Hadamard products and Schur–Szegő composition

Coefficientwise products of polynomials preserve Pólya frequency polynomials and interlacing. The same holds for the Schur–Szegő composition, which also controls root locations, and the finite Pólya–Schur theorem characterizes the diagonal operators that preserve real-rootedness.

Results

A PF polynomial is a real polynomial with nonnegative coefficients that is zero or real-rooted. For $f=\sum_kf_kx^k$ and $p=\sum_kp_kx^k$, the Hadamard product is $f\ast p=\sum_kf_kp_kx^k$ and the factorial Hadamard product is $f\circledast p=\sum_kk!\,f_kp_kx^k$. For $f$ and $g$ of degree at most $n$, the Schur–Szegő composition is $$ f*_ng=\sum_{k=0}^n\binom nk^{-1}f_kg_kx^k. $$

Theorem (Garloff–Wagner). [GW96] For PF polynomials $p,q$ the Hadamard product $p\ast q$ is PF. If $f \ll g$ and $p \ll q$ have nonnegative coefficients, then $f\ast p \ll g\ast q$ or one of $f\ast p$ and $g\ast q$ is zero. If $f,g,p$ are PF and $f\ll g$ (or one of $f,g$ is zero), then $f\circledast p\ll g\circledast p$ or one of them is zero.

Theorem (Schur; Maló). [Sch14] If $f$ and $g$ are real-rooted and the roots of $g$ are all nonpositive or all nonnegative, then $f\circledast g$ is zero or real-rooted. The termwise product of two Pólya frequency coefficient sequences of polynomials is a Pólya frequency sequence; equivalently, the entrywise product of their totally nonnegative Toeplitz matrices is totally nonnegative.

Theorem (Schur–Szegő; Kostov–Shapiro). [KS06]; [Kos10] If $f$ is a PF polynomial of degree at most $n$ and $g$ is real-rooted of degree at most $n$, then $f*_ng$ is zero or real-rooted. Now let $f$ and $g$ have degree $n$, let $a \neq 0$ be a root of $f$ of multiplicity $m$ and $b \neq 0$ a root of $g$ of multiplicity $l$, with $m + l \geq n$. Then $-ab$ is a root of $f *_n g$ of multiplicity exactly $m + l - n$. If moreover $f$ is real-rooted and $g$ has only negative roots, then $f *_n g$ is real-rooted with as many positive, zero and negative roots as $f$, every nonzero root of multiplicity at least two is such a product $-ab$ with $m+l>n$, and all other nonzero roots are simple. Real-rootedness still holds when $f$ and $g$ only have degree at most $n$.

Root intervals are mapped to root intervals [KS06]. If $g$ has degree $n$ with all roots in $[-b,-a]$, where $0<a\le b$, and $f$ has degree at most $n$ with all roots in $[-v,-u]$, where $0\le u\le v$, then $f*_ng$ has all roots in $[-bv,-au]$. If instead the roots of $f$ lie in $[-v,0]$ and those of $g$ in $[-b,0]$ with $b,v\ge0$, then $f*_ng$ is zero or has all roots in $[-bv,0]$, and it is nonzero when $\deg f=n$. Under the first hypotheses, with $\deg f=m$, let $\beta_1\le\dots\le\beta_m$ be the roots of the degree-$m$ reduction of $g$ and $\rho_1\le\dots\le\rho_m$ those of $f*_ng$; then $v\beta_i\le\rho_i\le u\beta_i$ for every $i$.

Theorem (finite Pólya–Schur). Let $\gamma_0,\dots,\gamma_n$ be real, let $T_\gamma(x^k) = \gamma_k x^k$ and $J(x)=\sum_{k=0}^n \binom{n}{k} \gamma_k x^k$. If every $\gamma_k \geq 0$, then $T_\gamma$ maps each real-rooted polynomial of degree at most $n$ to zero or a real-rooted polynomial if and only if $J$ is a PF polynomial. If every $(-1)^k\gamma_k \geq 0$, it does so if and only if $J$ is zero or real-rooted with all roots in $[0,\infty)$.

