---
title: McCullough–Wanderley Conjecture Overview
url: https://www.emergentmind.com/topics/mccullough-and-wanderley-conjecture
type: topic
---

# McCullough–Wanderley Conjecture Overview

Searching arXiv for the cited papers and closely related work on the McCullough–Wanderley conjecture, Markoff surfaces, and complete Pick kernels.
The McCullough–Wanderley conjecture is a family of conjectural classification statements for generating pairs of \(\mathrm{SL}_2(\mathbb F_q)\), their \(T_2\)-equivalence classes, and the Markoff-equivalence classes of triples in \(\mathbb F_q^3\) satisfying a Markoff-type equation. In the formulations emphasized in recent work, the central principle is that the relevant equivalence class should be determined by a commutator invariant on the group-theoretic side and by the Markoff parameter on the geometric side. Through the trace map and Fricke’s identity, the conjecture becomes an orbit-transitivity problem for Vieta involutions on finite-field Markoff surfaces, and this places it at the intersection of combinatorial group theory, arithmetic dynamics, and strong approximation [2508.21671, 2510.07577].

## 1. Conjectural formulations

McCullough and Wanderley formulated several closely related conjectures for \(\mathrm{SL}_2(\mathbb F_q)\). The invariant on the group-theoretic side is the **Higman invariant**, namely the extended conjugacy class of the commutator,
\[
\mathrm{cl}_G([g_1,g_2])\cup \mathrm{cl}_G([g_2,g_1]).
\]
For \(\mathrm{SL}_2(\mathbb F_q)\), this is essentially the same as the trace of the commutator, since matrices in that extended conjugacy class share the same trace [2510.07577].

| Conjecture | Object classified | Classifying datum |
|---|---|---|
| **Classification conjecture** | Nielsen classes of generating pairs in \(\mathrm{SL}_2(\mathbb F_q)\) | Higman invariant |
| **\(T\)-classification conjecture** | \(T_2\)-systems in \(\mathrm{SL}_2(\mathbb F_q)\) | Trace invariant |
| **\(Q\)-classification conjecture** | Markoff classes of essential triples | \(\kappa=x^2+y^2+z^2-xyz\) |

In the formulation stated for prime fields \(F=\mathbb F_p\), the \(Q\)-classification conjecture says: let \((A,B)\) be a generating pair of \(\mathrm{SL}(F)\). Then two generating pairs are Nielsen equivalent if and only if their commutators lie in the same extended conjugacy class of \(\mathrm{SL}(F)\). The same paper states this more concretely as
\[
\text{There is a single orbit of generating pairs at level } k\neq -2.
\]
Moreover, if \(p\equiv 3\pmod 4\), there is also a single orbit of generating pairs at level \(-2\); if \(p\equiv 1\pmod 4\), there are two orbits at level \(-2\). Here the **level** is
\[
k=\operatorname{tr}([A,B]).
\]
This is the form in which the conjecture is tied directly to orbit structure on finite-field Markoff surfaces [2508.21671].

A complementary Markoff-side formulation is that the Markoff class of an essential triple \((x,y,z)\) is uniquely determined by
\[
\kappa=x^2+y^2+z^2-xyz.
\]
The significance of the 2025 structural result is that the \(Q\)-classification conjecture implies the classification conjecture and the \(T\)-classification conjecture, so the Markoff-side orbit problem controls the other two formulations [2510.07577].

