The univariate Poisson product is symmetric
theorem starC_comm (c : ℝ) (f g : ℝ[X]) :
starC c f g = starC c g f
Formula for affine right inputs
theorem starC_affine_right (c a b : ℝ) (f : ℝ[X]) :
starC c f (C a + C b * X) =
C a * f + C b * X * (f + C c * f.derivative)
Nonnegative inputs give nonnegative coefficients
theorem starC_nonneg {c : ℝ} (hc : 0 ≤ c) {f g : ℝ[X]}
(hf : HasNonnegCoeffs f) (hg : HasNonnegCoeffs g) :
HasNonnegCoeffs (starC c f g)
Affine test for a single linear factor
theorem PreservesAffineRealRooted.linear_iff (c r : ℝ) :
PreservesAffineRealRooted c (X - C r) ↔ c * r ≤ 0
A root of the wrong sign obstructs real-rootedness
theorem starC_root_obstruction_not_splits {c r : ℝ} (hcr : 0 < c * r) :
¬(starC c (X - C r) (X - C (r + c))).Splits
Signed-symbol stability from the root-sign condition
theorem poissonSignedSymbol_stable {c : ℝ} {f : ℝ[X]} (hf : f ≠ 0)
(hfs : f.Splits) (hroot : ∀ r ∈ f.roots, c * r ≤ 0) :
MvUpperHalfPlaneStable (poissonSignedSymbol c f)
PF preservers at a negative parameter
theorem PreservesPF.iff_of_neg {c : ℝ} (hc : c < 0) (f : ℝ[X]) :
PreservesPF c f ↔ ∃ a : ℝ, 0 ≤ a ∧ f = C a
Real-rootedness preservers at parameter zero
theorem _root_.RealRooted.PreservesRealRootedOrZero.poissonStar_zero_iff {f : ℝ[X]} (hf : f ≠ 0) :
PreservesRealRootedOrZero (starCLinearMap 0 f) ↔ f.Splits
PF preservers at parameter zero
theorem PreservesPF.zero_iff (f : ℝ[X]) :
PreservesPF 0 f ↔ IsPFPolynomial f
Complete PF-preserver classification for every parameterMain result
theorem PreservesPF.iff (c : ℝ) (f : ℝ[X]) :
PreservesPF c f ↔
if 0 ≤ c then IsPFPolynomial f
else ∃ a : ℝ, 0 ≤ a ∧ f = C a
Real-rootedness preservation is equivalent to Euler-pencil stability
theorem realRootedOrZero_iff_eulerPencil_stable
{c : ℝ} {f : ℝ[X]} (hf0 : f ≠ 0) :
PreservesRealRootedOrZero (starCLinearMap c f) ↔
IsUpperHalfPlaneStablePencil (X * (f + C c * f.derivative)) f
Affine preservation is equivalent to Euler-pencil stability
theorem affineRealRooted_iff_eulerPencil_stable
{c : ℝ} {f : ℝ[X]} (hf0 : f ≠ 0) :
PreservesAffineRealRooted c f ↔
IsUpperHalfPlaneStablePencil (X * (f + C c * f.derivative)) f
Euler-pencil stability is equivalent to root-sign admissibility
theorem isUpperHalfPlaneStablePencil_euler_iff_isAdmissible
{c : ℝ} {f : ℝ[X]} (hf0 : f ≠ 0) :
IsUpperHalfPlaneStablePencil (X * (f + C c * f.derivative)) f ↔
IsAdmissible c f
Euler-pencil stability is equivalent to signed-symbol stability
theorem eulerPencil_stable_iff_signedSymbol_stable
{c : ℝ} {f : ℝ[X]} (hf0 : f ≠ 0) :
IsUpperHalfPlaneStablePencil (X * (f + C c * f.derivative)) f ↔
MvUpperHalfPlaneStable (poissonSignedSymbol c f)
Five-way classification for positive affine Poisson recurrences
theorem poisson_affine_recurrence_strict
{c : ℝ} {a b : ℕ → ℝ} {P : ℕ → ℝ[X]}
(hc : 0 < c) (ha : ∀ n, 0 < a n) (hb : ∀ n, 0 < b n)
(hP0 : P 0 = 1)
(hrec : ∀ n, P (n + 1) =
(C (a n) + C (b n) * X) * P n +
C (c * b n) * X * (P n).derivative) (n : ℕ) :
IsPFPolynomial (P n) ∧ HasSimpleRoots (P n) ∧
(∀ r, (P n).IsRoot r → r < 0) ∧ StrictInterl (P n) (P (n + 1)) ∧
∀ r, (P n).IsRoot r → ¬ (P (n + 1)).IsRoot r
