---
title: 'Hexablock: A 2×2 μ-Synthesis Domain'
url: https://www.emergentmind.com/topics/hexablock
type: topic
---

# Hexablock: A 2×2 μ-Synthesis Domain

The **hexablock** is a domain \(\mathbb H\subset \mathbb C^4\) associated with a special case of the structured singular value, or \( \mu \)-synthesis, problem for \(2\times 2\) matrices. It is attached to the upper-triangular matrix structure and occupies, in four complex variables, the role that the symmetrized bidisc \(\mathbb G_2\), the tetrablock \(\mathbb E\), and the pentablock \(\mathbb P\) play for earlier \(2\times 2\) \( \mu \)-synthesis models. Subsequent work established several equivalent realizations of \(\mathbb H\), determined its distinguished boundary and automorphism group, developed an operator theory of \(\mathbb H\)-contractions, and proved that the underlying structured singular value rigidly determines the upper-triangular structure from which the hexablock arises [2506.15149] [2507.16176] [2507.14589] [2603.26312].

## 1. Matrix-theoretic origin

The ambient framework is the structured singular value \(\mu_E\) attached to a linear subspace \(E\subseteq M_n(\mathbb C)\), defined by
\[
\mu_E(A)=\frac{1}{\inf\{\|X\|:X\in E,\ \det(I_n-AX)=0\}},
\]
with the convention \(\mu_E(A)=0\) if no such \(X\in E\) exists. For the hexablock, the relevant structure is the \(3\)-dimensional subspace
\[
E_{\mathrm{hexa}}=
\left\{
\begin{pmatrix}
z_1 & w\\
0 & z_2
\end{pmatrix}
:
z_1,z_2,w\in\mathbb C
\right\}
\subset M_2(\mathbb C),
\]
the space of all upper-triangular \(2\times 2\) matrices containing the scalar matrices [2603.26312].

The associated coordinate map is
\[
\pi:M_2(\mathbb C)\to \mathbb C^4,\qquad
\pi(A)=(a_{21},a_{11},a_{22},\det A),
\]
for \(A=(a_{ij})\). In this setting, the raw \(\mu\)-theoretic object is the image of the \(\mu_{\mathrm{hexa}}\)-unit ball under \(\pi\). The domain called the hexablock is the analytically useful domain extracted from that image; it encodes the \(2\times 2\) upper-triangular \(\mu\)-synthesis problem in several complex variables rather than directly in matrix space [2506.15149].

This construction places the hexablock in a standard \(2\times 2\) hierarchy. The scalar, diagonal, upper-triangular-with-equal-diagonal, and full upper-triangular structures lead respectively to \(\mathbb G_2\), \(\mathbb E\), \(\mathbb P\), and \(\mathbb H\) [2603.26312].

## 2. Equivalent realizations and Hartogs structure

A central realization of the hexablock uses the tetrablock
\[
\mathbb E=
\left\{
(x_1,x_2,x_3)\in\mathbb C^3:
1-x_1 z_1-x_2 z_2+x_3 z_1 z_2\neq 0
\ \forall\, |z_1|\le 1,\ |z_2|\le 1
\right\}.
\]
For \((a,x_1,x_2,x_3)\in \mathbb C\times \mathbb E\), define
\[
\psi_{z_1,z_2}(a,x_1,x_2,x_3)=
\frac{
a\sqrt{(1-|z_1|^2)(1-|z_2|^2)}
}{
1-x_1 z_1-x_2 z_2+x_3 z_1 z_2
},
\qquad z_1,z_2\in\mathbb D.
\]
Then
\[
\mathbb H=
\left\{
(a,x_1,x_2,x_3)\in \mathbb C\times\mathbb E:
\sup_{z_1,z_2\in\mathbb D}
\left|\psi_{z_1,z_2}(a,x_1,x_2,x_3)\right|<1
\right\}.
\]
Equivalently, if
\[
K_*(x)=
\sup_{z_1,z_2\in\mathbb D}
\frac{\sqrt{(1-|z_1|^2)(1-|z_2|^2)}}
{|1-x_1z_1-x_2z_2+x_3z_1z_2|},
\]
then
\[
(a,x_1,x_2,x_3)\in\mathbb H
\iff
(x_1,x_2,x_3)\in\mathbb E
\text{ and }
|a|\,K_*(x_1,x_2,x_3)<1.
\]
These descriptions exhibit \(\mathbb H\) as a Hartogs-type domain over \(\mathbb E\) [2506.15149].

A further formulation makes the fibration explicit:
\[
\mathbb H=\{(a,x)\in\mathbb C\times\mathbb E: |a|^2<e^{-u(x)}\},
\]
where \(u:\mathbb E\to\mathbb R^+\) is smooth and determined by maximizing points \((z_1(x),z_2(x))\in\mathbb D^2\). Thus, for each \(x\in\mathbb E\), the admissible \(a\)-values form a disc whose radius depends on \(x\). In this sense the hexablock is a one-dimensional Hartogs fiber over the tetrablock [2507.16176].

