---
title: 'Ring²: Square Operations in Ring Theory'
url: https://www.emergentmind.com/topics/ring-2
type: topic
---

# Ring²: Square Operations in Ring Theory

Ring\(^2\) is not a single standardized mathematical object. Across algebra, coding theory, graph theory, and discrete geometry, it denotes several square-related constructions: the additive square \(S^2\) of the symmetric part of a ring with involution; commutative ring extensions \(R\subset S\) of lattice-theoretic length \(2\); classes of rings constrained by conditions on \(u^2\) for \(u\in U(R)\); the square matrix ring \(M_2(F_p)\); rings with two commuting square-zero generators; the graph square \(R^2\) of a ring graph; and two-dimensional orthogonal ring patterns on square-grid combinatorics [1210.3353] [1803.11297] [2602.14521] [2508.06689] [1409.7228] [1405.5981] [1907.11905] [1911.07095]. The unifying motif is that “square” acts on different layers of structure—elements, distinguished subsets, extensions, matrices, or geometric configurations—so the meaning of Ring\(^2\) is context-dependent rather than canonical.

## 1. Square-generation inside rings with involution

In the setting of an associative unital \(F\)-algebra \(R\) with involution \( * \), the most direct reading of Ring\(^2\) is the additive subgroup \(S^2\), where
\[
S=\{r\in R\mid r^*=r\},\qquad K=\{r\in R\mid r^*=-r\},
\]
and for subsets \(A,B\subseteq R\), \(AB\) denotes the additive subgroup generated by all products \(ab\) with \(a\in A\), \(b\in B\). Thus \(S^2=SS\) is not the literal set of single products; it is the set of all finite sums \(\sum_i a_ib_i\) with \(a_i,b_i\in S\). Because \(1\in Z(R)\subseteq S\), one always has \(S\subseteq S^2\subseteq S^3\subseteq\cdots\). Herstein’s question asks whether, for a simple ring with involution of the first kind, one has \(S^2=R\) under suitable hypotheses [1210.3353].

The strongest universal statement proved in this direction is that if \(R\) is simple, \(\operatorname{Char}(F)\neq 2\), the involution is of the first kind, and \(S\) is not commutative, then
\[
R=S^3.
\]
The paper also gives two criteria that characterize when the square already suffices. In the simple noncommutative-symmetric setting,
\[
\exists x,y\in S:\ xy-yx\neq 0,\ xSy\subseteq S^2 \iff S^2=R,
\]
and similarly
\[
\exists x,y\in S:\ xy+yx\neq 0,\ xKy\subseteq S^2 \iff S^2=R.
\]
These statements localize the obstruction to square-generation: if suitable commutator or anticommutator data can be forced into \(S^2\), simplicity upgrades that inclusion to equality. The same framework yields parallel product-generation results for \(K^6\), \(K^4\), \(K+KSK\), \(KS+K^2\), \(SKS\), and \(S^2K\). In the matrix cases considered explicitly—transpose involution on \(M_n(F)\) and symplectic involution on \(M_{2m}(F)\)—the criteria are verified directly, so the symmetric square does generate the full matrix ring [1210.3353].

## 2. Ring\(^2\) as an extension of length \(2\)

In commutative algebra, Ring\(^2\) arises naturally as an extension \(R\subset S\) whose lattice of intermediate rings has length \(2\). If
\[
[R,S]=\{\,T\mid R\subseteq T\subseteq S\,\},
\]
then \(\ell[R,S]=2\) means that there exists a chain \(R\subset T\subset S\), but no longer chain \(R\subset T_1\subset T_2\subset S\). The central characterization is that, for a non-minimal extension, the following are equivalent: every proper intermediate ring \(T\) makes \(R\subset T\) minimal; every proper intermediate ring \(T\) makes \(T\subset S\) minimal; and \(\ell[R,S]=2\). In this sense, Ring\(^2\) is exactly the height-\(2\) case in the lattice \([R,S]\) [1803.11297].