## 2. Trace triples, Nielsen moves, and Vieta dynamics

The geometric framework is built from the trace map
\[
\operatorname{Tr}(A,B)=\big(\operatorname{tr}(A),\operatorname{tr}(B),\operatorname{tr}(AB)\big).
\]
Fricke’s trace identity gives
\[
\operatorname{tr}(A)^2+\operatorname{tr}(B)^2+\operatorname{tr}(AB)^2-\operatorname{tr}(A)\operatorname{tr}(B)\operatorname{tr}(AB)-2=\operatorname{tr}([A,B]).
\]
Accordingly, if \(\operatorname{tr}([A,B])=k\), then \(\operatorname{Tr}(A,B)\) lies on
\[
M_k:\quad x^2+y^2+z^2-xyz-2=k.
\]
In the alternative parameterization used elsewhere, setting
\[
\kappa=\operatorname{tr}[A,B]+2
\]
places the trace triple on
\[
x^2+y^2+z^2=xyz+\kappa.
\]
This is the basic bridge from generating pairs in \(\mathrm{SL}_2(\mathbb F_q)\) to affine cubic surfaces over finite fields [2508.21671, 2510.07577].

On pairs \((A,B)\), the papers use Nielsen moves such as
\[
r(A,B)=(B,A),\qquad s(A,B)=(A^{-1},AB),\qquad t(A,B)=(A^{-1},B),
\]
as well as the standard elementary Nielsen generators. On triples \((x,y,z)\), these induce the Vieta involutions
\[
(x,y,z)\mapsto (yz-x,y,z),\qquad (x,y,z)\mapsto (x,xz-y,z),\qquad (x,y,z)\mapsto (x,y,xy-z),
\]
together with coordinate permutations. The cited work states that these correspond exactly under the trace map:
\[
\operatorname{Tr}\circ \mathcal N=\mathcal N\circ \operatorname{Tr}
\]
for every Nielsen move \(\mathcal N\). Thus Nielsen dynamics on generating pairs and Markoff dynamics on trace triples are two realizations of the same action [2508.21671].

This correspondence also motivates the notion of an **essential** triple: a triple in \(\mathbb F_q^3\) is essential if it comes from a generating pair \((A,B)\) of \(\mathrm{SL}_2(\mathbb F_q)\). The conjectural picture is then that, after removing a short list of exceptional small orbits, all essential triples with fixed parameter should lie in one Markoff orbit [2510.07577].

## 3. Exceptional orbits and finite-orbit rigidity

A central theme in recent work is that the failure of transitivity is concentrated in explicitly classifiable exceptional orbits. One classification theorem partitions the corresponding non-projective subgroup types in \(\mathrm{PSL}(F)\) into affine, projective, and exceptional finite triangle-group types
\[
D_n,\quad A_4,\quad S_4,\quad A_5,
\]
with admissibility conditions
\[
D_n \text{ admissible} \iff n\mid p\pm 1,
\]
\[
A_4 \text{ always admissible},
\]
\[
S_4 \text{ admissible} \iff p\equiv \pm 1\pmod 8,
\]
\[
A_5 \text{ admissible} \iff p\equiv \pm 1\pmod 5 \text{ or } p=5.
\]
The same theorem gives a complete list of exceptional orbits on \(M_k(F)\) [2508.21671].

| Type | Representative and level | Size |
|---|---|---|
| **Dihedral \(D_n\)** | \((0,0,0)\) at level \(-2\); \((t,0,0)\) at level \(t^2-2\) | \(1\); \(6\) |
| **Tetrahedral \(A_4\)** | \((1,1,0)\) at level \(0\) | \(16\) |
| **Octahedral \(S_4\)** | \((\sqrt2,1,0)\) at level \(1\) | \(36\) |
| **Icosahedral \(A_5\)** | \(\left(\frac{1+\sqrt5}{2},\frac{1-\sqrt5}{2},0\right)\) at level \(1\); \(\left(\frac{1+\sqrt5}{2},1,0\right)\) at level \(\frac{1+\sqrt5}{2}\); \(\left(\frac{1-\sqrt5}{2},1,0\right)\) at level \(\frac{1-\sqrt5}{2}\) | \(72\); \(40\); \(40\) |

These exceptional finite-field orbits are stated to agree exactly with the finite orbits of the complex Markoff-type equations \(M_k\) found by Dubrovin and Mazzocco. The cited interpretation is that the small orbits are rigid and are not accidental finite-field phenomena [2508.21671].

