#Interlacing and interleaving roots

Interlacing is one of the most flexible ways to certify real-rootedness. This page separates four recurring viewpoints:

  • root-level interlacing, Sturm sequences, and derivative recurrences;

  • matrix certificates such as Bézout matrices, Hurwitz matrices, and Lace matrices;

  • compatible polynomials and common interlacings;

  • operators and decompositions that preserve real-rootedness.

Coefficient-level total-positivity tools are collected on the Pólya frequency page, while the multivariate operator viewpoint is covered on the stable-polynomial page.

We first fix the interlacing convention used throughout, following [Wag92]; see also [Br15].

Definition (Interleaving polynomials).

Let \(f\) and \(g\) be polynomials with positive leading coefficients and with real roots \(\{f_{i}\}\) and \(\{g_{i}\},\) respectively. We say that \(f\) interlaces \(g\) if \(\deg(f)+1=\deg(g)=d\) and \[g_1 \leq f_1 \leq g_2 \leq f_2 \leq \dotsm \leq f_{d-1}\leq g_{d}.\] Moreover, we say that \(f\) alternates left of \(g\) if \(\deg(f)=\deg(g)=d\) and \[f_1 \leq g_1 \leq f_2 \leq \dotsm \leq f_{d}\leq g_{d}.\] We say that \(f\) interleaves \(g\) if either \(f\) interlaces \(g\) or \(f\) alternates left of \(g.\) We write this as \(f \interl g.\) By convention, \(0 \interl 0,\) \(0 \interl h\) and \(h \interl 0\) whenever \(h\) is a polynomial with a positive leading coefficient.

Lemma (Folklore).

Let \(f, g\) be monic polynomials with \(\deg(f) = \deg(g)+1\) and all roots real and nonpositive. If \(g \interl f,\) then \(f-tg\) is real-rooted and \(g \interl f- tg \interl f.\)

Wagner’s lemma is one of the basic closure tools for interlacing induction.

Lemma (Wagner’s lemma).

Let \(f, g, h \in \setR[t]\) be real-rooted polynomials with only real, nonpositive roots and positive leading coefficients. Then

  • if \(f \interl h\) and \(g \interl h\) then \(f+g \interl h.\)

  • if \(h \interl f\) and \(h \interl g\) then \(h \interl f+g.\)

  • \(g \interl f\) if and only if \(f \interl tg.\)

D.G. Wagner also shows that \(f \interl g\) implies that \(f \interl \lambda f + \mu g \interl g\) for all \(\lambda, \mu \geq 0.\)

As an example, if \(f\) is real-rooted, then \(f' \interl f.\) Hence, recursions involving the derivative play well with interlacing roots. The differential recurrence for Eulerian polynomials is a classical example: it produces a strict Sturm sequence. The Eulerian root argument records the sign analysis. The next three criteria cover derivative recurrences, two-term recurrences, and affine combinations, respectively. Vector recurrences are treated by the matrix criterion below.

Theorem (Cauchy interlacing theorem, see [God17, Fis05]).

Let \(A\) be a real symmetric \(n \times n\) matrix with eigenvalues \[\lambda_1 \leq \lambda_2 \leq \dotsb \leq \lambda_n,\] and let \(B\) be a principal \((n-1)\times(n-1)\) submatrix of \(A\) with eigenvalues \[\mu_1 \leq \mu_2 \leq \dotsb \leq \mu_{n-1}.\] Then \[\lambda_1 \leq \mu_1 \leq \lambda_2 \leq \mu_2 \leq \dotsb \leq \mu_{n-1} \leq \lambda_n.\] Equivalently, the characteristic polynomial of \(B\) interlaces the characteristic polynomial of \(A.\)

A useful derivative criterion is due to S.-M. Ma and Y. Wang [MW08, Thm. 2].

Theorem (Ma–Wang, 2008).

Suppose \(f\) and \(F\) are polynomials with positive leading coefficients, such that \[F(x) = u(x) f(x) + v(x) f'(x), \qquad \deg(f) \leq \deg(F) \leq \deg(f) + 1\] with \(u(x),v(x)\) having real coefficients. If \(f\) is real-rooted and \(v(r) \leq 0\) whenever \(f(r)=0,\) then \(F\) is real-rooted and \(f \interl F.\)

The theorem states more regarding root placements and multiplicities.

Favard recurrences give a second standard interlacing source, especially for orthogonal polynomial families.

Theorem (Favard interlacing, see [Fav35]).

Let \(P_0(t)=1,\) \(P_1(t)=t-\alpha_0,\) and \[P_{n+2}(t)=(t-\alpha_{n+1})P_{n+1}(t)-\beta_{n+1}P_n(t),\] where \(\alpha_n\in\setR\) and \(\beta_{n+1}\gt{}0.\) Then \[P_n \interl P_{n+1}\] for all \(n.\) In particular, each \(P_n\) is real-rooted.

The following criterion from [LW07, Thm. 2.1] is often useful.

Theorem (Liu–Wang, 2007).

Let \(F,f,g \in \setR[t]\) satisfy \[F(x) = a(x)f(x) + b(x)g(x), \qquad \deg(F) \in \{\deg(f), \deg(f)+1\},\] where \(a(x),b(x) \in \setR[t].\) Assume that \(f\) and \(g\) are real-rooted, that \(g \interl f,\) and that \(F\) and \(g\) have leading coefficients of the same sign. If \(b(r) \leq 0\) whenever \(f(r)=0,\) then \(F\) is real-rooted and \(f \interl F.\) In particular, if \(g\) strictly interlaces \(f\) and \(b(r) \lt 0\) whenever \(f(r)=0,\) then \(f\) strictly interlaces \(F.\)

The following affine-coefficient criterion is useful when neither summand is a derivative.

Theorem (Wang–Yeh, 2005, [WY05, Thm. 1]).

Let \(f,g\in\setR[x]\) be real-rooted polynomials with positive leading coefficients and \(g\interl f.\) If \(a,b,c,d\in\setR\) satisfy \(ad\geq bc,\) then \[(bx+a)f(x)+(dx+c)g(x)\] is real-rooted or zero.

By applying this criterion to row-generating polynomials, Wang and Yeh also give conditions ensuring that bilinear triangular recurrences produce Pólya frequency sequences.

#Sturm sequences

There are two related uses of the term Sturm sequence. In enumerative combinatorics, it is fairly common to see sequences of polynomials \(\{ P_n(t) \}_{n \geq 0}\) with the property that \(P_j(t) \interl P_{j+1}(t)\) for all \(j.\) If the leading coefficient is always positive and \(\deg(P_n)=n,\) this is called a Sturm sequence, see [LW07]. If we drop the degree condition, the sequence is called a generalized Sturm sequence.

The notion of Sturm-unimodality is defined in the analogous manner.

There are many Sturm sequences on permutations and Catalan objects.

The classical Sturm sequence, or Sturm chain, instead starts from one polynomial \(f.\) Set \(S_0=f,\) \(S_1=f',\) and recursively take the negative remainders \[S_{j+1}=-\operatorname{rem}(S_{j-1},S_j)\] until the remainder is zero. For a real number \(x,\) let \(V_f(x)\) be the number of sign changes in \[S_0(x),S_1(x),\dotsc\] after zero entries are omitted.

Theorem (Sturm’s theorem).

If \(a\lt{}b\) and neither endpoint is a zero of \(f,\) then the number of distinct real zeros of \(f\) in \((a,b)\) is \[V_f(a)-V_f(b).\]

In particular, evaluating the signs at \(-\infty\) and \(+\infty\) gives the total number of distinct real zeros. This provides an exact coefficient-based test for real-rootedness: if \(g\) is the squarefree part of \(f,\) then \(f\) is real-rooted if and only if Sturm’s theorem counts \(\deg(g)\) distinct real zeros of \(g.\)

Example (Baxter and Hoggatt polynomials).