The falling-factorial transform is multiplicative for the Poisson product
theorem fallingFactorialTransform_starC (c : ℝ) (f g : ℝ[X]) :
fallingFactorialTransform c (starC c f g) =
fallingFactorialTransform c f * fallingFactorialTransform c g
Associativity of the univariate Poisson product
theorem starC_assoc (c : ℝ) (f g h : ℝ[X]) :
starC c (starC c f g) h = starC c f (starC c g h)
Functional calculus: S_g((t − r) p) = (g − r) ⋆ S_g(p)RealRooted original
theorem derivProductPowMap_X_sub_C_mul (hc0 : c 0 = 1) (r : ℝ) (p : ℝ[X]) :
derivProductPowMap w c g ((X - C r) * p) =
derivProduct w c (g - C r) (derivProductPowMap w c g p)
For an associative product, S_g(p q) = S_g(p) ⋆ S_g(q)RealRooted original
theorem derivProductPowMap_mul (p q : ℝ[X]) :
derivProductPowMap w c g (p * q) =
derivProduct w c (derivProductPowMap w c g p) (derivProductPowMap w c g q)
Levels rooted in a ⋆-closed class give images rooted in that classRealRooted original
theorem isRealRootedIn_derivProductPowMap (hc0 : c 0 = 1) {T J : Set ℝ}
(hclosed : ∀ f q : ℝ[X], IsRealRootedIn T f → IsRealRootedIn T q →
IsRealRootedIn T (derivProduct w c f q))
(hg : ∀ r ∈ J, IsRealRootedIn T (g - C r)) {p : ℝ[X]} (hp : IsRealRootedIn J p) :
IsRealRootedIn T (derivProductPowMap w c g p)
Source-interval reduction to the one-factor levels g − rRealRooted original
theorem forall_isRealRootedIn_derivProductPowMap_iff (hc0 : c 0 = 1) {T J : Set ℝ}
(hclosed : ∀ f q : ℝ[X], IsRealRootedIn T f → IsRealRootedIn T q →
IsRealRootedIn T (derivProduct w c f q)) :
(∀ p : ℝ[X], IsRealRootedIn J p → IsRealRootedIn T (derivProductPowMap w c g p)) ↔
∀ r ∈ J, IsRealRootedIn T (g - C r)
The image of a real-rooted input has degree deg(p) deg(g)RealRooted original
theorem natDegree_derivProductPowMap (hg : g.natDegree ≠ 0) {S : Set ℝ} {p : ℝ[X]}
(hp : IsRealRootedIn S p) :
derivProductPowMap w c g p ≠ 0 ∧
(derivProductPowMap w c g p).natDegree = p.natDegree * g.natDegree
Leading coefficient of the image for a weight with vanishing quadratic partRealRooted original
theorem leadingCoeff_derivProductPowMap (hg : g.natDegree ≠ 0) (hA0 : w.coeff 2 = 0)
{S : Set ℝ} {p : ℝ[X]} (hp : IsRealRootedIn S p) :
(derivProductPowMap w c g p).leadingCoeff = p.leadingCoeff * g.leadingCoeff ^ p.natDegree
If two constant levels of g split, every intermediate level splitsRealRooted original
theorem splits_sub_C_of_mem_Icc {f : ℝ[X]} {a b s : ℝ} (hdeg : f.natDegree ≠ 0)
(hs : s ∈ Icc a b) (ha_splits : (f - C a).Splits) (hb_splits : (f - C b).Splits) :
(f - C s).Splits
Polynomials rooted in (−∞, 0] are closed under the Poisson productRealRooted original
theorem isRealRootedIn_Iic_starC {c : ℝ} (hc : 0 ≤ c) {f g : ℝ[X]}
(hf : IsRealRootedIn (Iic 0) f) (hg : IsRealRootedIn (Iic 0) g) :
IsRealRootedIn (Iic 0) (starC c f g)
Two endpoint levels rooted in (−∞, 0] give all intermediate levelsRealRooted original
theorem isRealRootedIn_Iic_sub_C_of_mem_Icc {g : ℝ[X]} (hg : g.natDegree ≠ 0) {a b r : ℝ}
(hr : r ∈ Icc a b) (ha : IsRealRootedIn (Iic 0) (g - C a))
(hb : IsRealRootedIn (Iic 0) (g - C b)) : IsRealRootedIn (Iic 0) (g - C r)