For value sequences, Wagner's theorem says: if $f(0),f(1),f(2),\dotsc$ and $g(0),g(1),g(2),\dotsc$ are Pólya frequency sequences for real polynomials $f,g$, then so is the value sequence of $fg$ [Wag92]; [BFJ24]. Write $\sum_{n \ge 0} p(n)\, x^n = W(p)(x)/(1-x)^{\deg p + 1}$. By Brenti, if $p$ is real-rooted with all roots in $[-1, 0]$ and nonnegative leading coefficient, then $W(p)$ is PF; for $p = x + a$ with $0\le a\le1$ it is $a + (1-a)x$. By Liu and Mao [LM26], if the coefficients of $W(p)$ and $W(q)$ are nonnegative, log-concave and without internal zeros (up to $\deg p$ and $\deg q$), then so are those of $W(pq)$ (up to $\deg pq$); the example $W(p) = 1 + x^3$, $W(q) = x$, $W(pq) = 4x + 3x^3 + x^4$ shows that internal zeros must be excluded.

Further results: Kostov's coefficient map, which shows that a monic $P$ with $P(-1)=0$ written as $K_{a_1}*_n\dots*_nK_{a_{n-1}}$, $K_a=(x+1)^{n-1}(x+a)$, has interior coefficients affine in $\sigma_j(a_1,\dots,a_{n-1})$ with an invertible scaled Vandermonde linear part, so its coefficients determine $\sigma(a)$; and the fact that root-location preservation passes to rowwise Schur–Szegő and diagonal scaling of a polynomial sequence, but not to rowwise Hadamard products of two preserving sequence maps (one sends $(1+2x)^2$ to $1+3x+4x^2$) or to rowwise normalized Schur squares of one invertible preserving sequence (one sends $1+2x$ to $(9x^2+24x+17)/8$).

References

See also Hadamard products, Schur–Szegő composition and the finite multiplier criterion on symmetricfunctions.com.

