---
title: Markoff-like Surfaces
url: https://www.emergentmind.com/topics/markoff-like-surfaces
type: topic
---

# Markoff-like Surfaces

Markoff-like surfaces are families of affine cubic surfaces, and in some extensions K3 surfaces, organized around the Markoff polynomial and its close relatives. In the cited literature, the core examples are the affine cubic surfaces \(U_a: x^2+y^2+z^2-xyz=a\), the generalized Markoff surfaces \(X_m: x^2+y^2+z^2-3xyz=m\), and several character-variety and Wehler-surface analogues with the same coordinate symmetries and Vieta-type transformations [2408.06846] [2603.23306] [2302.11515]. Their study sits at the intersection of Diophantine geometry, arithmetic dynamics, finite-field expansion, Brauer–Manin theory, and low-dimensional character varieties [2110.11030] [2202.07142].

## 1. Defining equations and geometric models

In current usage, “Markoff-like” does not refer to a single equation but to a family of closely related level sets and deformations. A central normalization is the Markoff polynomial
\[
M(x,y,z)=x^2+y^2+z^2-xyz,
\]
whose level sets
\[
U_a:\quad x^2+y^2+z^2-xyz=a
\]
are called the family of affine Markoff type cubic surfaces [2408.06846]. Another recurring normalization is
\[
x^2+y^2+z^2-3xyz=m,
\]
viewed as a family of generalized Markoff surfaces \(X_m\) whose positive integral points are Markoff \(m\)-triples [2603.23306]. The one-parameter family
\[
V_k:\quad x_1^2+x_2^2+x_3^2-x_1x_2x_3=k
\]
is the framework in which integral points, “class numbers,” and almost-all Hasse principle results are developed [1706.06712].

| Family | Equation | Context |
|---|---|---|
| Affine Markoff type cubic surfaces | \(x^2+y^2+z^2-xyz=a\) | Integral Hasse principle and density results |
| Generalized Markoff surfaces | \(x^2+y^2+z^2-3xyz=m\) | Markoff \(m\)-triples, trees, and branches |
| Four-holed-sphere relative character varieties | \(x^2+y^2+z^2+xyz=ax+by+cz+d\) | Markoff-type affine cubic surfaces |
| Markoff-type K3 surfaces | \(ax^2y^2z^2+b(x^2y^2+x^2z^2+y^2z^2)+cxyz+d(x^2+y^2+z^2)+e=0\) | Wehler K3 analogues |

The geometry depends sharply on the normalization. For
\[
U_m:\quad x^2+y^2+z^2-xyz=m,
\]
the projective closure
\[
X_m:\quad t(x^2+y^2+z^2)-xyz=mt^3
\]
is smooth if and only if \(m(m-4)\neq 0\), and \(U_m\) is the complement of the hyperplane section \(H=L_1\cup L_2\cup L_3\), where the \(L_i\) are three lines at infinity [1808.01584]. In the \(\mathrm{SL}_2\)-trace normalization
\[
\mathcal M_k:\quad x^2+y^2+z^2-xyz=k,
\]
the surfaces are nonsingular affine cubic surfaces for \(k\neq 4\), and the literature quoted there identifies them as nonsingular log K3’s [2110.11030]. The four-holed-sphere family
\[
x^2+y^2+z^2+xyz=ax+by+cz+d
\]
is likewise treated as an affine cubic surface obtained from a smooth cubic surface by removing three coplanar lines [2202.07142].

This multiplicity of models is structural rather than terminological. Some papers work with cubic surfaces in \(\mathbb A^3\), some with compactifications in \(\mathbb P^3\), and some with \((2,2,2)\)-surfaces in \((\mathbf P^1)^3\); the common thread is the persistence of Vieta-type involutions, strong coordinate symmetry, and arithmetic problems on integral or finite-field points.

