagonal, Matrix.diagonal_mul] at hx0 hy0 hz0 have mul_cancel_both {u v : ℝ} (hu : u ≠ 0) (hv : v ≠ 0) {w : ℝ} (h : u * w * v = 0) : w = 0 := by have : u * (w * v) = 0 := by linarith cases mul_eq_zero.mp this with | inl h1 => exact False.elim (hu h1) | inr h1 => cases mul_eq_zero.mp h1 with | inl h2 => exact h2 | inr h2 => exact False.elim (hv h2) have ha01 : A 0 1 = 0 := mul_cancel_both (hd_ne 0) (hd_ne 1) hx0 have ha12 : A 1 2 = 0 := mul_cancel_both (hd_ne 1) (hd_ne 2) hy0 have ha02 : A 0 2 = 0 := mul_cancel_both (hd_ne 0) (hd_ne 2) hz0 apply hnondiag intro i j hij fin_cases i <;> fin_cases j · contradiction · exact ha01 · exact ha02 · rw [← hsymm.apply]; exact ha01 · contradiction · exact ha12 · rw [← hsymm.apply]; exact ha02 · rw [← hsymm.apply]; exact ha12 · contradiction
The q-permanent of B is a cubic polynomial in q, and q ↦ matrix_q_permanent q B is strictly increasing on ℝ.
Because B is symmetric with 1s on the diagonal, matrix_q_permanent_three_of_diag_one gives its q-permanent as 1 + (x^2 + y^2) q + 2 x y z q^2 + z^2 q^3. Since x^2 < 1, y^2 < 1, and (x, y, z) ≠ (0, 0, 0), cubic_poly_strict_mono_of_minors_lt_one implies this cubic polynomial is strictly increasing.
uses
cubic_poly_strict_mono_of_minors_lt_one,matrix_q_permanent_three_of_diag_onehave hmonoB := cubic_poly_strict_mono_of_minors_lt_one x y z hx hy hne have hBpoly : ∀ q, matrix_q_permanent q B = 1 + (x ^ 2 + y ^ 2) * q + 2 * x * y * z * q ^ 2 + z ^ 2 * q ^ 3 := fun q => matrix_q_permanent_three_of_diag_one B hBsymm hBdiag q have hB_strict : StrictMono (fun q => matrix_q_permanent q B) := by intro u v huv have := hmonoB huv dsimp only at this rw [← hBpoly u, ← hBpoly v] at this exact this
The function q ↦ matrix_q_permanent q A is strictly increasing on ℝ.
Conjugation by D satisfies matrix_q_permanent q B = (det D)^2 * matrix_q_permanent q A. Because each d k > 0, the determinant det D = ∏ d k is positive, so (det D)^2 > 0. Dividing the strict inequality matrix_q_permanent u B < matrix_q_permanent v B by (det D)^2 preserves the strict inequality for A.
uses
matrix_q_permanent_diagonal_conj,Matrix.det_diagonal,Finset.prod_pos,sq_pos_of_pos,mul_lt_mul_iff_of_pos_lefthave hconj : ∀ q, matrix_q_permanent q B = (Matrix.diagonal d).det ^ 2 * matrix_q_permanent q A := fun q => matrix_q_permanent_diagonal_conj q A d have hdet_pos : 0 < (Matrix.diagonal d).det := by rw [Matrix.det_diagonal] exact Finset.prod_pos fun i _ => hd_pos i have hc_pos : 0 < (Matrix.diagonal d).det ^ 2 := sq_pos_of_pos hdet_pos intro u v huv have hlt := hB_strict huv dsimp only at hlt rw [hconj u, hconj v] at hlt exact (mul_lt_mul_iff_of_pos_left hc_pos).mp hlt
The dimension n cannot be 0 or 1.