Let \[P\!B_n(t)=\sum_{\pi\in\mathcal B_n}t^{\des(\pi)}\] be the descent polynomial of the Baxter permutations. For \(m,n\geq1,\) define the Hoggatt polynomials by \[H_n^{[m]}(1,t)= \sum_{k=0}^{n-1} \frac{\displaystyle\prod_{i=0}^{m-1} \binom{n+m-2}{k+i}} {\displaystyle\prod_{j=1}^{m-1} \binom{n+m-2}{j}}t^k.\] The case \(m=3\) is \(P\!B_n(t).\) Consecutive Baxter polynomials strictly interlace. More generally, \(H_n^{[1]}(1,t)\) weakly interlaces \(H_{n+1}^{[1]}(1,t),\) while the interlacing is strict for every fixed \(m\geq2\) [FZZ25, Thms. 1.1–1.2]. The coefficient triangles for \(m=1,2,3\) are, respectively, A007318, A001263, and A056939; the row sums in the last case are the Baxter numbers A001181.

Example (Separable-permutation descent polynomials).

The descent polynomials \(S_n(t)\) of separable permutations have \(n-1\) simple negative zeros, and \(S_n\) strictly interlaces \(S_{n+1}\) for \(n\geq2,\) by Zhanhe Zhang’s preprint [Zha26, Thm. 1.1].

Example (Elephant polynomials).

For a real parameter \(a,\) define \(R_1(x)=x\) and \[R_{n+1}(x)=xR_n(x)-\frac an(1-x^2)R'_n(x).\] If \(a\gt{}0,\) then \(R_n\) has \(n\) distinct zeros in \((-1,1),\) and consecutive polynomials interlace. At \(a=0,\) one has \(R_n(x)=x^n.\)

If \(a\lt{}0,\) the rotated polynomial \(S_n(x)=(-\mathrm{i})^nR_n(\mathrm{i}x)\) is real-rooted, and consecutive \(S_n\) interlace. Its zeros are distinct: there are \(n\) of them unless \(a=-1,\) when there are \(|n-2|\) and \(S_2\) is constant [GLR24, Props. 1–2]. The parameter comes from the elephant random walk as \(a=2p-1\) for \(p\in[0,1],\) although the polynomial family is defined for every \(a\in\setR.\) The specialization \(a=-1/2\) uses the even columns of Pascal’s triangle A034839; at \(a=-1,\) a reciprocal rescaling gives the triangle A101280.

#Bézout matrices and Hurwitz–Lace

For arbitrary polynomials \(f(t)\) and \(g(t)\) with \(\deg(f) = \deg(g)+1,\) let \(B(f,g)\) be the Bézout matrix, whose \((i,j)\) entry is equal to the coefficient of \(x^{i-1} y^{j-1}\) in the polynomial \[\frac{f(x)g(y) - f(y)g(x)}{x-y}.\]

Theorem (Bézout’s theorem, [Fis06, Cor. 9.145]).

Let \(f(t)\) be a polynomial of degree \(d\) and \(g(t)\) a polynomial of degree \(d-1.\) Then \(f(t)\) and \(g(t)\) are both real-rooted and \(g\) strictly interlaces \(f\) if and only if \(B(f,g)\) is positive definite.

Theorem (Hurwitz–Lace criterion).

Let \[p_0(t)=a_0+a_1t+a_2t^2+\dotsb,\qquad p_1(t)=b_0+b_1t+b_2t^2+\dotsb\] have nonnegative coefficients. Form the Hurwitz matrix \[H(p_0,p_1)= \begin{pmatrix} b_0 & b_1 & b_2 & b_3 & \dotsb \\ a_0 & a_1 & a_2 & a_3 & \dotsb \\ 0 & b_0 & b_1 & b_2 & \dotsb \\ 0 & a_0 & a_1 & a_2 & \dotsb \\ 0 & 0 & b_0 & b_1 & \dotsb \\ 0 & 0 & a_0 & a_1 & \dotsb \\ \vdots & \vdots & \vdots & \vdots & \ddots \end{pmatrix}.\] Then \(H(p_0,p_1)\) is totally nonnegative if and only if \(p_1 \interl p_0.\)

This criterion packages interlacing roots as total nonnegativity of a structured Hurwitz matrix.

Definition (Hurwitz stability).

A nonzero polynomial \(P\in\setR[t]\) is weakly Hurwitz stable if every zero \(z\) of \(P\) satisfies \(\operatorname{Re}(z)\leq 0.\) It is strictly Hurwitz stable if every zero satisfies \(\operatorname{Re}(z)\lt{}0.\) The unqualified term Hurwitz stable is also commonly used for the strict version.

Proposition (Hurwitz stability implies coefficient positivity).

Let \(P\in\setR[t]\) have positive leading coefficient. If \(P\) is weakly Hurwitz stable, then all coefficients of \(P\) are nonnegative. If \(P\) is strictly Hurwitz stable, then all its coefficients are positive.

Proof

Over the reals, \(P\) is a positive constant times a product of linear factors \(t+a,\) with \(a\geq 0,\) and quadratic factors \[(t-z)(t-\overline z) =t^2-2\operatorname{Re}(z)t+|z|^2\] coming from nonreal conjugate pairs. Every factor has nonnegative coefficients when its zeros lie in the closed left half-plane. In the strict case every displayed coefficient is positive, and the same conclusions hold for their product.

The converse fails even when every coefficient is positive. For example, \[(t+3)(t^2-2t+10)=t^3+t^2+4t+30\] has positive coefficients but has the two zeros \(1\pm3i\) in the open right half-plane. Thus Hurwitz stability is a sufficient certificate for coefficient positivity, but is strictly stronger than coefficient positivity.

Theorem (Hermite–Biehler, see [Hol03]).

Let \[P(z)=a_0+a_1z+\dotsb+a_nz^n\] be a real polynomial, and define its even and odd parts by \[P_E(t)=\sum_{k\geq 0} a_{2k}t^k, \qquad P_O(t)=\sum_{k\geq 0} a_{2k+1}t^k.\] Assume that \(P_E P_O\not\equiv 0\) and that \(P_E\) and \(P_O\) have positive leading coefficients. Then \(P\) has all zeros in the closed left half-plane if and only if \(P_E\) and \(P_O\) have only real nonpositive zeros and \(P_O \interl P_E.\)

Thus the Hermite–Biehler and Routh–Hurwitz criteria also translate root-location questions into interlacing and total nonnegativity. The block-Toeplitz and network viewpoint is discussed in [Ang23, Sec. 4], and weak/quasi-stable boundary cases are treated in [AGT17].

#Interlacing sequences and matrix criteria

The following definitions are from [Br15].

Definition (Interlacing sequences of polynomials).

A sequence \(\{f_1,f_2,\dotsc,f_n\}\) of polynomials with positive leading coefficients is said to be an interlacing sequence if \(f_i \interl f_j\) for all \(1\leq i \lt j \leq n.\)

We let \(\mathscr{F}_n^+\) denote the family of interlacing sequences of length \(n\) with nonnegative coefficients.

P. Brändén gives useful examples of how interlacing sequences can be used to prove real-rootedness [Br15]. In particular, rook polynomials associated with Ferrers diagrams can be proved to be real-rooted using this method.

Theorem (Shelling–interlacing criterion, [HS22, Thm. 3.1]).