An important structural point is that the coordinate image of the operator-norm ball and the coordinate image of the \(\mu_{\mathrm{hexa}}\)-ball do not coincide. The normed hexablock is strictly contained in the \(\mu\)-hexablock, and the \(\mu\)-hexablock is connected but not open. The domain \(\mathbb H\) is recovered as
\[
\mathbb H=\operatorname{int}\bigl(\overline{\mathbb H_\mu}\bigr)
=\operatorname{int}\bigl(\widehat{\overline{\mathbb H_N}}\bigr),
\]
and also satisfies
\[
\mathbb H=\mathbb H_\mu\cup(\{0\}\times \mathbb E).
\]
This rules out the common simplification that the hexablock is merely the direct image of a single matrix ball under \(\pi\) [2506.15149] [2507.14589].

## 3. Geometric properties and boundary structure

The hexablock is a bounded domain in \(\mathbb C^4\) with a mixture of convexity and nonconvexity properties. It is connected, polynomially convex, linearly convex, and \((1,1,1,2)\)-quasi-balanced. At the same time, it is neither starlike about the origin nor circled, and it is not convex. Its boundary is also non-\(\mathcal C^1\) [2506.15149] [2507.16176].

The distinguished boundary has a particularly explicit form:
\[
b\mathbb H=
\left\{
(a,x_1,x_2,x_3)\in\mathbb C^4:
|a|^2+|x_1|^2=1,\ (x_1,x_2,x_3)\in b\mathbb E
\right\}.
\]
Equivalently,
\[
b\mathbb H=\pi(\mathcal U(2)),
\]
and more explicitly
\[
b\mathbb H=
\left\{
(-e^{i\theta}z,\ w,\ e^{i\theta}\overline w,\ e^{i\theta}):
|z|^2+|w|^2=1,\ 0\le \theta\le 2\pi
\right\}.
\]
It follows that \(b\mathbb H\) is homeomorphic to \(\mathbb T\times \partial\mathbb B_2\), or equivalently to \(\mathbb T\times \mathbb S^3\) [2506.15149].

A finer analysis decomposes the topological boundary as
\[
\partial\mathbb H=\partial_1\mathbb H\cup\partial_2\mathbb H\cup\partial_3\mathbb H.
\]
Here
\[
\partial_1\mathbb H=
\{(a,x)\in\mathbb C\times\mathbb E: |a|^2=e^{-u(x)}\}
\]
has codimension \(1\), every point is smooth, and this stratum contains no \(2\)-dimensional analytic discs but is foliated by \(1\)-dimensional analytic discs. The second codimension-\(1\) piece,
\[
\partial_2\mathbb H=
\{(a,x)\in\mathbb C\times\partial\mathbb E:
|a|^2<|x_1x_2-x_3|,\ |x_3|\neq 1\},
\]
admits a foliation by \(2\)-dimensional analytic discs. The residual piece \(\partial_3\mathbb H\) has topological codimension \(2\). These distinct analytic-disc structures are central to later rigidity results for proper self-maps [2507.16176].

## 4. Relation to the symmetrized bidisc, tetrablock, and pentablock

The hexablock belongs to the family of domains attached to \(2\times 2\) \( \mu \)-synthesis data. In the formulation emphasized in the rigidity literature, the scalar matrices yield the symmetrized bidisc \(\mathbb G_2\), the diagonal matrices yield the tetrablock \(\mathbb E\), upper-triangular matrices with equal diagonal entries yield the pentablock \(\mathbb P\), and all upper-triangular \(2\times 2\) matrices yield the hexablock \(\mathbb H\) [2603.26312].

Within this hierarchy, \(\mathbb H\) contains \(\mathbb G_2\), \(\mathbb E\), and \(\mathbb P\) as analytic retracts. The embeddings and holomorphic retractions show that the smaller domains sit inside the hexablock with holomorphic left inverses. This is one reason the hexablock serves as a unifying framework: it packages in a single \(4\)-dimensional domain several lower-dimensional \(2\times 2\) \(\mu\)-synthesis geometries [2506.15149].

The relation to nearby \(4\)-dimensional domains is subtler. The domain
\[
\mathbb F=
\left\{
(a_{11},a_{22},\det A,a_{12}+a_{21})\in\mathbb C^4:
A\in M_2(\mathbb C),\ |A|<1
\right\}
\]
shares strong formal similarities with \(\mathbb H\): both lie in \(\mathbb C^4\), both are tied to the tetrablock, and both interact with \(\mathbb G_2\), \(\mathbb E\), and \(\mathbb P\). Nonetheless, they are not biholomorphic. The decisive obstruction is that their distinguished boundaries are not homeomorphic; \(\partial_S\mathbb F\) differs topologically from the distinguished boundary of the hexablock [2603.01483].

This comparison is significant because it shows that close analogies at the level of matrix coordinates and \( \mu \)-synthesis interpretation do not collapse the geometry of these domains into a single biholomorphic type.