## 2. Vieta involutions, trees, and orbit structures

The defining formal feature of a Markoff-like surface is that the equation is quadratic in each variable separately, so one can replace one root by the other. For
\[
x^2+y^2+z^2-xyz=k,
\]
this yields the familiar Vieta involutions
\[
(x_1,x_2,x_3)\mapsto (x_1,x_2,x_1x_2-x_3),
\]
together with the analogous involutions in the other coordinates; permutations and double sign changes enlarge the symmetry group [2110.11030] [1706.06712]. For generalized Markoff \(m\)-triples on
\[
x^2+y^2+z^2=3xyz+m,
\]
the corresponding transformations are
\[
\nu_1(a,b,c)=(b,c,3bc-a),\quad \nu_2(a,b,c)=(a,c,3ac-b),\quad \nu_3(a,b,c)=\operatorname{ord}(3ab-c,a,b),
\]
and these preserve the Markoff parameter \(m\) [2603.23306].

In the integral theory, these involutions organize points into trees, branches, and fundamental domains. For \(m>0\) in the generalized \(3xyz\)-normalization, a minimal Markoff \(m\)-triple is defined by \(c\ge 3ab\), and the number of distinct \(m\)-trees equals the number of minimal Markoff \(m\)-triples [2603.23306]. In the classical one-parameter family \(V_k\), the Markoff morphisms act with finitely many orbits on \(V_k(\mathbb Z)\) for every \(k\neq 4\), while the Cayley cubic \(k=4\) is exceptional and has infinitely many inequivalent \(\Gamma\)-orbits [1706.06712].

Over finite fields, the same transformations become graph dynamics. For the classical surface
\[
x_1^2+x_2^2+x_3^2-3x_1x_2x_3=0,
\]
fixing one coordinate \(x_j=a\) cuts out a conic \(C_j(a)\), and the composition of a transposition with a Vieta involution acts on that conic by the matrix
\[
\operatorname{rot}(3x_1)=
\begin{pmatrix}
0&1\\
-1&3x_1
\end{pmatrix},
\]
with hyperbolic, elliptic, and parabolic cases distinguished by the quadratic character of \(x^2-4\) [1607.01530]. In the three-parameter deformation
\[
x_1^2+x_2^2+x_3^2+a_1x_2x_3+a_2x_1x_3+a_3x_1x_2=(3+a_1+a_2+a_3)x_1x_2x_3,
\]
the Vieta involutions become
\[
m_i:\ x_i\mapsto -x_i + sx_{i-1}x_{i+1} - a_{i+1}x_{i-1}-a_{i-1}x_{i+1},
\qquad s=3+a_1+a_2+a_3,
\]
and the proof of orbit-divisibility by \(p\) uses angle functions \(\Delta_i\) satisfying
\[
\sum_i \Delta_i(x)=s,\qquad \Delta_i(x)+\Delta_i(m_i x)=s
\]
on appropriate domains [2509.02187].

This suggests that “Markoff-like” denotes not only a shape of equation but also a specific dynamical package: quadratic-in-one-variable geometry, involutive mutations, and orbit decompositions that can be studied by conic fibrations, trees, or finite graphs.

## 3. Integral points, local-global principles, and the Brauer–Manin obstruction

For the affine Markoff type cubic surfaces
\[
U_a:\quad x^2+y^2+z^2-xyz=a,
\]
the local solubility criterion is completely explicit:
\[
\mathcal U_a(\mathbf A_{\mathbb Z})\neq \emptyset
\iff
a\not\equiv 3 \pmod 4
\quad\text{and}\quad
a\not\equiv \pm 3 \pmod 9,
\]
and the locally soluble parameters have natural density \(7/12\) [2408.06846]. In the closely related family
\[
V_k:\quad x_1^2+x_2^2+x_3^2-x_1x_2x_3=k,
\]
it is proved that for almost all admissible \(k\) the Hasse principle for integral points holds, while there are infinitely many \(k\) for which it fails [1706.06712]. Mishra sharpened the upper bound on the exceptional set to
\[
\#\mathcal E(A)\ll_\varepsilon \frac{A}{(\log A)^{2-\varepsilon}},
\]
and deduced density-\(1\) integral Hasse principle results in sparse prime-shifted subfamilies \(U_{p+c}\) [2408.06846].