Such extensions satisfy a sharp dichotomy. Every extension of length \(2\) is either pointwise minimal or simple, and these two possibilities are mutually exclusive. In addition, every length-\(2\) extension is quasi-Prüfer, and the support satisfies
\[
|\operatorname{Supp}(S/R)|\le 2.
\]
The classification then splits according to support size, integrality, and the canonical decomposition
\[
R\subset {}^+_R S \subset {}^t_R S \subset \overline R \subset S.
\]
When \(|\operatorname{Supp}(S/R)|=2\), the lattice \([R,S]\) has either \(3\) or \(4\) elements, and the extension is simple. In the \(M\)-crucial integral case, the possibilities divide into \(t\)-closed, seminormal infra-integral, subintegral, and mixed cases. The \(t\)-closed local case reduces to the residual field extension \(R/M\subset S/M\). For a finite separable field extension \(k\subset L\), length \(2\) is characterized by the distinct principal subfields \(E_1,\dots,E_t\): one has \(\ell[k,L]=2\) exactly when \(t>1\) and
\[
E_\alpha\cap E_\beta = k \qquad (\alpha\ne\beta).
\]
A major corollary is that every simple ring extension of length \(2\) has FIP, whereas the co-pointwise minimal cases are exactly the source of possible non-FIP behavior [1803.11297].

## 3. Square conditions on units: \(2\)-\(\sqrt{J}U\) and \(2\)-UNJ rings

A different interpretation of Ring\(^2\) focuses on the squares of units. One recent class is that of \(2\)-\(\sqrt{J}U\) rings, defined by
\[
\forall u\in U(R),\qquad u^2\in 1+\sqrt{J(R)},
\]
where
\[
\sqrt{J(R)}=\{x\in R: x^m\in J(R)\text{ for some }m\ge 1\}.
\]
This condition is weaker than \(\sqrt{J}U\), \(2\)-\(UJ\), and \(2\)-\(UU\), but it still imposes strong structure. Every homomorphic image and every finite direct product of \(2\)-\(\sqrt{J}U\) rings is again \(2\)-\(\sqrt{J}U\); if \(I\subseteq J(R)\), then \(R/I\) is \(2\)-\(\sqrt{J}U\) iff \(R\) is \(2\)-\(\sqrt{J}U\); every corner ring \(eRe\) and every unit-closed subring of a \(2\)-\(\sqrt{J}U\) ring is again \(2\)-\(\sqrt{J}U\). The theory also gives precise classification results: a division ring is \(2\)-\(\sqrt{J}U\) iff it is \(\mathbb F_2\) or \(\mathbb F_3\); a local ring is \(2\)-\(\sqrt{J}U\) iff its residue field is \(\mathbb F_2\) or \(\mathbb F_3\); and a semisimple ring is \(2\)-\(\sqrt{J}U\) iff it is a finite product of copies of \(\mathbb F_2\) and \(\mathbb F_3\). By contrast,
\[
M_n(R)\text{ is never }2\text{-}\sqrt{J}U\text{ for }n>1.
\]
The paper also proves
\[
R\text{ is a }\sqrt{J}U\text{ ring} \iff 2\in J(R)\text{ and }R\text{ is a }2\text{-}\sqrt{J}U\text{ ring},
\]
thereby locating the new class exactly between older unit conditions and radical structure [2602.14521].

A closely related generalization is the class of \(2\)-UNJ rings, defined by
\[
\forall u \in U(R),\quad u^2 \in 1 + Nil(R) + J(R).
\]
This includes every \(2\)-\(UJ\), \(2\)-\(UU\), and UNJ ring, but the converses fail. The class is stable under finite direct products, several quotient constructions, upper triangular matrix constructions, trivial extensions, skew and truncated polynomial extensions, and certain Morita contexts. It is not stable under full matrix amplification:
\[
M_n(R)\ \text{is not 2-UNJ for nonzero }R\text{ and }n\ge 2.
\]
Its semisimple shadow is again extremely small: a division ring is \(2\)-UNJ iff it is \(\mathbb Z_2\) or \(\mathbb Z_3\), and a semilocal ring is \(2\)-UNJ exactly when
\[
R/J(R)\cong \bigoplus_i R_i,\qquad R_i\cong \mathbb Z_2\text{ or }\mathbb Z_3.
\]
In stronger structural settings, the square condition forces polynomial identities. For regular rings,
\[
R\text{ is a regular 2-UNJ ring} \iff R\text{ is a tripotent ring}.
\]
For semi-potent rings, the conditions “\(R\) is 2-UNJ”, “\(R\) is 2-UJ”, and “\(R/J(R)\) is tripotent” become equivalent [2508.06689].