  1. [Sch14] Issai Schur. Zwei Sätze über algebraische Gleichungen mit lauter reellen Wurzeln. Journal für die reine und angewandte Mathematik, 144(2), 75-88, 1914.
    BibTeX
    @article{Schur1914,
      title = {Zwei S{\"a}tze {\"u}ber algebraische {G}leichungen mit
               lauter reellen {W}urzeln},
      author = {Schur, Issai},
      year = {1914},
      journal = {Journal f{\"u}r die reine und angewandte Mathematik},
      volume = {144},
      number = {2},
      pages = {75--88},
      doi = {10.1515/crll.1914.144.75},
      url2 = {https://doi.org/10.1515/crll.1914.144.75},
      publisher = {Walter de Gruyter}
    }
  2. [SP14] J Schur & G Polya. Über zwei Arten von Faktorenfolgen in der Theorie der algebraischen Gleichungen. J. Reine Angew. Math., 1914.
    BibTeX
    @article{SchurPolya1914,
      title={{\"U}ber zwei {A}rten von {F}aktorenfolgen in der {T}heorie der algebraischen {G}leichungen},
      author={Schur, J and Polya, G},
      year={1914},
      journal={J. Reine Angew. Math.},
      publisher={Walter de Gruyter, Berlin/New York Berlin, New York}
    }
  3. [Wag92] David G Wagner. Total positivity of Hadamard products. Journal of Mathematical Analysis and Applications, 163(2), 459-483, 1992.
    BibTeX
    @article{Wagner1992,
      doi = {10.1016/0022-247x(92)90261-b},
      url2 = {https://doi.org/10.1016/0022-247x(92)90261-b},
      year = {1992},
      month = jan,
      publisher = {Elsevier {BV}},
      volume = {163},
      number = {2},
      pages = {459--483},
      author = {David G Wagner},
      title = {Total positivity of {H}adamard products},
      journal = {Journal of Mathematical Analysis and Applications}
    }
  4. [GW96] Jürgen Garloff & David G. Wagner. Hadamard products of stable polynomials are stable. Journal of Mathematical Analysis and Applications, 202(3), 797-809, 1996.
    BibTeX
    @article{GarloffWagner1996,
      doi = {10.1006/jmaa.1996.0348},
      url2 = {https://doi.org/10.1006/jmaa.1996.0348},
      year = {1996},
      publisher = {Elsevier {BV}},
      volume = {202},
      number = {3},
      pages = {797--809},
      author = {J{\"u}rgen Garloff and David G. Wagner},
      title = {Hadamard Products of Stable Polynomials Are Stable},
      journal = {Journal of Mathematical Analysis and Applications}
    }
  5. [BFJ24] Petter Brändén, Luis Ferroni, & Katharina Jochemko. Preservation of inequalities under Hadamard products. Transactions of the American Mathematical Society, 2024.
    BibTeX
    @article{BrandenFerroniJochemko2024x,
      author = {Petter Br{\"{a}}nd{\'{e}}n and Luis Ferroni and Katharina Jochemko},
      title = {Preservation of inequalities under {H}adamard products},
      year = {2024},
      eprint = {2408.12386v2},
      url = {https://arxiv.org/abs/2408.12386v2},
      journal = {Transactions of the American Mathematical Society},
      note = {To appear}
    }
  6. [LM26] Yanxin Liu & Jianxi Mao. Preservation of log-concavity under Hadamard products. arXiv e-prints, 2026.
    BibTeX
    @article{LiuMao2026Hadamard,
      author = {Yanxin Liu and Jianxi Mao},
      title = {Preservation of log-concavity under {H}adamard products},
      year = {2026},
      eprint = {2609.11589v1},
      url = {https://arxiv.org/abs/2609.11589v1},
      journal = {arXiv e-prints}
    }
  7. [KS06] Vladimir Kostov & Boris Shapiro. On the Schur--Szegö composition of polynomials. Comptes Rendus. Mathématique, 343(2), 81-86, 2006.
    BibTeX
    @article{KostovShapiro2006,
      author = {Kostov, Vladimir and Shapiro, Boris},
      title = {On the {S}chur--{S}zeg{\"o} composition of polynomials},
      year = {2006},
      journal = {Comptes Rendus. Mathématique},
      volume = {343},
      number = {2},
      pages = {81--86},
      doi = {10.1016/j.crma.2006.06.007},
      url = {https://doi.org/10.1016/j.crma.2006.06.007}
    }
  8. [Kos10] Vladimir P. Kostov. Interlacing properties and the Schur--Szegő composition. Functional Analysis and Other Mathematics, 3(1), 65-74, 2010.
    BibTeX
    @article{Kostov2010,
      author = {Kostov, Vladimir P.},
      title = {Interlacing properties and the {S}chur--{S}zeg{\H{o}} composition},
      year = {2010},
      journal = {Functional Analysis and Other Mathematics},
      volume = {3},
      number = {1},
      pages = {65--74},
      doi = {10.1007/s11853-010-0039-2},
      url = {https://doi.org/10.1007/s11853-010-0039-2}
    }

Definitions

  • Schur–Szegő composition

    RealRooted.schurSzegoCompsource
    def schurSzegoComp (n : Nat) (f g : ℝ[X]) : ℝ[X] :=
      Finset.sum (Finset.range (n + 1))
        (fun k => monomial k (f.coeff k * g.coeff k / (Nat.choose n k : ℝ)))
  • Hadamard product

    RealRooted.hadamardProductsource
    def hadamardProduct (p q : ℝ[X]) : ℝ[X] :=
      p.sum fun n a => monomial n (a * q.coeff n)
  • PF polynomial

    RealRooted.IsPFPolynomialsource
    def IsPFPolynomial (p : ℝ[X]) : Prop :=
      HasNonnegCoeffs p ∧ (p = 0 ∨ p.Splits) ∧ ∀ r ∈ p.roots, r ≤ 0
  • Values of a polynomial at 0, 1, 2, …