Let \(C\) be a shellable polytopal complex with facets \(F_1,\dotsc,F_s,\) and let \(C'\to C\) be a subdivision. If the \(h\)-polynomials of the relative subdivision pieces associated with the facets can be ordered as an interlacing sequence, then \(h(C',x)\) is real-rooted.

The interlacing hypothesis is essential: an arbitrary positive sum of real-rooted polynomials need not be real-rooted. A useful specialization is that the barycentric subdivision of every cubical complex admitting a stable shelling has a real-rooted \(h\)-polynomial [HS22, Thm. 4.7].

Proposition ([Wag92]).

Suppose \(f_1 \interl f_2 \interl \dotsb \interl f_n\) and \(f_1 \interl f_n.\) Then \(\{f_1,\dotsc,f_n\}\) is an interlacing sequence.

#Matrices preserving interlacing sequences

The following matrix criterion is proved in [Br15, Thm. 7.8.5].

Theorem (Matrices preserving interlacing sequences).

Let \(G\) be an \(m \times n\) matrix of polynomials in \(t.\) Then \(G : \mathscr{F}_n^+ \to \mathscr{F}_m^+\) if and only if

  1. All entries of \(G\) have nonnegative coefficients, and

  2. For every \(2{\times}2\) submatrix \(\begin{bmatrix} a(t) & b(t) \\ c(t) & d(t) \end{bmatrix}\) of \(G\) and real numbers \(\lambda, \mu \gt{} 0,\) we have \[(\lambda t + \mu) b(t) + d(t) \interl (\lambda t + \mu) a(t) + c(t).\]

For a constant matrix, total non-negativity gives a simpler sufficient test.

Corollary (TNN mixing, [Fis06, Thm. 3.7]).

Let \(f_1,\dotsc,f_n\) be an interlacing sequence and let \(A\) be an \(m\times n\) totally non-negative real matrix. Then the nonzero coordinates of \[A(f_1,\dotsc,f_n)^{\mathsf T}\] form an interlacing sequence in row order.

Thus a coefficient matrix supplied by a planar network can be used directly to mix an interlacing refinement.

Example (Small matrices preserving interleaving sequences).

There are 81 matrices with entries in \(\{0,1,t\}.\) Remove those whose first and last rows are equal and those where one row is \((0,0).\) This leaves 56 matrices; among them, the following 18 satisfy the conditions in the theorem above. The nested lists give the two rows of each matrix.

[
  [[0, 1], [0, t]],
  [[0, 1], [t, 0]],
  [[1, 0], [0, 1]],
  [[1, 0], [t, 0]],
  [[t, 0], [0, t]],
  [[0, 1], [t, 1]],
  [[0, 1], [t, t]],
  [[1, 0], [1, 1]],
  [[1, 0], [t, 1]],
  [[1, 1], [0, 1]],
  [[1, 1], [t, 0]],
  [[t, 0], [t, t]],
  [[t, 1], [0, t]],
  [[t, 1], [t, 0]],
  [[t, t], [0, t]],
  [[1, 1], [t, 1]],
  [[1, 1], [t, t]],
  [[t, 1], [t, t]]
]

#Compatible polynomials

Chudnovsky and Seymour introduced a concept closely related to interleaving sequences of polynomials. The polynomials \(f_1,\dotsc,f_k \in \setR[z]\) have a common interlacing if some \(h \in \setR[z]\) exists with \(h \interl f_j\) for all \(j.\)

Theorem (Obreschkoff–Dedieu, see [Obr63, Ded92] and [Br04, Thm. 8]).

Let \(f,g \in \setR[x]\) have positive leading coefficients. Then every polynomial in the real pencil \[\{ \alpha f+\beta g : \alpha,\beta\in\setR\}\] is real-rooted or zero if and only if \(f\) and \(g\) are real-rooted and one of \(f \interl g\) or \(g \interl f\) holds.

This signed pencil criterion is the two-polynomial counterpart of the nonnegative compatibility theorem below.

Theorem (Chudnovsky–Seymour, 2007, [CS07]).

Let \(\{f_1,f_2,\dotsc,f_n\}\) be a set of real-rooted polynomials with positive leading coefficients. Such a set is said to be compatible if for all choices \(\lambda_j\geq 0,\) the sum \(\sum_{j=1}^n \lambda_j f_j\) is real-rooted.

Then the following are equivalent:

  • \(f_1,\dotsc,f_n\) are pairwise compatible;

  • \(f_1,\dotsc,f_n\) are compatible;

  • \(f_i,f_j\) have a common interlacing for all \(1\leq i \lt j \leq n\);

  • \(f_1,\dotsc,f_n\) have a common interlacing.

L. L. Liu studies this compatible-zero method further in [Liu12]. In particular, she gives a necessary and sufficient criterion for two real-rooted polynomials with opposite signs of their leading coefficients to be compatible, and a direct induction proof of the pairwise-compatible positive-leading-coefficient case. Her criterion is a root-count condition. For nonconstant \(f\) and \(g,\) the correct statement is that \(f\) and \(g\) are compatible if and only if either this condition holds, or they share a zero \(r\) such that \(f/(x-r)\) and \(g/(x-r)\) are compatible. The second alternative is missing from the published statement, and it is needed: \(x\) and \(-x^2\) are compatible but fail the root-count condition.

Theorem (Mixed-sign compatibility, [LR24, Thm. 1.2]).

Let \(f_1,\dotsc,f_n\in\setR[x]\) be real-rooted; their leading coefficients may have different signs. Then the family is compatible if and only if every triple \((f_i,f_j,f_k)\) is compatible. If no convex combination of the \(f_i\) is zero, these conditions are also equivalent to the existence of a real-rooted polynomial \(g\) such that every \(f_i\) is in proper position with \(g.\)

Thus mixed signs require a triple test rather than the pairwise test in the positive-leading-coefficient case.

Another useful technique is presented in [Ben18], where Christoffel–Darboux type identities are used to show interlacing. This exploits an idea similar to Bézout matrices. By taking the limit, \[\lim_{y\to x} \frac{f(x)g(y)-f(y)g(x)}{x-y} = f'(x)g(x) - g'(x)f(x)\] which is the Wronskian of \(f\) and \(g.\)

Proposition (Wronskian criterion).

Let \(f\) and \(g\) be real-rooted polynomials with positive leading coefficients and \(\deg(f)-\deg(g)\in\{0,1\}.\) If \(g \interl f,\) then \[f'(x)g(x)-f(x)g'(x)\geq 0\] for all \(x\in\setR.\) Conversely, if this Wronskian is nonnegative on \(\setR\) and every zero of \(f\) has multiplicity at most one more than its multiplicity as a zero of \(g\) (for instance, if \(f\) has only simple zeros), then \(g \interl f.\)

The multiplicity condition cannot be dropped: \(f=(x-1)^3\) and \(g=x^3\) have equal degree and the nonnegative Wronskian \(3x^2(x-1)^2,\) but \(x^3\) does not interleave \((x-1)^3.\)

Thus one way to prove an interlacing statement is to show that this Wronskian is nonnegative, perhaps by rewriting it as a sum of squares. This is especially useful when a Christoffel–Darboux identity gives a closed form for the difference quotient above.

Theorem (Interlacing families, see [Lau22]).

Let \(T\) be a rooted tree whose leaves are labeled by real-rooted polynomials of the same degree and with positive leading coefficients. Label every internal vertex by the sum of the polynomials labeling its children. Suppose that the children of every internal vertex have a common interlacing. Then, for each root index \(j,\) leaves \(a\) and \(b\) exist such that \[\lambda_j(p_a)\leq \lambda_j(p_{\emptyset})\leq \lambda_j(p_b),\] where \(\lambda_j\) denotes the \(j\)th largest zero. In particular, some leaf has largest zero at most the largest zero of the root polynomial.