## 4. Square matrix rings and square-zero generators in coding theory

In algebraic coding theory, Ring\(^2\) appears literally as the square matrix ring
\[
M_2(F_p),
\]
the full \(2\times 2\) matrix ring over the prime field. This ring is treated as a finite Frobenius ring, and its coding-theoretic structure is organized through a quadratic-field embedding
\[
\tau:F_{p^2}\longrightarrow M_2(F_p),
\qquad
\tau(a+b\omega)=
\begin{pmatrix}
a & b\\
b & a+(p-1)b
\end{pmatrix},
\]
where \(f(x)=x^2+x+(p-1)\) is irreducible over \(F_p\). The resulting subring \(\mathcal F_{p^2}=\tau(F_{p^2})\) yields the decomposition
\[
M_2(F_p)=\mathcal F_{p^2}+\mathbf v_p\mathcal F_{p^2},
\]
with \(\mathbf v_p^2=I\), and also a nilpotent presentation
\[
A_p:=M_2(F_p)=F_{p^2}+\mathbf u_pF_{p^2},
\qquad
\mathbf u_p^2=0.
\]
The paper derives the homogeneous weight on \(M_2(F_p)\), introduces the Bachoc weight
\[
w_B(A)=
\begin{cases}
0,&A=0,\\
1,&A\in GL(2,p),\\
p,&\text{otherwise,}
\end{cases}
\]
and constructs a left \(F_p\)-module isometry
\[
\Phi_p:M_2(F_p)\to F_{p^2}+uF_{p^2}
\]
that transports cyclic codes over the noncommutative matrix ring to additive cyclic codes over the chain ring \(F_{p^2}+uF_{p^2}\) while preserving distance. Under \(p\nmid n\), a cyclic code of length \(n\) over \(A_p\) has the form
\[
C=(F_1,\mathbf u_pF_2),
\qquad
F_0F_1F_2=X^n-I,
\]
with
\[
|C|=p^{2s},
\qquad
s=2\deg F_1+\deg F_2.
\]
This is a genuinely noncommutative Ring\(^2\) theory centered on \(M_2(F_p)\) itself [1409.7228].

A second coding-theoretic usage concerns the non-chain local ring
\[
R=\mathbb{Z}_p[u,v]/\langle u^2,\; v^2,\; uv-vu\rangle,
\]
whose defining feature is the presence of two commuting nilpotents of index \(2\). Every element has unique form
\[
a+ub+vc+uvd,\qquad a,b,c,d\in\mathbb Z_p,
\]
the ring has cardinality \(p^4\), maximal ideal \(M=\langle u,v\rangle\), and is not a chain ring because \(\langle u,v\rangle\) is not principal. Cyclic codes over
\[
R[x]/\langle x^n-1\rangle
\]
admit a unique four-generator canonical form
\[
C=\langle A_1(x),A_2(x),A_3(x),A_4(x)\rangle
\]
with divisibility chain
\[
a_3(x)\mid a_1(x)\mid g(x)\mid (x^n-1),
\qquad
a_3(x)\mid a_2(x)\mid g(x)\mid (x^n-1).
\]
The paper computes free rank, rank, minimal spanning sets, code size, and Hamming distance, and proves that the Hamming distance is controlled by the bottom torsion component
\[
C_{uv}=\langle a_3(x)\rangle.
\]
Its Gray map
\[
\phi_L(a+ub+vc+uvd)=(a+b+c+d,\ c+d,\ b+d,\ d)
\]
is a distance-preserving isometry from Lee distance on \(R^n\) to Hamming distance on \(\mathbb Z_p^{4n}\), and the Gray image of a cyclic code is \(4\)-quasi-cyclic. For \(p=3\) and \(n=3\), the construction yields all ternary optimal codes of length \(12\) except the code with parameters \([12,2,9]^*\) [1405.5981].