The Brauer-theoretic structure is unusually explicit for the cubic family
\[
U_m:\quad x^2+y^2+z^2-xyz=m.
\]
Writing \(d=m-4\), one has concrete algebraic Brauer classes such as
\[
(x-2,d),\qquad (y-2,d),\qquad (z-2,d),
\]
and the algebraic Brauer group \(\operatorname{Br}_1(U_m)/\operatorname{Br}_0(U_m)\) is computed case by case in terms of the square classes of \(m\), \(m-4\), and \(m(m-4)\); the transcendental quotient is described by a Kummer-type condition involving
\[
\frac{\sqrt{m-4}-\sqrt m}{2}
\]
[1808.01584]. For the related compactification and affine open, the paper on integral Hasse principle and strong approximation proves that only the squareclasses
\[
\{\pm 1,2,3,5\}\subset \mathbb Q^\*/\mathbb Q^{\*2}
\]
can support an integral Brauer–Manin obstruction, gives
\[
\gg \frac{B^{1/2}}{(\log B)^{1/2}}
\]
examples with a Brauer–Manin obstruction, and also
\[
\gg \frac{B^{1/2}}{\log B}
\]
examples with \(\mathcal U_m(\mathbf A_{\mathbb Z})^{\operatorname{Br}}\neq \varnothing\) but \(\mathcal U_m(\mathbb Z)=\varnothing\) [1807.10223].

A recurring misconception is that Brauer–Manin should account for all arithmetic failures in these families. The cited results show otherwise. For \(U_m\), strong approximation for integral points fails away from every finite set of places, and for \(m\neq 0,4\) the Brauer group does not control strong approximation [1808.01584]. In the four-holed-sphere family
\[
x^2+y^2+z^2+xyz=ax+by+cz+d,
\]
the generic algebraic Brauer group is \(\mathbb Z/2\), explicit corestricted quaternion classes are written down, and there are both positive-proportion strong-approximation failures explained by Brauer–Manin and explicit Hasse failures not explained by the algebraic Brauer group [2202.07142].

## 4. Finite-field, \(p\)-adic, and spectral dynamics

For the classical Markoff surface
\[
x_1^2+x_2^2+x_3^2-3x_1x_2x_3=0,
\]
the finite-field strong approximation conjecture is that for every prime \(p\),
\[
X(\mathbb F_p)=\{(0,0,0)\}\sqcup X^\*(p),
\]
with \(X^\*(p)\) a single orbit under the group generated by permutations and Vieta involutions [1607.01530]. What is proved unconditionally is already very strong: for every \(\varepsilon>0\) and \(p\) large there is a giant orbit \(\mathcal C(p)\subset X^\*(p)\) with
\[
|X^\*(p)\setminus \mathcal C(p)|\le p^\varepsilon,
\]
and the number of exceptional primes \(p\le T\) for which full transitivity fails is at most \(T^\varepsilon\) [1607.01530].

Computational evidence sharpens this picture. For every prime \(p\le 3000\), the nonzero mod-\(p\) Markoff graph is connected, confirming the Bourgain–Gamburd–Sarnak conjecture in that range [1812.07275]. The same paper reports that for \(p\equiv 3\pmod 4\), the second adjacency eigenvalue appears to approach \(2\sqrt2\), suggesting asymptotically Ramanujan behavior, whereas for \(p\equiv 1\pmod 4\) the data suggest a weaker limiting gap near \(2.875\ldots\); in both residue classes, the bulk spectrum matches the Kesten–McKay law [1812.07275].