A broader strong-approximation result for the cubic surfaces
\[
\mathcal S_{A,B,C,D}:\quad X^2+Y^2+Z^2=XYZ+AX+BY+CZ+D
\]
organizes the exceptional set differently. For nondegenerate parameters, the only small orbits that need to be removed are those arising from finite \(\Gamma\)-orbits over \(\mathbb C\), classified by Lisovyy–Tykhyy as Type I singleton orbits, Type II of size \(2\), Type III of size \(3\), Type IV of size \(4\), plus \(45\) exceptional orbits [2603.04096].

## 4. Strong approximation and the orbit-transitivity principle

The McCullough–Wanderley conjecture is closely tied to strong approximation for Markoff surfaces. One formulation states that every solution of \(M_k\) over \(\mathbb F_p\) descends from a solution over \(\mathbb Z\), and that the action of Vieta involutions on \(M_k(\mathbb F_p)\) is essentially transitive. In the finite-field orbit language, the strong approximation conjecture is that for each \(k\neq 2\) there should be a unique large orbit \(\mathcal C_k\) such that
\[
M_k(F)=\mathcal C_k\sqcup \mathcal E_k,
\]
where \(\mathcal E_k\) is the union of exceptional orbits at level \(k\). Equivalently, every non-exceptional point lies in one “cage” orbit \(\mathcal C_k\) [2508.21671].

A major result is that, when \(p\equiv 3\pmod 4\), the \(Q\)-classification conjecture is equivalent to strong approximation. The mechanism described in the literature is that strong approximation gives one generating orbit on each non-singular level, but a priori a tower over a triple could split into two Nielsen orbits; when \(p\equiv 3\pmod 4\), the relevant towers over the cage are connected. The same discussion records an important caveat at the special level \(k=-2\): if \(p\equiv 3\pmod 4\), there is one generating orbit, whereas if \(p\equiv 1\pmod 4\), there are two generating orbits. The equivalence is stated for all levels \(k\) with \(2-k\) not a square, and for \(k=-2\) under the appropriate congruence condition [2508.21671].

This perspective clarifies a common source of ambiguity. The conjecture is not merely a statement about isolated Nielsen classes; it is a global transitivity statement for the Vieta action after the explicit exceptional locus has been removed. Bourgain–Gamburd–Sarnak are cited as showing that there is a very large orbit \(\mathcal C_k\) with at most \(p^\epsilon\) points outside it, which reduces the problem to identifying the small exceptional orbits [2508.21671].

## 5. Proven cases and density-one results

Recent work has supplied two different kinds of progress. The first is a structural implication: \(Q\)-classification implies the classification conjecture and the \(T\)-classification conjecture. The second is an arithmetic verification for a large set of prime fields. One theorem states that if \(p\) is prime and
\[
27720\nmid (p^2-1),
\]
then for every \(\kappa\in\mathbb F_p\setminus\{4\}\), there is exactly one orbit of solutions to
\[
x^2+y^2+z^2=xyz+\kappa
\]
under the Vieta involutions, except for a short explicit list of exceptional orbits. Under the same arithmetic hypothesis there is a single orbit of generating pairs with fixed commutator trace, except in the classical \(\kappa=0\), \(p\equiv 1\pmod 4\) exception where there are two orbits. The paper describes this as proving the McCullough–Wanderley conjectures for prime fields \(\mathbb F_p\) whenever \(27720\nmid(p^2-1)\), covering a density
\[
1-\frac{2^5}{27720}\approx 0.999
\]
of primes [2510.07577].

