Overview

  • When is a composition irreducible?
  • When are all iterates irreducible?
  • How large are the Galois groups of iterates?
  • How do critical orbits affect these questions?

Composition and factorization

For polynomials f,g∈K[x]f,g\in K[x], their composition is (f∘g)(x)=f(g(x))(f\circ g)(x)=f(g(x)). A decomposition F=f∘gF=f\circ g, with both degrees greater than one, is different from a factorization F=uvF=uv. For example, x4−2=(x2−2)∘x2x^4-2=(x^2-2)\circ x^2 is decomposable but irreducible over Q\mathbb Q by Eisenstein’s criterion. Conversely, x2−1x^2-1 is reducible but cannot be decomposed into two polynomials of smaller positive degree.

Ritt’s 1922 work describes decompositions over C\mathbb C. Its connection with Galois theory is the correspondence between right composition factors, intermediate function fields, and blocks in the action on the roots of F(x)−tF(x)-t. The article on polynomial decomposition develops this correspondence.

References 1

Irreducibility of a composition

Let F∈K[x]F\in K[x] be irreducible, let g∈K[x]g\in K[x] have degree at least one, and let α\alpha be a root of FF. Capelli’s lemma reduces the composition problem to one extension field:

F(g(x)) is irreducible over K⟺g(x)−α is irreducible over K(α).F(g(x))\text{ is irreducible over }K\quad\Longleftrightarrow\quad g(x)-\alpha\text{ is irreducible over }K(\alpha).

Thus a degree-mdmd problem can become a degree-dd problem over a field of degree mm. For quadratics, the latter is a square test; norms can then give a sufficient condition in the ground field. See the proof and examples.

References 1

Irreducibility of iterates

Write f∘0(x)=xf^{\circ0}(x)=x and f∘(n+1)=f∘f∘nf^{\circ(n+1)}=f\circ f^{\circ n}. A polynomial is stable over KK if every f∘nf^{\circ n}, n≥1n\ge1, is irreducible over KK. It is eventually stable if the numbers of irreducible factors of these iterates, counted with multiplicity, are bounded.

Stability concerns one factor at every depth. Eventual stability allows several factors, but bounds their number. Both depend on the field and, for the equations f∘n(x)−a=0f^{\circ n}(x)-a=0, on the target aa. The stability article treats Capelli’s lemma, critical-orbit tests, local criteria, and open cases.

References 1

Galois groups on a tree

Fix a∈Ka\in K and let d=deg⁡f≥2d=\deg f\ge2. Assume that every f∘n(x)−af^{\circ n}(x)-a is separable. Place its roots at level nn, using a separate copy at each level, and join β\beta to f(β)f(\beta). This gives a regular rooted tree with dd children at each vertex. Galois automorphisms preserve these edges.

The resulting group lies in the automorphism group of the tree. Through level nn, this is the nn-fold iterated wreath product of SdS_d, of order

∣Aut⁡(Td,n)∣=(d!)1+d+⋯+dn−1.|\operatorname{Aut}(T_{d,n})|=(d!)^{1+d+\cdots+d^{n-1}}.

Irreducibility at every level means transitivity on every level. Equality with the full tree group is a stronger condition. See arboreal Galois groups.

References 2

Example

For f=x2+1f=x^2+1, the adjusted critical values begin

−1,2,5,26,….-1,\quad 2,\quad 5,\quad 26,\ldots.

Stoll’s criterion expresses maximality through level nn as independence of the first nn values in Q×/Q×2\mathbb Q^\times/\mathbb Q^{\times2}. In particular, −1-1 and 22 are independent, so the second iterate has group C2≀C2C_2\wr C_2, the dihedral group of order 88. Stoll also proves maximality at every level for this polynomial; the calculation above illustrates the finite-level criterion.

References 1

Critical orbits

A polynomial is postcritically finite if every critical point has finite forward orbit. The polynomial x2−2x^2-2 has the critical orbit 0↦−2↦2↦20\mapsto-2\mapsto2\mapsto2. Its iterates are all irreducible over Q\mathbb Q, but their Galois groups are cyclic of order 2n2^n, far smaller than the full binary tree groups. The cyclotomic calculation proves both assertions.