    RealRooted.polynomialValueSeqsource
    def polynomialValueSeq (p : ℝ[X]) : ℕ → ℝ :=
      fun n => p.eval (n : ℝ)
  • Factorial Hadamard product

    RealRooted.factorialHadamardProductsource
    def factorialHadamardProduct (p q : ℝ[X]) : ℝ[X] :=
      diagonalOperator (fun k => (Nat.factorial k : ℝ) * q.coeff k) p

Theorems

  • Schur–Szegő composition preserves real-rootedness

    RealRooted.finiteSchurSzegoCompositionsource
    theorem finiteSchurSzegoComposition {n : ℕ} {f p : ℝ[X]}
        (hf : IsPFPolynomial f) (hfdeg : f.natDegree ≤ n)
        (hpdeg : p.natDegree ≤ n) (hsplit : p.Splits) :
        schurSzegoComp n f p = 0 ∨ (schurSzegoComp n f p).Splits
  • Kostov–Shapiro: exact multiplicity of the root −ab of a Schur–Szegő composition

    RealRooted.rootMultiplicity_schurSzegoCompsource
    theorem rootMultiplicity_schurSzegoComp {n m l : ℕ} {f g : ℝ[X]} {a b : ℝ}
        (hf : f.natDegree = n) (hg : g.natDegree = n) (ha : a ≠ 0) (hb : b ≠ 0)
        (hm : f.rootMultiplicity a = m) (hl : g.rootMultiplicity b = l) (hn : n ≤ m + l) :
        (schurSzegoComp n f g).rootMultiplicity (-(a * b)) = m + l - n
  • Kostov–Shapiro: every nonzero multiple root of P *_n Q is a forced product −ab

    RealRooted.exists_forced_factors_of_rootMultiplicity_schurSzegoCompsource
    theorem exists_forced_factors_of_rootMultiplicity_schurSzegoComp
        {n : ℕ} {P Q : ℝ[X]} (hP : P.natDegree = n) (hPs : P.Splits)
        (hQ : Q.natDegree = n) (hQs : Q.Splits) (hneg : ∀ b ∈ Q.roots, b < 0) :
        ∀ r, r ≠ 0 → 1 < (schurSzegoComp n P Q).rootMultiplicity r →
          ∃ a b : ℝ, a ≠ 0 ∧ b ≠ 0 ∧ P.IsRoot a ∧ Q.IsRoot b ∧
            n < P.rootMultiplicity a + Q.rootMultiplicity b ∧ r = -(a * b) ∧
            (schurSzegoComp n P Q).rootMultiplicity r =
              P.rootMultiplicity a + Q.rootMultiplicity b - n
  • Kostov–Shapiro: composition with a negative-rooted polynomial preserves root signs

    RealRooted.schurSzegoComp_roots_sign_counts_of_roots_negsource
    theorem schurSzegoComp_roots_sign_counts_of_roots_neg {n : ℕ} {P Q : ℝ[X]}
        (hP : P.natDegree = n) (hPs : P.Splits) (hQ : Q.natDegree = n) (hQs : Q.Splits)
        (hneg : ∀ b ∈ Q.roots, b < 0) :
        (schurSzegoComp n P Q).roots.countP (0 < ·) = P.roots.countP (0 < ·) ∧
          (schurSzegoComp n P Q).roots.count 0 = P.roots.count 0 ∧
          (schurSzegoComp n P Q).roots.countP (· < 0) = P.roots.countP (· < 0)
  • Composition with a negative-rooted polynomial preserves real-rootedness

    RealRooted.splits_schurSzegoComp_of_roots_negsource
    theorem splits_schurSzegoComp_of_roots_neg {n : ℕ} {P Q : ℝ[X]} (hP : P.natDegree ≤ n)
        (hPs : P.Splits) (hQ : Q.natDegree ≤ n) (hQs : Q.Splits) (hneg : ∀ b ∈ Q.roots, b < 0) :
        (schurSzegoComp n P Q).Splits
  • Kostov–Shapiro: roots in [−b, −a] and [−v, −u] give roots in [−bv, −au]