For example, if \(A\) is a real symmetric matrix and \(A_S\) denotes the principal submatrix obtained by deleting the rows and columns indexed by \(S,\) then the characteristic polynomials \(\{\det(xI-A_S): |S|=k\}\) form an interlacing family. The root polynomial is the \(k\)th derivative of the characteristic polynomial of \(A.\) This refines the Cauchy interlacing theorem from a single principal submatrix to a whole tree of principal submatrices.

The method was introduced by A. Marcus, D. Spielman, and N. Srivastava for graph signings. The Godsil–Gutman identity says that the expected characteristic polynomial of a uniformly random signed adjacency matrix is the matching polynomial. They proved that the signed characteristic polynomials form an interlacing family. Combined with the matching-polynomial root bound, this selects a signing whose largest eigenvalue is at most \(2\sqrt{d-1}\) for a graph of maximum degree \(d.\) Applied repeatedly to \(2\)-lifts of bipartite graphs, this gives infinite families of bipartite Ramanujan graphs of every degree [MSS15].

Their sequel combines interlacing families with mixed characteristic polynomials and barrier functions; this yields a positive solution of the Kadison–Singer problem [MSS15].

#Fully interlacing sequences and the \(\operatorname{Lace}\) matrix

A stronger coefficient-level version of interlacing was introduced by C. A. Athanasiadis and D. G. Wagner [AW24].

Definition (The \(\operatorname{Lace}\) matrix).

Let \(A=(A_{ij}(x))\) be a \(p \times q\) matrix of formal power series, indexed by \(0 \leq i \lt p\) and \(0 \leq j \lt q,\) and write \[A_{ij}(x)=\sum_{n\geq 0} a_{ij}(n)x^n,\] with \(a_{ij}(n)=0\) for \(n \lt 0.\) The infinite Lace matrix \(\operatorname{Lace}(A)\) has integer rows and columns. If \[u=pu' + i,\qquad v=qv' + j,\qquad 0\leq i \lt p,\quad 0\leq j \lt q,\] then \[\operatorname{Lace}(A)_{u,v}=a_{ij}(v'-u').\] Equivalently, \(\operatorname{Lace}(A)\) is the block Toeplitz coefficient-convolution matrix associated to multiplication by \(A.\)

For a column vector \(A=(A_0,A_1,\dotsc,A_{p-1})^T,\) Athanasiadis–Wagner call the sequence fully interlacing if \(\operatorname{Lace}(A)\) is totally nonnegative. Thus every finite minor of this infinite coefficient-convolution matrix is nonnegative.

Full interlacing implies ordinary pairwise interlacing, but the converse is false. The distinction is useful: pairwise interlacing is a root-level condition, whereas full interlacing is a total-nonnegativity condition on a structured coefficient matrix.

Whenever the product \(AB\) is defined, \[\operatorname{Lace}(AB)=\operatorname{Lace}(A)\operatorname{Lace}(B).\] This is convenient, since totally nonnegative matrices are closed under multiplication.

Example (Pairwise interlacing need not be fully interlacing).

The following example is from [AW24, Ex. 3.4]. Let \[P(x)=t+x,\qquad Q(x)=(b+x)(d+x),\qquad R(x)=(a+x)(c+x).\] If \[a \leq b \leq t \leq c \leq d,\] then \(P,Q,R\) are pairwise interlacing. However, the \(3\times 3\) minor \[\begin{pmatrix} t & 1 & 0 \\ bd & b+d & 1 \\ ac & a+c & 1 \end{pmatrix}\] of \(\operatorname{Lace}((P,Q,R)^T)\) has determinant \[t(b+d-a-c)-(bd-ac),\] which can be negative. For example, with \(a=1,\) \(b=2,\) \(t=2,\) \(c=3,\) and \(d=4,\) this determinant is \(-1.\) Thus ordinary pairwise interlacing does not imply full interlacing.

#Preserving operators and symmetric decompositions

The previous sections treat interlacing as a certificate. The next examples show how the certificate is produced by structural operations: symmetric decompositions in Ehrhart theory, differential operators, products of Ehrhart functions, and basis transformations such as the derangement and Narayana transformations.

#Symmetric decompositions

The following setup from [BS18] is especially useful for \(h\)-polynomials arising in Ehrhart theory.

Given \(p \in \setR[x]\) with \(\deg(p) \leq d,\) define \[I_d(p)(x) \coloneqq x^d p(1/x).\] Then unique polynomials \(a,b \in \setR[x]\) exist such that \[p(x) = a(x) + x b(x), \qquad I_d(a)=a, \qquad I_{d-1}(b)=b.\] This is the symmetric \(I_d\)-decomposition of \(p,\) and it is given explicitly by \[a = \frac{p - x I_d(p)}{1-x}, \qquad b = \frac{I_d(p)-p}{1-x}.\]

It is also convenient to associate to \(p\) the polynomial \[f(x) \coloneqq (1+x)^d p\!\left( \frac{x}{1+x}\right),\] and to define \[R_d(f)(x) \coloneqq (-1)^d f(-1-x).\] The paper uses the proper-position relation \(\prec\); in the present nonnegative coefficient setting, this can be read as an interlacing relation.

Theorem (Brändén–Solus, [BS18, Thm. 2.7]).

Let \(p \in \setR[x]\) have degree at most \(d,\) and let \((a,b)\) be its symmetric \(I_d\)-decomposition. Assume that both \(a\) and \(b\) have nonnegative coefficients. Then the following are equivalent:

  • \(b \prec a\);

  • \(a \prec p\);

  • \(b \prec p\);

  • \(I_d(p) \prec p\);

  • \(R_d(f) \prec f.\)

In particular, these conditions imply that \(p,\) \(a,\) and \(b\) are all real-rooted.

Thus, an interlacing statement between a polynomial and its reciprocal forces the symmetric pieces of the decomposition to be real-rooted as well. This is one reason the theorem shows up naturally for reflexive \(h^*\)-polynomials; see magic positivity on the polytope page.

For integers \(r\geq1\) and \(d\geq0,\) define the Veronese numerator operator \(U_r^{d+1}\) by \[\sum_{n\geq0}a_{rn}t^n =\frac{U_r^{d+1}h(t)}{(1-t)^{d+1}} \qquad\text{whenever}\qquad \sum_{n\geq0}a_nt^n=\frac{h(t)}{(1-t)^{d+1}}.\]

Theorem (Jochemko, [Joc22, Thms. 1.1–1.2]).

Let \(h(t)=h_0+\dotsb+h_st^s,\) where \(s\leq d,\) have nonnegative coefficients, and put \(h_i=0\) outside \(0\leq i\leq s.\) Suppose that \[\sum_{j=0}^i h_j\geq\sum_{j=0}^{i-1}h_{d-j} \quad\text{and}\quad \sum_{j=0}^i h_j\leq\sum_{j=0}^i h_{s-j} \qquad\text{for every }i\geq0.\] Then \(U_r^{d+1}h\) has a nonnegative real-rooted symmetric decomposition for \(r\geq\max\{s,d+1-s\}.\) If instead \(h(t)=t^sh(1/t),\) its symmetric decomposition is nonnegative and interlacing in the same range.

For a \(d\)-dimensional lattice polytope, \(U_r^{d+1}h_P^*=h_{rP}^*.\) Consequently every such dilation in the stated range has a real-rooted symmetric decomposition; it is interlacing when \(P\) is Gorenstein [Joc22, Cor. 6.1–6.2].