Critical-orbit equations also define polynomials in the parameter. For x2+cx^2+c, periodic critical points give Gleason polynomials; strictly preperiodic critical points give Misiurewicz polynomials. Their irreducibility is a separate arithmetic problem.

References 2

Development of the subject

Odoni proved that a generic monic polynomial in characteristic zero has the full iterated wreath product at every finite level. Arithmetic specialization asks for one polynomial over a number field that retains this property at all levels simultaneously. Looper proved existence in every prime degree over Q\mathbb Q. Benedetto–Juul prove existence over a number field KK when dd is even, or when dd and d−2d-2 are not both squares in KK. In particular, all degrees are covered when [K:Q][K:\mathbb Q] is odd. Specter’s preprint gives every degree over every number field.

The base field matters: Dittmann–Kadets construct Hilbertian fields of characteristic zero over which no polynomial of degree at least two has finite-index arboreal image. This does not contradict the number-field results.

References 5

Current questions

The main open problems ask when finitely many iterates certify stability or maximality, and when the arboreal image has finite index.

Recent work includes explicit quadratic representations at strictly preperiodic targets, descriptions of postcritically finite groups, and Stoll’s November 2026 finite-level criteria. See Arboreal Galois groups.

References 3

Further reading

Silverman’s The Arithmetic of Dynamical Systems develops the background on critical points, reduction, heights, and dynatomic polynomials. Jones’s 2013 survey organizes the arboreal questions and their arithmetic applications. His 2007: An Arboreal Odyssey, posted in 2026, is an unchanged historical manuscript; its open questions describe the subject as it stood in 2007.

References 3

Questions

OpenComposition and iteration

Can one fixed iterate detect the full arboreal group?

Scope. f_a=x²+a with a∈Q∖{−1,0}; target point 0; Stoll, November 2026.

Is there a fixed NN such that maximality of Gal⁡(fa∘N/Q)\operatorname{Gal}(f_a^{\circ N}/\mathbb Q) implies maximality at every level, for every a∈Q∖{−1,0}a\in\mathbb Q\setminus\{-1,0\}? Hindes proposed N=4N=4; Stoll repeats this as Conjecture 8.1. The values a=2/3a=2/3 and −6/7-6/7 rule out N<4N<4. Stoll reports the implication from level 44 to level 2020 for all a=r/sa=r/s of height max⁡(∣r∣,s)≤106\max(|r|,s)\le10^6, in lowest terms with s>0s>0. The conjecture concerns every level and every rational parameter.

References 2
  • Hindes2015IterationCurves · §3, Remark 3.4 and following discussion, p. 26Asks whether the first failure of maximality can occur at arbitrarily large levels, and whether all first failures occur before level 5.Source
  • Stoll2026Arboreal · Conjecture 8.1, Proposition 8.2, and Remark 8.5; arXiv:2611.14014v1 (10 November 2026)Source
Discuss this question →
SettledComposition and iteration

Does every Hilbertian field admit a surjective polynomial arboreal representation of degree at least two?

Scope. Polynomials of degree d≥2 over Hilbertian fields of characteristic zero

No. Dittmann–Kadets give counterexample fields. These counterexamples are compatible with the positive existence results over number fields and with generic maximality at each fixed finite level.

References 1
Discuss this question →
OpenComposition and iteration

Do iterates of x²+1/c always have at most four irreducible factors?

Scope. c∈Z∖{0,−1}; factorization over Q; all iterate depths

Conjecture 1.7 predicts this uniform bound and gives a precise list of factor counts. The bound is attained in known cases.

References 1
Discuss this question →
OpenComposition and iteration

Does irreducibility of the second iterate imply stability for x²+1/c?

Scope. c∈Z∖{0,−1}; all irreducibility statements over Q

If f∘2f^{\circ2} is irreducible, must every f∘nf^{\circ n} be irreducible? This is Conjecture 1.8, distinct from the conjecture on maximal Galois groups.

References 1
Discuss this question →
OpenComposition and iteration

Does every quadratic polynomial with infinite critical orbit have finite-index arboreal image?