    RealRooted.IsRealRootedIn.schurSzegoComp_Iccsource
    theorem IsRealRootedIn.schurSzegoComp_Icc {D : ℕ} {f G : ℝ[X]} {a b u v : ℝ}
        (ha : 0 < a) (hab : a ≤ b) (hu : 0 ≤ u) (huv : u ≤ v)
        (hf : IsRealRootedIn (Icc (-v) (-u)) f) (hG : IsRealRootedIn (Icc (-b) (-a)) G)
        (hfdeg : f.natDegree = D) (hGdeg : G.natDegree = D) :
        IsRealRootedIn (Icc (-(b * v)) (-(a * u))) (schurSzegoComp D f G)
  • Kostov–Shapiro: roots in [−b, 0] and [−v, 0] give roots in [−bv, 0]

    RealRooted.IsRealRootedIn.schurSzegoComp_Icc_zerosource
    theorem IsRealRootedIn.schurSzegoComp_Icc_zero {D : ℕ} {f G : ℝ[X]} {b v : ℝ} (hb : 0 ≤ b)
        (hv : 0 ≤ v) (hf : IsRealRootedIn (Icc (-v) 0) f) (hG : IsRealRootedIn (Icc (-b) 0) G)
        (hfdeg : f.natDegree = D) (hGdeg : G.natDegree = D) :
        IsRealRootedIn (Icc (-(b * v)) 0) (schurSzegoComp D f G)
  • Root intervals for an input of degree at most n

    RealRooted.IsRealRootedIn.schurSzegoComp_Icc_of_natDegree_lesource
    theorem IsRealRootedIn.schurSzegoComp_Icc_of_natDegree_le {D : ℕ} {f G : ℝ[X]}
        {a b u v : ℝ} (ha : 0 < a) (hab : a ≤ b) (hu : 0 ≤ u) (huv : u ≤ v)
        (hf : IsRealRootedIn (Icc (-v) (-u)) f) (hG : IsRealRootedIn (Icc (-b) (-a)) G)
        (hfdeg : f.natDegree ≤ D) (hGdeg : G.natDegree = D) :
        IsRealRootedIn (Icc (-(b * v)) (-(a * u))) (schurSzegoComp D f G)
  • Root rays [−b, 0] for an input of degree at most n; the output may be zero

    RealRooted.IsRealRootedIn.schurSzegoComp_Icc_zero_of_natDegree_lesource
    theorem IsRealRootedIn.schurSzegoComp_Icc_zero_of_natDegree_le {D : ℕ} {f G : ℝ[X]}
        {b v : ℝ} (hb : 0 ≤ b) (hv : 0 ≤ v) (hf : IsRealRootedIn (Icc (-v) 0) f)
        (hG : IsRealRootedIn (Icc (-b) 0) G) (hfdeg : f.natDegree ≤ D)
        (hGdeg : G.natDegree = D) :
        schurSzegoComp D f G = 0 ∨
          IsRealRootedIn (Icc (-(b * v)) 0) (schurSzegoComp D f G)
  • Fixed Schur–Szegő composition maps strict root intervals

    RealRooted.IsRealRootedIn.maps_schurSzegoComp_Iccsource
    theorem IsRealRootedIn.maps_schurSzegoComp_Icc {D : ℕ} {G : ℝ[X]} {a b u v : ℝ}
        (ha : 0 < a) (hab : a ≤ b) (hu : 0 ≤ u) (huv : u ≤ v)
        (hG : IsRealRootedIn (Icc (-b) (-a)) G) (hGdeg : G.natDegree = D) :
        MapsRealRootedIn (schurSzegoCompRightLinearMap D G) D
          (Icc (-v) (-u)) (Icc (-(b * v)) (-(a * u)))
  • Fixed Schur–Szegő composition maps root rays, allowing zero output