Poisson powers map [a, b]-rooted inputs to (−∞, 0]-rooted outputs iff both ends doRealRooted original
theorem forall_isRealRootedIn_starCPowMap_Icc_iff {c : ℝ} (hc : 0 ≤ c) {g : ℝ[X]}
(hg : g.natDegree ≠ 0) {a b : ℝ} (hab : a ≤ b) :
(∀ p : ℝ[X], IsRealRootedIn (Icc a b) p → IsRealRootedIn (Iic 0) (starCPowMap c g p)) ↔
IsRealRootedIn (Iic 0) (g - C a) ∧ IsRealRootedIn (Iic 0) (g - C b)
The same criterion, degree by degreeRealRooted original
theorem forall_mapsRealRootedIn_starCPowMap_Icc_iff {c : ℝ} (hc : 0 ≤ c) {g : ℝ[X]}
(hg : g.natDegree ≠ 0) {a b : ℝ} (hab : a ≤ b) :
(∀ n, MapsRealRootedIn (starCPowMap c g) n (Icc a b) (Iic 0)) ↔
IsRealRootedIn (Iic 0) (g - C a) ∧ IsRealRootedIn (Iic 0) (g - C b)
Degree and leading coefficient of the Poisson sequence mapRealRooted original
theorem natDegree_leadingCoeff_starCPowMap {c : ℝ} (hc : 0 ≤ c) {g : ℝ[X]}
(hg : g.natDegree ≠ 0) {S : Set ℝ} {p : ℝ[X]} (hp : IsRealRootedIn S p) :
starCPowMap c g p ≠ 0 ∧ (starCPowMap c g p).natDegree = p.natDegree * g.natDegree ∧
(starCPowMap c g p).leadingCoeff = p.leadingCoeff * g.leadingCoeff ^ p.natDegree
Self-product lemma: (q ⋆ q)(x) > 0 wherever the weight is positiveRealRooted original
theorem eval_assocProduct_self_pos (ha : 0 < a) (hA : 0 ≤ w.coeff 2) (hq : q ≠ 0) {x : ℝ}
(hx : 0 < w.eval x) : 0 < (assocProduct a w q q).eval x
A real-rooted self-product has all its roots in {w ≤ 0}RealRooted original
theorem isRealRootedIn_of_splits_assocProduct_self (ha : 0 < a) (hw : w.natDegree ≤ 2)
(hA : 0 ≤ w.coeff 2) (hq : q ≠ 0) (hs : (assocProduct a w q q).Splits) :
IsRealRootedIn {r | w.eval r ≤ 0} (assocProduct a w q q)
The square of left multiplication preserves real-rootedness iff q ⋆ q is real-rootedRealRooted original
theorem assocProduct_self_splits_iff (ha : 0 < a) (hw : w.natDegree ≤ 2) (hA : 0 ≤ w.coeff 2)
(hq : q ≠ 0) :
(assocProduct a w q q).Splits ↔ ∀ p : ℝ[X], IsRealRootedIn univ p →
IsRealRootedIn univ (assocProduct a w q (assocProduct a w q p))
All product powers are real-rooted iff q and q ⋆ q areRealRooted original
theorem forall_isRealRootedIn_derivProductPow_iff (ha : 0 < a) (hw : w.natDegree ≤ 2)
(hA : 0 ≤ w.coeff 2) (hq : q ≠ 0) :
(∀ n : ℕ, IsRealRootedIn univ (derivProductPow w (assocCoeff a (w.coeff 2)) q n)) ↔
q.Splits ∧ (assocProduct a w q q).Splits
A singleton source root is tested by g − r and (g − r) ⋆ (g − r)RealRooted original
theorem forall_isRealRootedIn_derivProductPowMap_singleton_iff (ha : 0 < a)
(hw : w.natDegree ≤ 2) (hA : 0 ≤ w.coeff 2) {g : ℝ[X]} {r : ℝ} (hgr : g - C r ≠ 0) :
(∀ p : ℝ[X], IsRealRootedIn {r} p →
IsRealRootedIn univ (derivProductPowMap w (assocCoeff a (w.coeff 2)) g p)) ↔
(g - C r).Splits ∧ (assocProduct a w (g - C r) (g - C r)).Splits
If w ≤ 0 on ℝ, the one-factor levels decide every source setRealRooted original
theorem derivProductPowMap_realRooted_iff_of_nonpos (ha : 0 < a)
(hw : w.natDegree ≤ 2) (hA : 0 ≤ w.coeff 2) (hW : ∀ x : ℝ, w.eval x ≤ 0) {J : Set ℝ}
{g : ℝ[X]} :
(∀ p : ℝ[X], IsRealRootedIn J p →
IsRealRootedIn univ (derivProductPowMap w (assocCoeff a (w.coeff 2)) g p)) ↔
∀ r ∈ J, IsRealRootedIn univ (g - C r)