Sendov's conjecture

Celebrated

Let n2n \ge 2 be an integer and let

f(z)=(zr1)(zr2)(zrn),r1,,rnC,f(z)=(z-r_{1})(z-r_{2})\cdots (z-r_{n}), \qquad r_{1},\dots ,r_{n}\in \mathbb{C},

be a monic polynomial all of whose roots lie in the closed unit disk, i.e. rk1|r_{k}|\le 1 for every k{1,,n}k \in \{1,\dots ,n\}. Since degf=n11\deg f' = n-1 \ge 1, the derivative ff' has at least one zero in C\mathbb{C}.

Then for every j{1,,n}j \in \{1,\dots ,n\} there exists zCz \in \mathbb{C} with

f(z)=0andzrj1;f'(z)=0 \qquad \text{and} \qquad |z-r_{j}| \le 1;

that is, for each root rjr_{j} of ff the closed disk {zC:zrj1}\{z \in \mathbb{C} : |z-r_{j}| \le 1\} contains a critical point of ff.

Equivalent formulations 5

Other statements of this same problem, merged from separate entries. Each is equivalent to the statement above — proving any one settles them all.

  1. Sendov's conjecture

    In mathematics, Sendov's conjecture, sometimes also called Ilieff's conjecture, concerns the relationship between the locations of roots and critical points of a polynomial function of a complex variable. It is named after Blagovest Sendov.

    source: Wikipedia

  2. Sendov's conjecture on critical points of polynomials

    Let

    p(z)=i=1n(zλi),p(z)=nj=1n1(zμj),p(z)=\prod_{i=1}^n(z-\lambda_i),\qquad p'(z)=n\prod_{j=1}^{n-1}(z-\mu_j),

    where λi1|\lambda_i|\leq1 for all ii. Sendov's conjecture. For every root λi\lambda_i of pp, there is a root μj\mu_j of pp' such that

    λiμj1.|\lambda_i-\mu_j|\leq1.

    The source says this conjecture has been open since 1959 and also calls it Ilyeff's conjecture; it gives no resolution.

    source: K. M. R. Audenaert and F. Kittaneh, “Problems and Conjectures in Matrix and Operator Inequalities”, arXiv:1201.5232 (2012).

  3. Circulant-matrix formulation of Sendov's conjecture

    Let CMnC\in M_n be a circulant matrix with C1\lVert C\rVert\leq1, let λ\lambda be an eigenvalue of CC, and let CC' be the principal submatrix obtained by deleting the first row and first column. Circulant formulation of Sendov's conjecture. The matrix CλIC'-\lambda I has an eigenvalue in the closed unit disk. This is presented as a third formulation of Sendov's conjecture, whose original polynomial statement is described as open in the source.

    source: K. M. R. Audenaert and F. Kittaneh, “Problems and Conjectures in Matrix and Operator Inequalities”, arXiv:1201.5232 (2012).

  4. D-companion-matrix formulation of Sendov's conjecture

    Let λ1,,λn\lambda_1,\ldots,\lambda_n belong to the closed unit disk, let D=diag(λ1λn,,λn1λn)D=\operatorname{diag}(\lambda_1-\lambda_n,\ldots,\lambda_{n-1}-\lambda_n), and let II and JJ be respectively the identity matrix and the all-ones matrix of order n1n-1. D-companion formulation of Sendov's conjecture. The matrix

    D(IJn)D\left(I-\frac{J}{n}\right)

    has an eigenvalue in the closed unit disk. This is presented as a reformulation of Sendov's conjecture; the source gives no resolution beyond stating Sendov's problem as open.

    source: K. M. R. Audenaert and F. Kittaneh, “Problems and Conjectures in Matrix and Operator Inequalities”, arXiv:1201.5232 (2012).

  5. Sendov's conjecture in the language of expansivity

    Let P(x)=anxn+an1xn1++a1x+a0P(x)=a_nx^n+a_{n-1}x^{n-1}+\cdots+a_1x+a_0 be a polynomial of degree n2n\geq 2. Let {bi}i=1n\{b_i\}_{i=1}^{n} be the zeros of P(x)P(x) satisfying bi1|b_i|\leq 1, and let S=(anxn,an1xn1,,a1x,a0)\mathcal{S}=(a_nx^n,a_{n-1}x^{n-1},\ldots,a_1x,a_0) be a tuple representation of P(x)P(x). For each bib_i, let Se\mathcal{S}_e denote the identity element used in the expansivity construction. Sendov's conjecture. For each bib_i, there exists some SaZ[(γ1βγ)1(S)]\mathcal{S}_a\in\mathcal{Z}[(\gamma^{-1}\circ\beta\circ\gamma\circ\nabla)^1(\mathcal{S})] such that

    Idn+1(biSeSa)1.|\operatorname{Id}_{n+1}(b_i\mathcal{S}_e-\mathcal{S}_a)|\leq 1.

    This is the expansivity formulation of the classical assertion that every zero in the unit disk lies within unit distance of a zero of the derivative. The paper reports partial variants and known degree-specific and asymptotic results, but the conjecture itself is presented as the central problem.

    source: Theophilus Agama, “Expansivity theory and Sendov's conjecture”, arXiv:1907.12825 (2026).

Sources & referencesView supporting material

Primary source

Wikipedia

Additional references

  1. B. Sendov, in Research Problems in Function Theory (W. K. Hayman, ed.), Athlone Press (1967), problem 4.5.
  2. T. Tao, "Sendov's conjecture for sufficiently high degree polynomials," Acta Mathematica 229 (2022), 347-392.
  3. G. Schmeisser, "The conjectures of Sendov and Smale," in Approximation Theory, DARBA (2002), 353-369.
  4. Wikipedia, Sendov's conjecture, the article this problem comes from.

Progress summary

Refreshed
Solved

A machine-checked proof now appears to settle the conjecture in every degree, although the underlying argument has not yet been independently published.

Sendov’s conjecture asserts that every zero of a polynomial whose zeros lie in the unit disk has a zero of its derivative within distance one. The classical question was known only through partial results before the recent full-generality formalization.

Known results

  • Degree n8n\le 8: proved by Brown and Xiang, with Xiang’s result dated 1999.
  • Arbitrary degree with at most eight distinct zeros: proved by Brown.
  • Sufficiently large degree: proved by Tao; the bound is effective but not the exact threshold n=8n=8.
  • Further special cases include all-real zeros and zeros on the unit circle.

August 2026 Lean-checked proof

On August 12, 2026, Tao reported that Lech Mazur’s AI-assisted argument resolves the conjecture for all n2n\ge 2 and has been verified in Lean. The accompanying formalization and proof document state the exact full-generality theorem, so this is best treated as resolved, though not yet as a publication-ready human proof.

Current status (as of August 2026): Sendov’s conjecture is settled for every degree by a Lean-checked formalization, while a publication-ready human proof remains forthcoming.

Sources

Solutions 0

No solutions have been posted yet.