Case analysis on n ≤ 3 leaves n ∈ {0, 1, 2, 3}. If n = 0, there are no matrix entries, so A is vacuously diagonal. If n = 1, there are no distinct indices i ≠ j, so A is also diagonal. In both cases, this contradicts ¬ A.IsDiag.
interval_cases n · exfalso apply hnondiag intro i exact i.elim0 · exfalso apply hnondiag intro i j hij exfalso apply hij ext omega
In the remaining cases n = 2 and n = 3, q ↦ matrix_q_permanent q A is strictly increasing.
These cases are dispatched directly by the helper lemmas for 2 × 2 and 3 × 3 matrices.
uses
matrix_q_permanent_strict_mono_of_order_le_three_case_two,matrix_q_permanent_strict_mono_of_order_le_three_case_three· exact matrix_q_permanent_strict_mono_of_order_le_three_case_two A hpos hnondiag · exact matrix_q_permanent_strict_mono_of_order_le_three_case_three A hpos hnondiag
Matrix.isHermitian_iff_isSymm- A matrix with real entries is Hermitian if and only if it is symmetric.
matrix_q_permanent_two_of_is_symm- For a symmetric 2 × 2 matrix A, its q-permanent equals A 0 0 * A 1 1 + q * (A 0 1)^2.
sq_pos_of_ne_zero- The square of a non-zero real number is strictly positive.
Matrix.mul_diagonal- Multiplying a matrix on the right by a diagonal matrix scales each column by the corresponding diagonal entry.
Matrix.diagonal_mul- Multiplying a matrix on the left by a diagonal matrix scales each row by the corresponding diagonal entry.
Real.sq_sqrt- The square of the square root of a non-negative real number is the number itself.
mul_inv_cancel₀- Multiplying a non-zero element by its inverse yields 1.
Matrix.PosDef.det_pos- A positive definite matrix has strictly positive determinant.
Matrix.det_fin_two- The determinant of a 2 × 2 matrix is the product of its diagonal entries minus the product of its off-diagonal entries.
div_lt_one- A fraction with positive denominator is strictly less than 1 if and only if its numerator is strictly less than its denominator.
Prod.mk.inj- Equality of pairs implies equality of their respective components.
mul_eq_zero.mp- If the product of two elements is zero, at least one of the factors is zero.
False.elim- From a contradiction (False), any proposition follows.
cubic_poly_strict_mono_of_minors_lt_one- The cubic polynomial 1 + (x^2 + y^2) q + 2 x y z q^2 + z^2 q^3 is strictly increasing on ℝ if x^2 < 1, y^2 < 1, and (x, y, z) ≠ (0, 0, 0).
matrix_q_permanent_three_of_diag_one- The q-permanent of a 3 × 3 symmetric matrix with 1s on the diagonal is 1 + (x^2 + y^2) q + 2 x y z q^2 + z^2 q^3, where x, y, z are the off-diagonal entries.
matrix_q_permanent_diagonal_conj- Conjugating a matrix symmetrically by a diagonal matrix D multiplies its q-permanent by (det D)^2.
Matrix.det_diagonal- The determinant of a diagonal matrix is the product of its diagonal entries.
Finset.prod_pos- The product of strictly positive elements over a finite set is strictly positive.
sq_pos_of_pos- The square of a strictly positive element is strictly positive.
mul_lt_mul_iff_of_pos_left- Multiplying an inequality by a strictly positive number preserves the strict inequality.
matrix_q_permanent_strict_mono_of_order_le_three_case_two- The q-permanent of a non-diagonal real positive definite 2 × 2 matrix is strictly increasing on ℝ.
matrix_q_permanent_strict_mono_of_order_le_three_case_three- The q-permanent of a non-diagonal real positive definite 3 × 3 matrix is strictly increasing on ℝ.
- Finite index exhaustion (`fin_cases`), linear integer arithmetic (`omega`), and nonlinear real inequalities (`nlinarith`) are handled by automation.