There is a parallel tool for colored barycentric subdivisions. Given \(h(x)=\sum_{i=0}^n c_ix^i,\) set \[f(x)=\sum_{i=0}^n c_ix^i(1+x)^{n-i}, \qquad f(0)+\sum_{m\geq1}\bigl(f(rm)-f(rm-1)\bigr)x^m =\frac{D_{n,r}h(x)}{(1-x)^n}.\] For every \(r\geq1,\) \(D_{n,r}h\) is real-rooted whenever the \(c_i\) are nonnegative. If additionally \(\sum_{j=0}^i c_j\leq\sum_{j=0}^i c_{n-j}\) for \(0\leq i\leq\lfloor n/2\rfloor,\) its symmetric decomposition is also nonnegative and real-rooted [AT21, Thm. 1.3].

#Operators preserving real-rootedness

A multivariate approach to preserving real-rootedness is provided by stable polynomials, where linear operators preserving stability imply real-rootedness of specializations.

Theorem (Hermite–Poulain, see [Lev64, Obr63, Sch14] and [Br04, Thm. 11]).

Let \[f(x)=a_0+a_1x+\dotsb+a_nx^n\] and \(g(x)\) be real-rooted polynomials. Then \[f\!\left(\frac{d}{dx}\right)g \coloneqq a_0g(x)+a_1g'(x)+\dotsb+a_ng^{(n)}(x)\] is real-rooted.

Corollary (Hermite–Poulain preserves interleaving).

Let \(f_1,f_2,g\in\setR[x],\) and assume that \(g\) is real-rooted. If \(f_1\) and \(f_2\) are real-rooted and interleave, then \(f_1(d/dx)g\) and \(f_2(d/dx)g\) also interleave.

Proof

By the Obreschkoff–Dedieu theorem, the assumption on \(f_1\) and \(f_2\) is equivalent to saying that every polynomial in the pencil \[\alpha f_1+\beta f_2,\qquad \alpha,\beta\in\setR,\] is real-rooted or zero. Applying the Hermite–Poulain theorem to this polynomial and to \(g\) gives \[(\alpha f_1+\beta f_2)\!\left(\frac{d}{dx}\right)g = \alpha f_1\!\left(\frac{d}{dx}\right)g +\beta f_2\!\left(\frac{d}{dx}\right)g,\] which is real-rooted or zero for all real \(\alpha\) and \(\beta.\) Applying Obreschkoff–Dedieu once more shows that \(f_1(d/dx)g\) and \(f_2(d/dx)g\) interleave. This proves the claim.

D.G. Wagner proves the following result [Wag92].

Theorem

Suppose \(f,g\) are polynomials for which \[\sum_{n \geq 0} f(n) z^n = \frac{W(f)(z)}{(1-z)^{1+\deg f}}, \quad \sum_{n \geq 0} g(n) z^n = \frac{W(g)(z)}{(1-z)^{1+\deg g}},\] where \(W(f)(z)\) and \(W(g)(z)\) are real-rooted with nonpositive roots. Then \(W(fg)(z)\) also has real and nonpositive roots.

In particular, this shows that if \(f\) and \(g\) are Ehrhart functions of two polytopes \(P_f\) and \(P_g\) with real-rooted \(h^*\)-polynomials, then the \(h^*\)-polynomial of the polytope product \(P_f * P_g\) is also real-rooted. For examples of polytopes with real-rooted \(h^*\)-polynomials, see tableaux and polytopes.

Theorem (See [Br04, Thm. 9]).

If \(\phi : \setR[x] \to \setR[x]\) is a linear operator that preserves real-rootedness, then \(\phi\) preserves the interleaving property up to order: if \(f \interl g,\) then \[\phi(f) \interl \phi(g) \quad\text{or}\quad \phi(g) \interl \phi(f),\] after multiplying \(\phi(f)\) and \(\phi(g)\) by nonzero constants so that their leading coefficients are positive.

The order cannot be fixed in general. The operator \(p(x)\mapsto p(-x)\) preserves real-rootedness and \(x \interl x-1,\) but the normalized images satisfy \(x+1 \interl x\) rather than \(x \interl x+1.\)

#Schur–Szegő composition and interlacing

At a fixed degree bound, Schur–Szegő composition with a fixed polynomial is a diagonal operator, so the finite multiplier criterion applies. Kostov [Kos10] proves interlacing properties for polynomials arising from eigenvectors of the coefficient map defined by Schur–Szegő factorization.

For separable-permutation descent polynomials, Zhang compares two degree-dependent diagonal compression maps. He represents their outputs through one common finite Schur–Szegő multiplier and explicit kernels [Zha26, Thm. 4.1 and Lem. 4.2].

#Derangement and Narayana transformations

Another operator with strong interlacing properties is the following. Let \(d_n(x)\) denote the classical derangement polynomial, and define the derangement transformation \(D : \setR[x] \to \setR[x]\) by \[D(x^n) \coloneqq d_n(x).\] If \(p\) and \(q\) are degree-\(n\) polynomials with roots \(\alpha_1 \leq \dotsb \leq \alpha_n\) and \(\beta_1 \leq \dotsb \leq \beta_n\) in \([-1,0],\) write root order \(p \leq_{\mathrm{root}} q\) if \(\alpha_k \leq \beta_k\) for all \(k.\)

Theorem (Brändén–Solus, [BS18, Thm. 3.6]).

If \(p\) and \(q\) have all roots in \([-1,0]\) and \(p \leq_{\mathrm{root}} q,\) then \(D(p) \prec D(q).\)

In particular, \(D(p)\) and \(D(q)\) are real-rooted and have interlacing zeros. As a consequence, if \[p(x)=\sum_{k=0}^n c_k x^k (x+1)^{n-k}, \qquad c_k \geq 0,\] then [BS18, Cor. 3.7] gives \[A_n \prec D(p) \prec d_n.\]

For \(0\leq k\leq n,\) define \[d_{n,k}(x) \coloneqq D\bigl(x^k(1+x)^{n-k}\bigr) =\sum_{\substack{\pi\in\symS_n\\ \operatorname{Fix}(\pi)\subseteq[n-k]}} x^{\exc(\pi)}.\] The excedance interpretation is due to C. A. Athanasiadis [Ath25, Prop. 4.1(a)]. Thus \(d_{n,0}(x)=A_n(x),\) while \(d_{n,n}(x)=d_n(x).\)

More generally, suppose that fixed points are forbidden on any specified set \(S\subseteq[n]\) with \(|S|=k.\) Fixed-point deletion preserves the excedance number, so inclusion-exclusion gives \[\sum_{\substack{\pi\in\symS_n\\ \operatorname{Fix}(\pi)\cap S=\varnothing}} x^{\exc(\pi)} =\sum_{T\subseteq S}(-1)^{|T|}A_{n-|T|}(x) =\sum_{i=0}^k(-1)^i\binom{k}{i}A_{n-i}(x) =d_{n,k}(x).\] Consequently, only the cardinality of the forbidden set matters.

The polynomials \(x^k(1+x)^{n-k}\) form a chain in root order. Hence, for every \(0\leq i\lt{}j\leq n,\) [BS18, Thm. 3.6 and Cor. 3.7] imply \(d_{n,i}(x)\prec d_{n,j}(x)\); in particular, \[d_{n,0}(x)\prec d_{n,1}(x)\prec\dotsb\prec d_{n,n}(x).\] Therefore \((d_{n,k}(x))_{0\leq k\leq n}\) is an interlacing sequence and every \(d_{n,k}(x)\) is real-rooted. The proof of [BS18, Lem. 3.5] already keeps separate parameters for the individual fixed points, giving a stronger fixed-point-weight refinement before specialization.