A different density-one theorem concerns the more general cubic surface
\[
X^2+Y^2+Z^2=XYZ+AX+BY+CZ+D.
\]
For nondegenerate integer parameters, and for a density one set of primes \(p\) depending on the parameters, \(\Gamma\) acts transitively on the complement of the small orbits coming from finite orbits over \(\mathbb C\). In the special family
\[
X^2+Y^2+Z^2=XYZ+k,\qquad k\neq 4,
\]
the paper states that its results “very nearly prove the \(Q\)-classification conjecture of McCullough and Wanderley for density 1 of all primes,” and that via Martin this also “very nearly proves their Classification and \(T\)-Classification conjectures for density 1 of all primes.” The qualification is explicit: the density-one prime set depends on the parameter, and the proof requires a lower bound of the form
\[
\frac12\log_{20+k}(p)^{1/3}>1000,
\]
or the general parameter version with \(20+2|A|+2|B|+2|C|+|D|\). This parameter dependence prevents a full uniform proof in the strongest form [2603.04096].

The same source also records Martin’s transfer theorem: if the solutions to
\[
X^2+Y^2+Z^2=XYZ+k
\]
over \(\mathbb F_p\), with \(k\neq 4\), which are not reductions of finite orbits over \(\mathbb C\) form a single orbit under \(\Gamma\), then the Higman invariant fully classifies Nielsen equivalence classes of generating pairs of \(\mathrm{SL}_2(\mathbb F_p)\). This is the precise bridge from orbit transitivity on the Markoff-type surface to the Classification Conjecture [2603.04096].

## 6. Obstructions, degenerate parameters, and terminological distinctions

The modern finite-field theory emphasizes that transitivity can genuinely fail for degenerate parameters. For the general family
\[
X^2+Y^2+Z^2=XYZ+AX+BY+CZ+D,
\]
the degeneracy condition is
\[
\text{degenerate} \iff \text{equivalent to a quadruple with } A=B \text{ and } 4D+A^2=8C+16.
\]
The equivalence here is by permuting \(A,B,C\) and changing signs of two of them in the permitted way. For nondegenerate parameters, the density-one giant-orbit theorem applies; for degenerate parameters, there must be at least \(2\) large orbits, and in some cases \(4\) large orbits, so transitivity fails [2603.04096].

In the one-parameter family
\[
X^2+Y^2+Z^2=XYZ+k,
\]
the degeneracy condition forces \(k=4\), so the only degenerate surface in that subfamily is the Cayley cubic. This isolates the main obstruction to a uniform Markoff-plus-\(k\) transitivity statement. In the generalized cluster-algebra family
\[
x_1^2+x_2^2+x_3^2+a_1x_2x_3+a_2x_1x_3+a_3x_1x_2=(3+a_1+a_2+a_3)x_1x_2x_3,
\]
the same paper states that
\[
(A,B,C,D)\text{ is degenerate mod }p \iff \exists i,\ a_i^2\equiv 4\pmod p,
\]
and over the integers
\[
(A,B,C,D)\text{ is degenerate} \iff \exists i,\ a_i^2=4.
\]
This matches the de Courcy-Ireland–Litman–Mizuno obstruction [2603.04096].

The name “McCullough” also appears in an unrelated operator-theoretic context: the McCullough–Quiggin characterization of complete Pick kernels. A 2019 note gives “a short and simple proof of necessity in the McCullough–Quiggin characterization of complete Pick kernels” and concerns positive semidefinite kernels, reproducing kernel Hilbert spaces, and the positivity of
\[
F_z(x,y)=1-\frac{k(x,z)k(z,y)}{k(z,z)\,k(x,y)}.
\]
That line of work is distinct from the McCullough–Wanderley conjectures on Markoff triples and Nielsen equivalence, even though both literatures involve McCullough’s earlier contributions [1912.13068].

Taken together, the current picture is technically sharp. The conjecture is best viewed as an orbit-classification principle linking generating pairs of \(\mathrm{SL}_2(\mathbb F_q)\), Vieta dynamics on Markoff surfaces, and strong approximation. Its exceptional locus is now highly explicit, its dependence on subgroup structure and finite complex orbits is understood in detail, and large classes of prime fields are covered by rigorous theorems. What remains open is the strongest uniform form: a fully parameter-independent orbit-transitivity statement for all nondegenerate levels and all sufficiently large primes.

Source: https://www.emergentmind.com/topics/mccullough-and-wanderley-conjecture