Several recent works quantify the dynamics further. “Bounding Lifts of Markoff Triples mod \(p\)” derives explicit upper bounds for the size of integral lifts of mod-\(p\) points by analyzing path growth in the Markoff graphs, including a bound
\[
\operatorname{size}(\tilde{\mathbf x})<(3\epsilon)^{96(2p+1)^4},
\qquad
\epsilon=\frac{3+\sqrt5}{2},
\]
under a large-order hypothesis [2311.11468]. On the \(p\)-adic side, “Residual Transitivity implies Minimality for Markoff Surfaces over \(p\)-adic Integers” proves that if \(p>3\) and either \(D\equiv 0\pmod{p^2}\) or \(\left(\frac{D-4}{p}\right)=1\), then transitivity of \(\operatorname{Aut}(X_D^\ast)\) on \(X_D^\ast(\mathbb F_p)\) implies minimality on \(X_D^\ast(\mathbb Z_p)\), using \(p\)-adic analytic flows [2502.18976].

Beyond transitivity, Markoff-like dynamics exhibit other local-global phenomena. For the normalized Markoff surface
\[
M:\quad x^2+y^2+z^2=xyz,
\]
the compositions \(\phi_i\phi_j\) of two reflections are strongly residually periodic: \((M,\phi_i\phi_j,\{(0,0,0)\})\) is \(\mathrm{SRP}(3)\), and the periodic points modulo almost every prime come from the periodic conics \(z=\pm 1\), which have no \(\mathbf Q\)-rational points [1504.07099]. In the off-diagonal deformation
\[
x_1^2+x_2^2+x_3^2+a_1x_2x_3+a_2x_1x_3+a_3x_1x_2=(3+a_1+a_2+a_3)x_1x_2x_3,
\]
one has a different finite-field rigidity theorem: if \(p\ge 5\), \(s=3+a_1+a_2+a_3\neq 0\), and \(a_i^2\neq 4\) for all \(i\), then every nontrivial orbit has size divisible by \(p\); on the Cayley-cubic exceptional locus there are parameter families with at least two or four nontrivial orbits [2509.02187].

## 5. Character varieties, recurrence constraints, and arithmetic slices

A major source of Markoff-like surfaces is trace geometry. For the commutator equation
\[
[X,Y]=XYX^{-1}Y^{-1}=Z
\quad\text{in}\quad \mathrm{SL}_2,
\]
if
\[
x_1=\operatorname{Tr}X,\qquad x_2=\operatorname{Tr}Y,\qquad x_3=\operatorname{Tr}(XY),
\]
then the Fricke identity gives
\[
x_1^2+x_2^2+x_3^2-x_1x_2x_3=\operatorname{Tr}([X,Y])+2.
\]
Thus the cubic surfaces
\[
\mathcal M_k:\quad x^2+y^2+z^2-xyz=k
\]
arise as \(\mathrm{SL}_2\)-trace surfaces attached to commutator equations and to the once-punctured-torus character variety [2110.11030]. The four-holed-sphere relative character variety produces the different but closely related family
\[
x^2+y^2+z^2+xyz=ax+by+cz+d,
\]
which is explicitly treated as a Markoff-type cubic surface in Brauer–Manin theory [2202.07142].

A different kind of specialization comes from recurrence sequences. On the generalized surfaces
\[
x^2+y^2+z^2=3xyz+m,
\]
the paper “Branches of Markoff \(m\)-triples with two \(k\)-Fibonacci components” studies integral points with at least two coordinates in a fixed \(k\)-Fibonacci sequence [2603.23306]. It proves that every non-minimal such triple has the form
\[
\left(\alpha_{k,r},\,F_k(N-r),\,F_k(N+r)\right),
\qquad
\alpha_{k,r}=\frac{F_k(4r)}{3F_k(2r)}=\frac{L_k(2r)}{3},
\]
where \(r\) is odd, \(3\nmid k\), \(N>3r\), and \(N\) is odd when \(k\in\{1,2\}\) [2603.23306]. It also proves that every infinite path of Markoff \(m\)-triples with at least two \(k\)-Fibonacci components is contained in a principal \((2,k)\)-Fibonacci branch, and that for fixed \(r\) these branches are distributed among exactly \(2r\) distinct trees [2603.23306].