## 5. Automorphisms and proper holomorphic self-maps

A substantial subgroup of \(\operatorname{Aut}(\mathbb H)\) was constructed by Biswas–Pal–Tomar:
\[
G(\mathbb H)=
\left\{
T_{\nu,\chi,\omega},\ T_{\nu,\chi,F,\omega}:
\nu,\chi\in\operatorname{Aut}(\mathbb D),\ \omega\in\mathbb T
\right\}.
\]
These automorphisms are built from automorphisms of the tetrablock together with a unimodular rotation in the fiber variable \(a\). Their explicit formulas involve Young’s tetrablock automorphisms \(\tau_{\nu,\chi}\) and \(\tau_{\nu,\chi,F}\) and a fractional-linear factor in the first coordinate [2506.15149] [2507.16176].

The principal rigidity theorem for mapping theory states that every proper holomorphic self-map of \(\mathbb H\) is an automorphism. Consequently,
\[
\operatorname{Aut}(\mathbb H)=G(\mathbb H).
\]
This settles the conjecture
\[
G(\mathbb H)=\operatorname{Aut}(\mathbb H)
\]
posed by Biswas–Pal–Tomar [2507.16176].

The proof combines several structural inputs. Because \(\mathbb H\) is \((1,1,1,2)\)-quasi-balanced, a proper self-map extends holomorphically to a neighborhood of \(\overline{\mathbb H}\) and carries \(\partial\mathbb H\) into itself. The analytic-disc foliations of \(\partial_1\mathbb H\) and \(\partial_2\mathbb H\) force the induced map on the base variables to be a proper holomorphic self-map of the tetrablock, hence an automorphism. After normalization, the remaining fiber dynamics are reduced on the subdomain
\[
\Omega=\{(a,\lambda,\lambda,\lambda^2)\in\mathbb H\}
=\{(a,\lambda,\lambda,\lambda^2)\in\mathbb C^4: |a|^2+|\lambda|^2<1\},
\]
which is essentially the unit ball in \(\mathbb C^2\). Alexander’s theorem then yields the final rigidity step in the fiber variable [2507.16176].

## 6. Operator theory and rigidity of the underlying matrix structure

The operator-theoretic counterpart of the domain studies commuting operator quadruples
\[
(A,X_1,X_2,X_3)
\]
for which \(\overline{\mathbb H}\) is a spectral set. Such a tuple is called an \(\mathbb H\)-contraction. Because \(\overline{\mathbb H}\) is polynomially convex, the spectral-set condition is equivalent to the polynomial von Neumann inequality
\[
\|p(A,X_1,X_2,X_3)\|
\le
\|p\|_{\infty,\overline{\mathbb H}}
\qquad
\text{for all }p\in\mathbb C[z_0,z_1,z_2,z_3].
\]
The resulting theory includes characterizations of \(\mathbb H\)-unitaries and \(\mathbb H\)-isometries, canonical decompositions, a Wold-type decomposition, and two dilation theorems [2507.14589].

One of the basic structural characterizations is that
\[
(A,X_1,X_2,X_3)\text{ is an }\mathbb H\text{-unitary}
\]
if and only if
\[
(X_1,X_2,X_3)\text{ is an }E\text{-unitary}
\quad\text{and}\quad
A^*A=I-X_1^*X_1.
\]
The theory is tightly linked to pre-existing operator models: if \((A,X_1,X_2,X_3)\) is an \(\mathbb H\)-contraction, then \((A,X_1)\) and \((A,X_2)\) are \(\mathbb B_2\)-contractions, \((X_1,X_2,X_3)\) is an \(E\)-contraction, and \((A,X_1+X_2,X_3)\) is a \(P\)-contraction. Conversely, the classes associated with \(\mathbb B_2\), \(\mathbb E\), \(\mathbb P\), and \(\Gamma\) embed into the hexablock framework through explicit coordinate identifications [2507.14589].

Independent of the operator theory, the matrix structure underlying the hexablock is itself rigid. If \(E\subseteq M_2(\mathbb C)\) is a linear subspace containing the scalar matrices and
\[
\mu_E=\mu_{\mathrm{hexa}}
\quad\text{on }M_2(\mathbb C),
\]
then necessarily
\[
E=E_{\mathrm{hexa}}.
\]
The proof first excludes any matrix with nonzero \((2,1)\)-entry by testing against the rank-one matrix \(A=e_1e_2^t\), and then rules out the only possible proper intermediate \(2\)-dimensional subspaces inside \(E_{\mathrm{hexa}}\). In particular, neither the diagonal subspace nor the penta-subspace
\[
E_{\mathrm{penta}}=
\left\{
\begin{pmatrix}
z & w\\
0 & z
\end{pmatrix}
:
z,w\in\mathbb C
\right\}
\]
has the same structured singular value as \(E_{\mathrm{hexa}}\) [2603.26312].

This rigidity theorem identifies a defining feature of the hexablock construction: the function \(\mu_E\) associated with the domain does not merely arise from an upper-triangular structure; among scalar-containing subspaces of \(M_2(\mathbb C)\), it determines that structure uniquely.

Source: https://www.emergentmind.com/topics/hexablock