This is one route to the interlacing results for colored Eulerian and colored derangement polynomials on the permutation page.

There is an analogous basis-transformation viewpoint for other classical polynomial triangles. The Eulerian transformation \(A(x^n)=A_n(x)\) does not preserve real-rootedness in complete generality [BJ22], but it does preserve real-rootedness on the nonnegative cone discussed on the gamma-positivity page [Ath25]. In parallel, J. Mao and L. Wang prove that the Narayana transformation \(T_{N_m}(x^n)=N_{n,m}(x),\) interpolating between the type \(B\) and type \(A\) Narayana polynomials, preserves real-rootedness for polynomials with nonnegative coefficients; see Narayana transformations and [MW26].

L. L. Liu and X. Yan [LY25] refine this picture by studying the triangular matrices arising from binomial-type transforms. For the derangement transformation, for every fixed \(k\) the column \[d_{k,k}(x),\,d_{k+1,k}(x),\,d_{k+2,k}(x),\dotsc\] is a generalized Sturm sequence. Their recurrence framework also recovers the fixed-\(n\) row interlacing above. They prove analogous statements for the type \(B\) and \(r\)-colored derangement transformations, and obtain sufficient conditions for such row and column interlacing in a more general recurrence setup.

Remark (Pólya frequency sequences and TNN).

For the coefficient-level total-positivity background used above, see Pólya frequency sequences. In that language, ordinary PF sequences are TNN Toeplitz matrices, while fully interlacing sequences are built from block Toeplitz matrices.

Bibliography

  1. [AGT17]Mohammad Adm, Jürgen Garloff and Mikhail Tyaglov. Total nonnegativity of finite Hurwitz matrices and root location of polynomials. Journal of Math. Ann. Appl., 467 (2018), no. 1, pp. 148 - 170, 2017.
    .bib
    @article{AdmGarloffTyaglov2017,
      author = {Mohammad Adm and J{\"u}rgen Garloff and Mikhail Tyaglov},
      title = {Total nonnegativity of finite {H}urwitz matrices and root location
        of polynomials},
      year = {2017},
      journal = {Journal of Math. Ann. Appl., 467 (2018), no. 1, pp. 148 - 170},
      doi = {10.1016/j.jmaa.2018.06.065},
      url = {https://doi.org/10.1016/j.jmaa.2018.06.065}
    }
    
  2. [Ang23]Robert Angarone. Cylindrical Networks and Total Nonnegativity. arXiv:2312.09385v2, 2023.
    .bib
    @article{Angarone2023x,
      author = {Robert Angarone},
      title = {Cylindrical {N}etworks and {T}otal {N}onnegativity},
      year = {2023},
      eprint = {2312.09385v2},
      url = {https://arxiv.org/abs/2312.09385v2},
      journal = {arXiv e-prints}
    }
    
  3. [Ath25]Christos A. Athanasiadis. On the real-rootedness of the Eulerian transformation. Journal of the London Mathematical Society, 111(2), 2025.
    .bib
    @article{Athanasiadis2025EulerianTransformation,
      author = {Athanasiadis, Christos A.},
      title = {On the real-rootedness of the {E}ulerian transformation},
      journal = {Journal of the London Mathematical Society},
      volume = {111},
      number = {2},
      year = {2025},
      doi = {10.1112/jlms.70083},
      url2 = {https://doi.org/10.1112/jlms.70083},
      eprint = {2302.00754},
      archivePrefix = {arXiv}
    }
    
  4. [AT21]Christos A. Athanasiadis and Eleni Tzanaki. Symmetric decompositions, triangulations and real-rootedness. Mathematika, 67(4):840–859, 2021.
    .bib
    @article{AthanasiadisTzanaki2021,
      author = {Christos A. Athanasiadis and Eleni Tzanaki},
      title = {Symmetric decompositions, triangulations and real-rootedness},
      year = {2021},
      journal = {Mathematika},
      volume = {67},
      number = {4},
      pages = {840--859},
      doi = {10.1112/mtk.12109},
      eprint = {2103.03639v1}
    }
    
  5. [AW24]Christos A. Athanasiadis and David G. Wagner. Veronese sections and interlacing matrices of polynomials and formal power series. arXiv:2404.12989, 2024.
    .bib
    @article{AthanasiadisWagner2024x,
      author = {Christos A. Athanasiadis and David G. Wagner},
      title = {Veronese sections and interlacing matrices of polynomials and formal power series},
      journal = {arXiv e-prints},
      year = {2024},
      eprint = {2404.12989},
      doi = {10.48550/arXiv.2404.12989}
    }
    
  6. [Ben18]Ferenc Bencs. Christoffel-Darboux type identities for the independence polynomial. Combinatorics, Probability and Computing, 27(5):716–724, 2018.
    .bib
    @article{Bencs2014,
      author = {Ferenc Bencs},
      title = {Christoffel-{D}arboux type identities for the independence
        polynomial},
      year = {2018},
      journal = {Combinatorics, Probability and Computing},
      volume = {27},
      number = {5},
      pages = {716--724},
      doi = {10.1017/s0963548318000135},
      url = {https://doi.org/10.1017/s0963548318000135},
      eprint = {1409.2527}
    }
    
  7. [Br04]Petter Brändén. On operators on polynomials preserving real-rootedness and the Neggers-Stanley conjecture. Journal of Algebraic Combinatorics, 20(2):119–130, September 2004.
    .bib
    @article{Branden2004operators,
      title = {On Operators on Polynomials Preserving Real-Rootedness and the {N}eggers-{S}tanley Conjecture},
      volume = {20},
      ISSN = {0925-9899},
      url2 = {http://dx.doi.org/10.1023/B:JACO.0000047295.93525.df},
      DOI = {10.1023/b:jaco.0000047295.93525.df},
      number = {2},
      journal = {Journal of Algebraic Combinatorics},
      publisher = {Springer Science and Business Media LLC},
      author = {Petter Br{\"{a}}nd{\'{e}}n},
      year = {2004},
      month = sep,
      pages = {119–130}
    }
    
  8. [Br15]Petter Brändén. Unimodality, log-concavity, real-rootedness and beyond. Handbook of enumerative combinatorics:437–483, March 2015.
    .bib
    @incollection{Branden2015,
      doi = {10.1201/b18255-10},
      year = {2015},
      month = mar,
      publisher = {Chapman and Hall/{CRC}},
      pages = {437--483},
      author = {Petter Br{\"{a}}nd{\'{e}}n},
      title = {Unimodality, Log-Concavity, Real-Rootedness and Beyond},
      booktitle = {Handbook of Enumerative Combinatorics}
    }
    
  9. [BJ22]Petter Brändén and Katharina Jochemko. The Eulerian transformation. Transactions of the American Mathematical Society, 375(3):1917–1931, 2022.
    .bib
    @article{BrandenJochemko2022,
      author = {Petter Br{\"a}nd{\'e}n and Katharina Jochemko},
      title = {The {E}ulerian transformation},
      journal = {Transactions of the American Mathematical Society},
      volume = {375},
      number = {3},
      pages = {1917--1931},
      year = {2022},
      doi = {10.1090/tran/8539},
      eprint = {2103.00890},
      url = {https://arxiv.org/abs/2103.00890}
    }
    