    RealRooted.IsRealRootedIn.maps_schurSzegoComp_Icc_zerosource
    theorem IsRealRootedIn.maps_schurSzegoComp_Icc_zero {D : ℕ} {G : ℝ[X]} {b v : ℝ}
        (hb : 0 ≤ b) (hv : 0 ≤ v) (hG : IsRealRootedIn (Icc (-b) 0) G)
        (hGdeg : G.natDegree = D) :
        MapsRealRootedIn (schurSzegoCompRightLinearMap D G) D
          (Icc (-v) 0) (Icc (-(b * v)) 0)
  • Coordinatewise bounds for the ordered roots of a Schur–Szegő composition

    RealRooted.IsRealRootedIn.orderedRoot_schurSzegoComp_boundssource
    theorem IsRealRootedIn.orderedRoot_schurSzegoComp_bounds
        {D : ℕ} {f G : ℝ[X]} {a b u v : ℝ}
        (ha : 0 < a) (hab : a ≤ b) (hu : 0 ≤ u) (huv : u ≤ v)
        (hf : IsRealRootedIn (Icc (-v) (-u)) f)
        (hG : IsRealRootedIn (Icc (-b) (-a)) G)
        (hfdeg : f.natDegree ≤ D) (hGdeg : G.natDegree = D) :
        ∀ i : Fin f.natDegree,
          v * orderedRoot (schurSzegoReduction D f.natDegree G) f.natDegree i ≤
            orderedRoot (schurSzegoComp D f G) f.natDegree i ∧
          orderedRoot (schurSzegoComp D f G) f.natDegree i ≤
            u * orderedRoot (schurSzegoReduction D f.natDegree G) f.natDegree i
  • Fixed Schur–Szegő composition postcomposes a polynomial-sequence operator

    RealRooted.basisTransformₗ_schurSzegoCompsource
    theorem basisTransformₗ_schurSzegoComp (D : ℕ) (G : ℝ[X]) (P : ℕ → ℝ[X]) :
        basisTransformₗ (fun n => schurSzegoComp D (P n) G) =
          schurSzegoCompRightLinearMap D G ∘ₗ basisTransformₗ P
  • Termwise sequence scaling precomposes by a diagonal operator

    RealRooted.basisTransformₗ_C_mulsource
    theorem basisTransformₗ_C_mul (γ : ℕ → ℝ) (P : ℕ → ℝ[X]) :
        basisTransformₗ (fun n => C (γ n) * P n) =
          basisTransformₗ P ∘ₗ diagonalLinearMap γ
  • Combined diagonal precomposition and Schur–Szegő postcomposition

    RealRooted.basisTransformₗ_schurSzegoComp_C_mulsource
    theorem basisTransformₗ_schurSzegoComp_C_mul (D : ℕ) (G : ℝ[X])
        (γ : ℕ → ℝ) (P : ℕ → ℝ[X]) :
        basisTransformₗ (fun n => schurSzegoComp D (P n) (C (γ n) * G)) =
          schurSzegoCompRightLinearMap D G ∘ₗ
            basisTransformₗ P ∘ₗ diagonalLinearMap γ
  • Root-location preservation for the combined sequence operation

    RealRooted.MapsRealRootedIn.basisTransform_schurSzegoComp_C_mulsource
    theorem MapsRealRootedIn.basisTransform_schurSzegoComp_C_mul
        {D n m : ℕ} {I J K L : Set ℝ} {G : ℝ[X]}
        {γ : ℕ → ℝ} {P : ℕ → ℝ[X]}
        (hG : MapsRealRootedIn (schurSzegoCompRightLinearMap D G) m K L)
        (hP : MapsRealRootedIn (basisTransformₗ P) n J K)
        (hγ : MapsRealRootedIn (diagonalLinearMap γ) n I J)
        (hdeg : ∀ p : ℝ[X], p.natDegree ≤ n → (basisTransform P p).natDegree ≤ m) :
        MapsRealRootedIn
          (basisTransformₗ (fun k => schurSzegoComp D (P k) (C (γ k) * G))) n I L
  • Two preserving sequence maps can have a non-real-rooted rowwise Hadamard product