## 5. Graph-theoretic Ring\(^2\) and the square of a ring graph

In graph theory, a ring is a graph whose vertex set is partitioned into \(k\ge 4\) nonempty sets
\[
X_1,\dots,X_k
\]
such that each \(X_i\) is a clique, each \(X_i\) is anticomplete to all nonconsecutive bags, some vertex of each \(X_i\) is complete to \(X_{i-1}\cup X_{i+1}\), and the vertices of \(X_i\) are linearly ordered by domination. Hyperholes are the special case in which every vertex of \(X_i\) is complete to \(X_{i-1}\cup X_{i+1}\). The main theorem states that if \(R\) is a \(k\)-ring, then
\[
\chi(R)=\max\{\chi(H)\mid H\text{ is a }k\text{-hyperhole in }R\}.
\]
Even rings are perfect, and there is an \(O(n^6)\) algorithm that either returns an optimal coloring or certifies that the input is not a ring [1907.11905].

The same paper isolates several direct consequences for the graph square \(R^2\). If \(R\) has ring partition \((X_1,\dots,X_k)\), then in \(R^2\) a vertex of \(X_i\) can be adjacent only to vertices in
\[
X_{i-2}\cup X_{i-1}\cup X_i\cup X_{i+1}\cup X_{i+2}.
\]
Moreover, for every \(i\), the bags \(X_i\) and \(X_{i+2}\) are complete to each other in \(R^2\). This yields a \(5\)-local cyclic structure for the square. A plausible implication is that \(R^2\) is naturally compared with a blow-up of \(C_k^2\), and for hyperholes this comparison is exact in the sense recorded in the paper’s synthesis [1907.11905].

## 6. Two-dimensional ring patterns and square-grid geometry

In discrete differential geometry, Ring\(^2\) appears in an entirely different sense: orthogonal ring patterns in the plane. Here one starts with a cell complex \(G\) formed from a subset of quadrilaterals of the square lattice \(\mathbb Z^2\), and each vertex carries a ring consisting of an inner circle \(c_i\) of signed radius \(r_i\) and an outer circle \(C_i\) of radius \(R_i\), both concentric. Neighboring rings satisfy crossed orthogonality conditions, and each elementary square satisfies a touching condition involving two inner circles and two outer circles. Orthogonality forces all rings in a connected pattern to have the same area,
\[
\pi(R_i^2-r_i^2)=\pi\ell_0^2,
\]
so one may write
\[
R_i=\ell_0\cosh(\rho_i),\qquad r_i=\ell_0\sinh(\rho_i).
\]
The central result is that the \(\rho\)-variables satisfy exactly the same interior equation as orthogonal circle patterns:
\[
\sum_{j:\, v_j\sim v_i} 2\arctan(e^{\rho_i-\rho_j}) = 2\pi.
\]
Thus orthogonal ring patterns are governed by the same integrable equation as orthogonal circle patterns [1911.07095].

Because the equation depends only on differences \(\rho_i-\rho_j\), the shift
\[
\rho_i^\delta=\rho_i+\delta
\]
produces a one-parameter family of orthogonal ring patterns. After rescaling, the limit \(\delta\to+\infty\) yields the ordinary orthogonal circle pattern with radii \(e^{\rho_i}\), while \(\delta\to-\infty\) yields its dual with radii \(e^{-\rho_i}\). The paper constructs ring-pattern analogues of the Doyle spiral, Erf, and \(z^\alpha\) functions, and develops a variational principle with Hessian
\[
D^2S = \sum_{v_i\sim v_j} \frac{1}{\cosh(\rho_i-\rho_j)}(d\rho_i-d\rho_j)^2.
\]
This gives existence and uniqueness for Dirichlet boundary data and a one-parameter family for Neumann boundary data. In this geometric setting, Ring\(^2\) is a planar square-grid annular geometry interpolating between a circle pattern and its dual [1911.07095].

The cumulative picture is therefore plural rather than singular. In one direction, Ring\(^2\) means \(S^2\), the square-generated additive span of symmetric elements, with the central problem of deciding when \(S^2=R\). In another, it means a height-\(2\) extension \(R\subset S\). In radical theory it encodes constraints on unit squares such as \(u^2\in 1+\sqrt{J(R)}\) or \(u^2\in 1+Nil(R)+J(R)\). In coding theory it points either to the square matrix ring \(M_2(F_p)\) or to rings with two commuting square-zero generators. In graph theory it refers to the square \(R^2\) of a ring graph, and in discrete geometry to ring patterns on square-grid combinatorics. The notation is therefore best understood as a family resemblance built around “square” operations on ring-theoretic or ring-like objects, not as a single universally fixed definition.

Source: https://www.emergentmind.com/topics/ring-2