This suggests that recurrence constraints can cut out highly rigid arithmetic subgraphs inside a Markoff-like surface: rather than producing sporadic integral points, they can force an explicit branch structure governed simultaneously by Lucas-sequence identities and Vieta dynamics.

## 6. K3 analogues and higher-dimensional Markoff-type geometry

The Markoff paradigm extends beyond cubic surfaces. A Wehler surface is a hypersurface of multidegree \((2,2,2)\) in
\[
\mathbf P^1\times \mathbf P^1\times \mathbf P^1,
\]
and when smooth it is a K3 surface. A Markoff-type K3 surface is a Wehler surface invariant under permutations of \(x,y,z\) and double sign changes, hence given in affine coordinates by
\[
ax^2y^2z^2+b(x^2y^2+x^2z^2+y^2z^2)+cxyz+d(x^2+y^2+z^2)+e=0,
\]
with nondegeneracy conditions
\[
c\neq 0,\qquad be\neq d^2,\qquad ad\neq b^2
\]
[2302.11515]. In one explicit family,
\[
x^2+y^2+z^2+4(x^2y^2+y^2z^2+z^2x^2)-16x^2y^2z^2-k=0,
\]
arithmetic hypotheses imply that the compactification \(W\) is a smooth K3 surface with
\[
\rho(W_{\overline{\mathbf Q}})=18,
\]
while for the affine open \(U=W\setminus\{rst=0\}\),
\[
\operatorname{Br}_1W/\operatorname{Br}_0W\cong (\mathbf Z/2\mathbf Z)^3,
\qquad
\operatorname{Br}_1U/\operatorname{Br}_0U\cong (\mathbf Z/2\mathbf Z)^4,
\]
with explicit algebraic classes such as
\[
A_1=(4x^2+1,\,-2(4k+1)),
\qquad
A_2=(4y^2+1,\,-2(4k+1))
\]
[2302.11515]. The same work constructs infinite families of integral Hasse principle failures and explicit Brauer-obstructed failures of strong approximation for three one-parameter MK3 families [2302.11515].

A later paper isolates a different MK3 family
\[
W_k:\quad (x^2-36)(y^2-36)(z^2-36)-m_0(xyz+C_0)^2-k=0,
\qquad
m_0=-468,\ C_0=-4330,
\]
and proves that it contains surfaces with Zariski-dense rational points but no integral points, the failure of the integral Hasse principle being explained by an algebraic Brauer–Manin obstruction [2504.10992]. This separates the rational and integral theories in a particularly sharp way.

Finite-field dynamics on K3 analogues can also depart from the cubic picture. For the tri-involutive K3 family
\[
\mathcal W_k:\quad x^2+y^2+z^2+x^2y^2z^2=kxyz,
\]
O’Dorney explains a phenomenon observed numerically by Fuchs, Litman, Silverman, and Tran: when \(q\equiv 1\pmod 8\), the points of \(W_4(\mathbb F_q)\) do not form a single large orbit under the natural symmetry group \(\Gamma\), but admit a partition into two disjoint \(\Gamma\)-invariant subsets, each of size
\[
\frac12 q^2+O(q^{3/2}),
\]
and the mechanism is an explicit double cover of \(W_4\) [2209.10436]. A plausible implication is that, once one passes from cubic surfaces to K3 surfaces, the Markoff combination of Vieta dynamics and local-global arithmetic survives, but hidden covering structures can become a first-order obstruction to single-orbit behavior.

Source: https://www.emergentmind.com/topics/markoff-like-surfaces