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
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
lemma matrix_q_permanent_strict_mono_of_order_le_three_case_two (A : Matrix (Fin 2) (Fin 2) ℝ) (hpos : A.PosDef) (hnondiag : ¬ A.IsDiag) : StrictMono (fun q : ℝ => matrix_q_permanent q A) := by have hsymm : A.IsSymm := by rw [← Matrix.isHermitian_iff_isSymm] exact hpos.isHermitian have hperm : ∀ q, matrix_q_permanent q A = A 0 0 * A 1 1 + q * (A 0 1) ^ 2 := fun q => matrix_q_permanent_two_of_is_symm q A hsymm have hne : A 0 1 ≠ 0 := by intro hzero apply hnondiag intro i j hij fin_cases i <;> fin_cases j · contradiction · exact hzero · rw [← hsymm.apply]; exact hzero · contradiction have hsq_pos : 0 < (A 0 1) ^ 2 := sq_pos_of_ne_zero hne intro x y hxy dsimp only rw [hperm x, hperm y] nlinarithlemma matrix_q_permanent_strict_mono_of_order_le_three_case_three (A : Matrix (Fin 3) (Fin 3) ℝ) (hpos : A.PosDef) (hnondiag : ¬ A.IsDiag) : StrictMono (fun q : ℝ => matrix_q_permanent q A) := by have hsymm : A.IsSymm := by rw [← Matrix.isHermitian_iff_isSymm] exact hpos.isHermitian let d : Fin 3 → ℝ := fun k => (Real.sqrt (A k k))⁻¹ have hd_pos : ∀ i, 0 < d i := fun i => inv_pos.mpr (Real.sqrt_pos.mpr hpos.diag_pos) have hd_ne : ∀ i, d i ≠ 0 := fun i => ne_of_gt (hd_pos i) let B := Matrix.diagonal d * A * Matrix.diagonal d have hBsymm : B.IsSymm := by ext i j dsimp [B] simp only [Matrix.mul_diagonal, Matrix.diagonal_mul] rw [hsymm.apply i j] ring have hBdiag : ∀ i, B i i = 1 := by intro i dsimp [B] simp only [Matrix.mul_diagonal, Matrix.diagonal_mul] have hposi : 0 < A i i := hpos.diag_pos have hsq : (Real.sqrt (A i i)) ^ 2 = A i i := Real.sq_sqrt (le_of_lt hposi) have : (Real.sqrt (A i i))⁻¹ * A i i * (Real.sqrt (A i i))⁻¹ = A i i * ((Real.sqrt (A i i))⁻¹ * (Real.sqrt (A i i))⁻¹) := by ring rw [this, ← mul_inv, ← sq, hsq, mul_inv_cancel₀ (ne_of_gt hposi)] let x := B 0 1 let y := B 1 2 let z := B 0 2 have hminor : ∀ i j : Fin 3, i ≠ j → (B i j) ^ 2 < 1 := by intro i j hij have hposi : 0 < A i i := hpos.diag_pos have hposj : 0 < A j j := hpos.diag_pos have hij_sq : A i j ^ 2 < A i i * A j j := by let e : Fin 2 → Fin 3 := ![i, j] have he : Function.Injective e := by intro a b hab fin_cases a <;> fin_cases b · rfl · exfalso; apply hij; exact hab · exfalso; apply hij; exact hab.symm · rfl have hsub := hpos.submatrix he have hdet := Matrix.PosDef.det_pos hsub rw [Matrix.det_fin_two] at hdet have h10 : A j i = A i j := hsymm.apply i j change (A i i) * (A j j) - (A i j) * (A j i) > 0 at hdet rw [h10] at hdet linarith dsimp [B] simp only [Matrix.mul_diagonal, Matrix.diagonal_mul] have hsqi : (Real.sqrt (A i i)) ^ 2 = A i i := Real.sq_sqrt (le_of_lt hposi) have hsqj : (Real.sqrt (A j j)) ^ 2 = A j j := Real.sq_sqrt (le_of_lt hposj) have : ((Real.sqrt (A i i))⁻¹ * A