  10. [BS18]Petter Brändén and Liam Solus. Symmetric decompositions and real-rootedness. International Mathematics Research Notices, 2018.
    .bib
    @article{BrandenSolus2018,
      author = {Petter Br{\"{a}}nd{\'{e}}n and Liam Solus},
      title = {Symmetric decompositions and real-rootedness},
      year = {2018},
      journal = {International Mathematics Research Notices},
      doi = {10.1093/imrn/rnz059},
      url = {https://doi.org/10.1093/imrn/rnz059},
      eprint = {1808.04141},
      archivePrefix = {arXiv}
    }
    
  11. [CS07]Maria Chudnovsky and Paul Seymour. The roots of the independence polynomial of a clawfree graph. Journal of Combinatorial Theory, Series B, 97(3):350–357, May 2007.
    .bib
    @article{ChudnovskySeymour2007,
      doi = {10.1016/j.jctb.2006.06.001},
      url2 = {https://doi.org/10.1016/j.jctb.2006.06.001},
      year = {2007},
      month = may,
      publisher = {Elsevier {BV}},
      volume = {97},
      number = {3},
      pages = {350--357},
      author = {Maria Chudnovsky and Paul Seymour},
      title = {The roots of the independence polynomial of a clawfree graph},
      journal = {Journal of Combinatorial Theory,  Series B}
    }
    
  12. [Ded92]Jean-Pierre Dedieu. Obreschkoff’s theorem revisited: What convex sets are contained in the set of hyperbolic polynomials?. Journal of Pure and Applied Algebra, 81(3):269–278, 1992.
    .bib
    @article{Dedieu1992,
      author = {Dedieu, Jean-Pierre},
      title = {Obreschkoff's theorem revisited: What convex sets are contained in the set of hyperbolic polynomials?},
      journal = {Journal of Pure and Applied Algebra},
      volume = {81},
      number = {3},
      pages = {269--278},
      year = {1992}
    }
    
  13. [FZZ25]Hanqian Fang, Candice X. T. Zhang and James J. Y. Zhao. Analytic properties arising from the Baxter numbers. arXiv:2505.05873v1, 2025.
    .bib
    @article{FangZhangZhao2025x,
      author = {Hanqian Fang and Candice X. T. Zhang and James J. Y. Zhao},
      title = {Analytic properties arising from the {B}axter numbers},
      year = {2025},
      eprint = {2505.05873v1},
      url = {https://arxiv.org/abs/2505.05873v1},
      journal = {arXiv e-prints}
    }
    
  14. [Fav35]Jean Favard. Sur les polynômes de tchebicheff. C. R. Acad. Sci. Paris, 200:2052–2053, 1935.
    .bib
    @article{Favard1935,
      author = {Jean Favard},
      title = {Sur les polyn{\^o}mes de Tchebicheff},
      journal = {C. R. Acad. Sci. Paris},
      volume = {200},
      year = {1935},
      pages = {2052--2053}
    }
    
  15. [Fis05]Steve Fisk. A very short proof of Cauchy’s interlace theorem for eigenvalues of Hermitian matrices. arXiv:math/0502408, 2005.
    .bib
    @article{Fisk2005x,
      author = {Steve Fisk},
      title = {A very short proof of {C}auchy's interlace theorem for eigenvalues of {H}ermitian matrices},
      year = {2005},
      eprint = {math/0502408},
      url = {https://arxiv.org/abs/math/0502408},
      journal = {arXiv e-prints},
      journalref = {Amer. Math. Monthly, Vol 112, No. 2, Feb. 2005, page 118}
    }
    
  16. [Fis06]Steve Fisk. Polynomials, roots, and interlacing. arXiv:math/0612833, 2006.
    .bib
    @article{Fisk2006x,
      Author = {Steve Fisk},
      Title = {Polynomials, roots, and interlacing},
      Year = {2006},
      Eprint = {arXiv:math/0612833},
      journal = {arXiv e-prints}
    }
    
  17. [God17]C. D. Godsil. Algebraic combinatorics. Routledge, 2017.
    .bib
    @book{Godsil2017,
      doi = {10.1201/9781315137131},
      url2 = {https://doi.org/10.1201/9781315137131},
      year = {2017},
      month = oct,
      publisher = {Routledge},
      author = {C.D. Godsil},
      title = {Algebraic Combinatorics}
    }
    
  18. [GLR24]Hélène Guérin, Lucile Laulin and Kilian Raschel. Elephant polynomials. arXiv:2401.04959v2, 2024.
    .bib
    @article{GuerinLaulinRaschel2024x,
      author = {H{\'e}l{\`e}ne Gu{\'e}rin and Lucile Laulin and Kilian Raschel},
      title = {Elephant polynomials},
      year = {2024},
      eprint = {2401.04959v2},
      url = {https://arxiv.org/abs/2401.04959v2},
      journal = {arXiv e-prints}
    }
    
  19. [HS22]Max Hlavacek and Liam Solus. Subdivisions of shellable complexes. Journal of Combinatorial Theory, Series A, 186:105553, 2022.
    .bib
    @article{HlavacekSolus2022,
      author = {Max Hlavacek and Liam Solus},
      title = {Subdivisions of shellable complexes},
      year = {2022},
      journal = {Journal of Combinatorial Theory, Series A},
      volume = {186},
      pages = {105553},
      doi = {10.1016/j.jcta.2021.105553},
      eprint = {2003.07328v2}
    }
    
  20. [Hol03]Olga Holtz. Hermite–Biehler, Routh–Hurwitz, and total positivity. Linear Algebra and its Applications, 372:105–110, 2003.
    .bib
    @article{Holtz2003,
      author = {Holtz, Olga},
      title = {Hermite--{B}iehler, {R}outh--{H}urwitz, and total positivity},
      year = {2003},
      journal = {Linear Algebra and its Applications},
      volume = {372},
      pages = {105--110},
      publisher = {Elsevier BV},
      doi = {10.1016/s0024-3795(03)00501-9},
      url = {http://dx.doi.org/10.1016/S0024-3795(03)00501-9},
      issn = {0024-3795}
    }
    
  21. [Joc22]Katharina Jochemko. Symmetric decompositions and the Veronese construction. International Mathematics Research Notices, 2022(15):11427–11447, 2022.
    .bib
    @article{Jochemko2022SymmetricDecompositions,
      author = {Katharina Jochemko},
      title = {Symmetric decompositions and the {V}eronese construction},
      year = {2022},
      journal = {International Mathematics Research Notices},
      volume = {2022},
      number = {15},
      pages = {11427--11447},
      doi = {10.1093/imrn/rnab031},
      eprint = {2004.05423v3}
    }
    
  22. [Kos10]Vladimir P. Kostov. Interlacing properties and the Schur–Szegő composition. Functional Analysis and Other Mathematics, 3(1):65–74, 2010.
    .bib
    @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}
    }
    
  23. [Lau22]Lap Chi Lau. Method of interlacing polynomials. 2022. Lecture notes for CS 860
    .bib
    @misc{Lau2022InterlacingFamilies,
      author = {Lau, Lap Chi},
      title = {Method of interlacing polynomials},
      year = {2022},
      note = {Lecture notes for CS 860},
      url = {https://cs.uwaterloo.ca/~lapchi/cs860-2022/notes/12-interlacing-family.pdf}
    }
    
  24. [LR24]Jonathan Leake and Nick Ryder. Compatibility of real-rooted polynomials with mixed signs. arXiv:2407.16226v2, 2024.
    .bib
    @article{LeakeRyder2024x,
      Author  = {Jonathan Leake and Nick Ryder},
      Title   = {Compatibility of Real-Rooted Polynomials with Mixed Signs},
      Year    = {2024},
      journal = {arXiv e-prints},
      Eprint  = {2407.16226v2}
    }
    