Scope. f=x²+c over a number field K; the forward orbit of 0 is infinite; target point 0

Jones’s conjecture predicts [Aut⁡(T∞):G∞(f,0)]<∞[\operatorname{Aut}(T_\infty):G_\infty(f,0)]<\infty when f=x2+cf=x^2+c over a number field has infinite critical orbit. It allows a proper group at early levels. This is the polynomial case of Conjecture 3.11; the 2026 quadratic preprint still cites the finite-index problem as conjectural.

References 2
Discuss this question →

Related topics

References

  1. Prime and composite polynomials

    Ritt, Joseph F. Transactions of the American Mathematical Society, 23(1), 51–66, 1922.

    Read the source
  2. Polynomials with Special Regard to Reducibility

    Schinzel, Andrzej. Cambridge University Press, Encyclopedia of Mathematics and its Applications, vol. 77, 2000.

    Read the source
  3. Eventually stable rational functions

    Jones, Rafe and Levy, Alon. International Journal of Number Theory, 13(9), 2299–2318, 2017.

    Read the source
  4. The Galois Theory of Iterates and Composites of Polynomials

    Odoni, R. W. K. Proceedings of the London Mathematical Society (3), 51(3), 385–414, 1985.

    Read the source
  5. Galois representations from pre-image trees: an arboreal survey

    Jones, Rafe. Publications mathématiques de Besançon. Algèbre et théorie des nombres, 107–136, 2013.

    Read the source
  6. Galois groups over Q of some iterated polynomials

    Stoll, Michael. Archiv der Mathematik, 59, 239–244, 1992.

    Read the source
  7. Arboreal Galois groups of postcritically finite quadratic polynomials: The strictly preperiodic case

    Benedetto, Robert L. and Ghioca, Dragos and Juul, Jamie and Tucker, Thomas J. Canadian Journal of Mathematics, 2026.

    Read the source
  8. Moduli Spaces of Quadratic Maps: Arithmetic and Geometry

    Ramadas, Rohini. International Mathematics Research Notices, 2024(15), 11364–11370, 2024.

    Read the source
  9. Dynamical Galois groups of trinomials and Odoni's conjecture

    Looper, Nicole R. Bulletin of the London Mathematical Society, 51(2), 278–292, 2019.

    Read the source
  10. Odoni's conjecture for number fields

    Benedetto, Robert L. and Juul, Jamie. Bulletin of the London Mathematical Society, 51(2), 237–250, 2019.

    Read the source
  11. Polynomials with Surjective Arboreal Galois Representations Exist in Every Degree

    Specter, Joel. arXiv preprint, 2018.

    Read the source
  12. Odoni's conjecture on arboreal Galois representations is false

    Dittmann, Philip and Kadets, Borys. Proceedings of the American Mathematical Society, 150(8), 3335–3343, 2022.

    Read the source
  13. Examples of Quadratic Polynomials over Q with Surjective Arboreal Galois Representations

    Henderson, Luck and Juul, Jamie and Lattin, Brenner and Mercado, Enrique and Schaefer, Mia. arXiv preprint, 2026.

    Read the source
  14. Arboreal Galois groups of postcritically finite quadratic polynomials: The periodic case

    Benedetto, Robert L. and Ghioca, Dragos and Juul, Jamie and Tucker, Thomas J. Journal of the London Mathematical Society, 112(2), e70257, 2025.

    Read the source
  15. When do all iterates of x²+a have maximal Galois group?

    Stoll, Michael. arXiv preprint, 2026.

    Read the source
  16. The Arithmetic of Dynamical Systems

    Silverman, Joseph H. Graduate Texts in Mathematics 241, Springer, vol. 241, x+511, 2007.

    Read the source
  17. 2007: An Arboreal Odyssey: A View of Arboreal Galois Representations and Applications, from Early in the Subject’s History

    Jones, Rafe. arXiv historical survey manuscript, 2026.

    Read the source
  18. The arithmetic of curves defined by iteration

    Hindes, Wade. Acta Arithmetica, 169(1), 1–27, 2015.

    Read the source
  19. Eventually stable quadratic polynomials over Q

    DeMark, David and Hindes, Wade and Jones, Rafe and Misplon, Moses and Stoll, Michael and Stoneman, Michael. New York Journal of Mathematics, 26, 526–561, 2020.

    Read the source