i j * (Real.sqrt (A j j))⁻¹) ^ 2 = A i j ^ 2 / (A i i * A j j) := by rw [mul_pow, mul_pow, inv_pow, inv_pow, hsqi, hsqj] ring rw [this] have hprod_pos : 0 < A i i * A j j := mul_pos hposi hposj exact (div_lt_one hprod_pos).mpr hij_sq have hx : x ^ 2 < 1 := hminor 0 1 (by decide) have hy : y ^ 2 < 1 := hminor 1 2 (by decide) have hne : (x, y, z) ≠ (0, 0, 0) := by intro hzero obtain ⟨hx0, hyz⟩ := Prod.mk.inj hzero obtain ⟨hy0, hz0⟩ := Prod.mk.inj hyz dsimp only [x, y, z, B] at hx0 hy0 hz0 rw [Matrix.mul_diagonal, Matrix.diagonal_mul] at hx0 hy0 hz0 have mul_cancel_both {u v : ℝ} (hu : u ≠ 0) (hv : v ≠ 0) {w : ℝ} (h : u * w * v = 0) : w = 0 := by have : u * (w * v) = 0 := by linarith cases mul_eq_zero.mp this with | inl h1 => exact False.elim (hu h1) | inr h1 => cases mul_eq_zero.mp h1 with | inl h2 => exact h2 | inr h2 => exact False.elim (hv h2) have ha01 : A 0 1 = 0 := mul_cancel_both (hd_ne 0) (hd_ne 1) hx0 have ha12 : A 1 2 = 0 := mul_cancel_both (hd_ne 1) (hd_ne 2) hy0 have ha02 : A 0 2 = 0 := mul_cancel_both (hd_ne 0) (hd_ne 2) hz0 apply hnondiag intro i j hij fin_cases i <;> fin_cases j · contradiction · exact ha01 · exact ha02 · rw [← hsymm.apply]; exact ha01 · contradiction · exact ha12 · rw [← hsymm.apply]; exact ha02 · rw [← hsymm.apply]; exact ha12 · contradiction have hmonoB := cubic_poly_strict_mono_of_minors_lt_one x y z hx hy hne have hBpoly : ∀ q, matrix_q_permanent q B = 1 + (x ^ 2 + y ^ 2) * q + 2 * x * y * z * q ^ 2 + z ^ 2 * q ^ 3 := fun q => matrix_q_permanent_three_of_diag_one B hBsymm hBdiag q have hB_strict : StrictMono (fun q => matrix_q_permanent q B) := by intro u v huv have := hmonoB huv dsimp only at this rw [← hBpoly u, ← hBpoly v] at this exact this have hconj : ∀ q, matrix_q_permanent q B = (Matrix.diagonal d).det ^ 2 * matrix_q_permanent q A := fun q => matrix_q_permanent_diagonal_conj q A d have hdet_pos : 0 < (Matrix.diagonal d).det := by rw [Matrix.det_diagonal] exact Finset.prod_pos fun i _ => hd_pos i have hc_pos : 0 < (Matrix.diagonal d).det ^ 2 := sq_pos_of_pos hdet_pos intro u v huv have hlt := hB_strict huv dsimp only at hlt rw [hconj u, hconj v] at hlt exact (mul_lt_mul_iff_of_pos_left hc_pos).mp hlt/-- The q-permanent of a non-diagonal real positive definite matrix of order at most 3 is strictly increasing on ℝ. -/theorem matrix_q_permanent_strict_mono_of_order_le_three (n : ℕ) (hn : n ≤ 3) (A : Matrix (Fin n) (Fin n) ℝ) (hpos : A.PosDef) (hnondiag : ¬ A.IsDiag) : StrictMono (fun q : ℝ => matrix_q_permanent q A) := by interval_cases n · exfalso apply hnondiag intro i exact i.elim0 · exfalso apply hnondiag intro i j hij exfalso apply hij ext omega · exact matrix_q_permanent_strict_mono_of_order_le_three_case_two A hpos hnondiag · exact matrix_q_permanent_strict_mono_of_order_le_three_case_three A hpos hnondiag