  25. [Lev64]B. Ja. Levin. Distribution of zeros of entire functions. Translations of mathematical monographs. American Mathematical Society, 1964.
    .bib
    @book{Levin1964,
      author = {Levin, B. Ja.},
      title = {Distribution of Zeros of Entire Functions},
      series = {Translations of Mathematical Monographs},
      volume = {5},
      publisher = {American Mathematical Society},
      address = {Providence, RI},
      year = {1964},
      doi = {10.1090/mmono/005},
      url2 = {https://doi.org/10.1090/mmono/005}
    }
    
  26. [Liu12]Lily L. Liu. Polynomials with real zeros and compatible sequences. The Electronic Journal of Combinatorics, 19(3):P33, 2012.
    .bib
    @article{Liu2012CompatibleSequences,
      author = {Liu, Lily L.},
      title = {Polynomials with real zeros and compatible sequences},
      year = {2012},
      journal = {The Electronic Journal of Combinatorics},
      volume = {19},
      number = {3},
      pages = {P33},
      doi = {10.37236/2674}
    }
    
  27. [LW07]Lily L. Liu and Yi Wang. A unified approach to polynomial sequences with only real zeros. Advances in Applied Mathematics, 38(4):542–560, May 2007.
    .bib
    @article{LiuWang2007,
      doi = {10.1016/j.aam.2006.02.003},
      url2 = {https://doi.org/10.1016/j.aam.2006.02.003},
      year = {2007},
      month = may,
      publisher = {Elsevier {BV}},
      volume = {38},
      number = {4},
      pages = {542--560},
      author = {Lily L. Liu and Yi Wang},
      title = {A unified approach to polynomial sequences with only real zeros},
      journal = {Advances in Applied Mathematics}
    }
    
  28. [LY25]Lily Li Liu and Xue Yan. Some interlacing properties related to the Eulerian and derangement polynomials. Advances in Applied Mathematics, 162:102776, 2025.
    .bib
    @article{LiuYan2025InterlacingEulerianDerangement,
      author = {Liu, Lily Li and Yan, Xue},
      title = {Some interlacing properties related to the {E}ulerian and derangement polynomials},
      journal = {Advances in Applied Mathematics},
      volume = {162},
      pages = {102776},
      year = {2025},
      doi = {10.1016/j.aam.2024.102776},
      url2 = {https://doi.org/10.1016/j.aam.2024.102776}
    }
    
  29. [MW08]Shi-Mei Ma and Yi Wang. $q$-Eulerian polynomials and polynomials with only real zeros. The Electronic Journal of Combinatorics, 15(1):Research Paper 17, January 2008.
    .bib
    @article{MaWang2008,
      doi = {10.37236/741},
      url2 = {https://doi.org/10.37236/741},
      year = {2008},
      month = jan,
      publisher = {The Electronic Journal of Combinatorics},
      volume = {15},
      number = {1},
      author = {Shi-Mei Ma and Yi Wang},
      title = {$q$-{E}ulerian Polynomials and Polynomials with Only Real Zeros},
      pages = {Research Paper 17},
      journal = {The Electronic Journal of Combinatorics}
    }
    
  30. [MW26]Jianxi Mao and Lijie Wang. The Narayana transformation. arXiv:2607.01572, 2026.
    .bib
    @article{MaoWang2026x,
      author = {Jianxi Mao and Lijie Wang},
      title = {The {N}arayana transformation},
      year = {2026},
      eprint = {2607.01572},
      url = {https://arxiv.org/abs/2607.01572},
      journal = {arXiv e-prints}
    }
    
  31. [MSS15]Adam Marcus, Daniel A. Spielman and Nikhil Srivastava. Interlacing families I: Bipartite Ramanujan graphs of all degrees. Annals of Mathematics, 182(1):307–325, 2015.
    .bib
    @article{MarcusSpielmanSrivastava2015I,
      author = {Adam Marcus and Daniel A. Spielman and Nikhil Srivastava},
      title = {Interlacing families {I}: Bipartite {R}amanujan graphs of all degrees},
      year = {2015},
      journal = {Annals of Mathematics},
      volume = {182},
      number = {1},
      pages = {307--325},
      doi = {10.4007/annals.2015.182.1.7},
      url = {https://doi.org/10.4007/annals.2015.182.1.7},
      eprint = {1304.4132}
    }
    
  32. [MSS15]Adam Marcus, Daniel A. Spielman and Nikhil Srivastava. Interlacing families II: Mixed characteristic polynomials and the Kadison–Singer problem. Annals of Mathematics, 182(1):327–350, 2015.
    .bib
    @article{MarcusSpielmanSrivastava2015II,
      author = {Adam Marcus and Daniel A. Spielman and Nikhil Srivastava},
      title = {Interlacing families {II}: Mixed characteristic polynomials and the
        {K}adison--{S}inger problem},
      year = {2015},
      journal = {Annals of Mathematics},
      volume = {182},
      number = {1},
      pages = {327--350},
      doi = {10.4007/annals.2015.182.1.8},
      url = {https://doi.org/10.4007/annals.2015.182.1.8},
      eprint = {1306.3969}
    }
    
  33. [Obr63]Nikola Obreschkoff. Verteilung und berechnung der nullstellen reeller polynome. Hochschulbücher für mathematik. VEB Deutscher Verlag der Wissenschaften, 1963.
    .bib
    @book{Obreschkoff1963,
      author = {Obreschkoff, Nikola},
      title = {Verteilung und Berechnung der Nullstellen reeller Polynome},
      series = {Hochschulb{\"u}cher f{\"u}r Mathematik},
      volume = {55},
      publisher = {VEB Deutscher Verlag der Wissenschaften},
      address = {Berlin},
      year = {1963}
    }
    
  34. [Sch14]Issai Schur. Zwei sätze über algebraische Gleichungen mit lauter reellen Wurzeln. Journal für die reine und angewandte Mathematik, 144(2):75–88, 1914.
    .bib
    @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}
    }
    
  35. [Wag92]David G Wagner. Total positivity of Hadamard products. Journal of Mathematical Analysis and Applications, 163(2):459–483, January 1992.
    .bib
    @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}
    }
    
  36. [WY05]Yi Wang and Yeong-Nan Yeh. Polynomials with real zeros and Pólya frequency sequences. Journal of Combinatorial Theory, Series A, 109(1):63–74, 2005.
    .bib
    @article{WangYeh2005,
      author = {Yi Wang and Yeong-Nan Yeh},
      title = {Polynomials with real zeros and {P}{\'o}lya frequency sequences},
      year = {2005},
      journal = {Journal of Combinatorial Theory, Series A},
      volume = {109},
      number = {1},
      pages = {63--74},
      doi = {10.1016/j.jcta.2004.07.008},
      url = {https://doi.org/10.1016/j.jcta.2004.07.008},
      eprint = {math/0611825}
    }
    
  37. [Zha26]Zhanhe Zhang. Strict interlacing of the descent polynomials of separable permutations. 2026. SSRN preprint 7510941, posted September 25, 2026
    .bib
    @unpublished{Zhang2026SeparableInterlacing,
      author = {Zhang, Zhanhe},
      title = {Strict Interlacing of the Descent Polynomials of Separable Permutations},
      date = {2026-09-10},
      year = {2026},
      note = {SSRN preprint 7510941, posted September 25, 2026},
      doi = {10.2139/ssrn.7510941},
      url = {https://ssrn.com/abstract=7510941}
    }
    

I use cookies to detect website issues and track search terms.