    RealRooted.rowwiseHadamardProduct_counterexamplesource
    theorem rowwiseHadamardProduct_counterexample :
        basisTransform rowwiseHadamardCounterexampleR ((1 + C 2 * X) ^ 2) =
            C 4 * X ^ 2 + C 3 * X + 1 ∧
          ¬ (C 4 * X ^ 2 + C 3 * X + 1 : ℝ[X]).Splits
  • Identical invertible maps can have non-real-rooted rowwise Schur squares

    RealRooted.rowwiseNormalizedSchurProduct_counterexamplesource
    theorem rowwiseNormalizedSchurProduct_counterexample :
        basisTransform normalizedSchurCounterexampleR (1 + C 2 * X) =
            C (9 / 8 : ℝ) * X ^ 2 + C 3 * X + C (17 / 8 : ℝ) ∧
          ¬ (C (9 / 8 : ℝ) * X ^ 2 + C 3 * X + C (17 / 8 : ℝ)).Splits
  • Kostov: Schur–Szegő factorizations have coefficients affine in σ(a)

    RealRooted.Kostov.interiorCoeff_compositionsource
    theorem interiorCoeff_composition (n : ℕ) (a : List ℝ)
        (ha : a.length = n - 1) :
        interiorCoeff n (composition n a) =
          (coeffMatrix n).mulVec (sigma n a) + coeffConstant n
  • Kostov's coefficient map recovers the symmetric functions of a factorization

    RealRooted.Kostov.coeffMap_eq_sigmasource
    theorem coeffMap_eq_sigma (n : ℕ) (c : Fin (n - 1) → ℝ) (a : List ℝ)
        (ha : a.length = n - 1)
        (hfactor : composition n a = monicPolynomial n c) :
        coeffMap n c = sigma n a
  • Finite Pólya–Schur theorem

    RealRooted.Hadamard.finiteMultiplierSequence_iff_jensen_isPFPolynomialsource
    theorem Hadamard.finiteMultiplierSequence_iff_jensen_isPFPolynomial {n : ℕ}
        {gamma : ℕ → ℝ} (hgamma : ∀ k, 0 ≤ gamma k) :
        IsFiniteMultiplierSequence n gamma ↔ IsPFPolynomial (jensenPolynomial n gamma)
  • Finite Pólya–Schur theorem, alternating signs

    RealRooted.isFiniteMultiplierSequence_iff_jensenPolynomial_roots_nonnegsource
    theorem isFiniteMultiplierSequence_iff_jensenPolynomial_roots_nonneg
        {n : ℕ} {gamma : ℕ → ℝ} (hgamma : ∀ k, 0 ≤ (-1) ^ k * gamma k) :
        IsFiniteMultiplierSequence n gamma ↔
          (jensenPolynomial n gamma = 0 ∨ (jensenPolynomial n gamma).Splits) ∧
            ∀ r ∈ (jensenPolynomial n gamma).roots, 0 ≤ r
  • Garloff–Wagner: Hadamard products preserve interlacingMain result

    RealRooted.StrictInterl.interl_hadamardProductsource
    theorem StrictInterl.interl_hadamardProduct {f g p q : ℝ[X]}
        (hf : HasNonnegCoeffs f) (hg : HasNonnegCoeffs g)
        (hp : HasNonnegCoeffs p) (hq : HasNonnegCoeffs q)
        (hfg : StrictInterl f g) (hpq : StrictInterl p q) :
        Interl (hadamardProduct f p) (hadamardProduct g q)
  • Maló: termwise products of Pólya frequency coefficient sequences are PF

    RealRooted.IsPolyaFreqSeq.mul_coeffsource
    theorem IsPolyaFreqSeq.mul_coeff {p q : ℝ[X]}
        (hp : IsPolyaFreqSeq p.coeff) (hq : IsPolyaFreqSeq q.coeff) :
        IsPolyaFreqSeq (fun n => p.coeff n * q.coeff n)
  • Products of polynomial value sequences are PF

    RealRooted.isPolyaFreqSeq_polynomialValueSeq_mul_of_polyaFreqSeqsource
    theorem isPolyaFreqSeq_polynomialValueSeq_mul_of_polyaFreqSeq
        {f g : Polynomial ℝ} (hf : IsPolyaFreqSeq (polynomialValueSeq f))
        (hg : IsPolyaFreqSeq (polynomialValueSeq g)) :
        IsPolyaFreqSeq (polynomialValueSeq (f * g))
  • Brenti: zeros in [−1, 0] give a PF Euler numerator

    RealRooted.IsPFPolynomial.polynomialValueEulerNumerator_of_roots_mem_Iccsource
    theorem IsPFPolynomial.polynomialValueEulerNumerator_of_roots_mem_Icc {p : ℝ[X]}
        (hs : p.Splits) (hlc : 0 ≤ p.leadingCoeff) (hroots : ∀ r ∈ p.roots, -1 ≤ r ∧ r ≤ 0) :
        IsPFPolynomial (polynomialValueEulerNumerator p)
  • Garloff–Wagner: Hadamard products of PF polynomials are PF

    RealRooted.IsPFPolynomial.hadamardProductsource
    theorem IsPFPolynomial.hadamardProduct {p q : ℝ[X]}
        (hp : IsPFPolynomial p) (hq : IsPFPolynomial q) :
        IsPFPolynomial (hadamardProduct p q)
  • Garloff–Wagner: the factorial Hadamard product preserves interlacing

    RealRooted.Interl.factorialHadamardProduct_rightsource
    theorem Interl.factorialHadamardProduct_right {f g p : ℝ[X]}
        (hf : IsPFPolynomial f) (hg : IsPFPolynomial g) (hp : IsPFPolynomial p)
        (hfg : Interl f g) :
        Interl (factorialHadamardProduct f p) (factorialHadamardProduct g p)
  • Schur: the factorial product preserves real-rootedness

    RealRooted.factorialHadamardProduct_eq_zero_or_splitssource
    theorem factorialHadamardProduct_eq_zero_or_splits {f g : ℝ[X]} (hf : f.Splits)
        (hg : g.Splits) (hsign : (∀ r ∈ g.roots, r ≤ 0) ∨ ∀ r ∈ g.roots, 0 ≤ r) :
        factorialHadamardProduct f g = 0 ∨ (factorialHadamardProduct f g).Splits
  • Liu–Mao: log-concave Euler numerators without internal zeros are closed under products

    RealRooted.LiuMao.coeffLCNIZ_polynomialValueEulerNumerator_mulsource
    theorem coeffLCNIZ_polynomialValueEulerNumerator_mul (p q : ℝ[X])
        (hp : CoeffNonnegUpTo p.natDegree (polynomialValueEulerNumerator p).coeff ∧
          CoeffLogConcaveUpTo p.natDegree (polynomialValueEulerNumerator p).coeff ∧
          CoeffNoInternalZerosUpTo p.natDegree (polynomialValueEulerNumerator p).coeff)
        (hq : CoeffNonnegUpTo q.natDegree (polynomialValueEulerNumerator q).coeff ∧
          CoeffLogConcaveUpTo q.natDegree (polynomialValueEulerNumerator q).coeff ∧
          CoeffNoInternalZerosUpTo q.natDegree (polynomialValueEulerNumerator q).coeff) :
        CoeffNonnegUpTo (p * q).natDegree (polynomialValueEulerNumerator (p * q)).coeff ∧
          CoeffLogConcaveUpTo (p * q).natDegree (polynomialValueEulerNumerator (p * q)).coeff ∧
          CoeffNoInternalZerosUpTo (p * q).natDegree
            (polynomialValueEulerNumerator (p * q)).coeff

Lean source RealRooted/Challenges/PolyaFrequency/Hadamard.lean at revision d6670742. Formalized 2026-07-03.