Triangular Coefficient Matrix Rings

• Triangular coefficient matrix rings are upper-triangular structures with diagonal rings and off-diagonal bimodules governed by specialized multiplication rules.
• They feature block decompositions and central idempotents that simplify the analysis of derivations, radicals, and other ring-theoretic properties.
• Applications span polynomial, combinatorial, and module-category frameworks, offering deep insights and practical tools for modern ring theory research.

Searching arXiv for recent and foundational papers on triangular coefficient matrix rings and closely related triangular matrix ring structures. A triangular coefficient matrix ring is an upper-triangular matrix ring in which diagonal entries are taken from prescribed rings and off-diagonal entries from prescribed bimodules, with multiplication determined by the ring structures on the diagonal, the bimodule actions, and, in higher-block forms, additional composition maps between off-diagonal blocks (Chen et al., 2020). In the $2\times2$ case this is the formal triangular matrix ring $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$, while in the $3\times3$ case it is the triangular $3$-matrix ring $\mathcal T_3(R_i;M_{ij})$ (Bennis et al., 2021). The phrase is also used for upper triangular rings viewed as trivial extensions of product rings (Cheraghpour et al., 2024), and a 2025 paper introduces a different upper-triangular ring $\hat H_n(R)$ with binomially weighted multiplication under the same name (Danchev et al., 20 Jul 2025). This suggests a terminological family of closely related coefficient-sensitive triangular constructions rather than a single universally fixed definition.

1. Terminology and defining constructions

In the three-block setting, let $R_1,R_2,R_3$ be unital rings and, for $1\le i<j\le 3$, let $M_{ij}$ be $(R_i,R_j)$-bimodules, with $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$0. For each $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$1, suppose there are bimodule homomorphisms

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$2

and write $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$3. If each $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$4 is faithful as a left $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$5-module and as a right $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$6-module, the triangular $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$7-matrix ring is

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$8

with usual addition and multiplication induced by the ring multiplications on the diagonal, the module actions, and the maps $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$9 (Chen et al., 2020).

This construction specializes in several standard ways. The usual upper triangular $3\times3$0 matrix ring $3\times3$1 is the case $3\times3$2, $3\times3$3, with standard multiplication (Chen et al., 2020). For $3\times3$4, one recovers the classical triangular ring

$3\times3$5

and the same pattern extends to triangular $3\times3$6-matrix rings in the sense of Ferreira (Chen et al., 2020).

In the formal $3\times3$7 setting, with rings $3\times3$8 and a $3\times3$9-bimodule $3$0, the triangular matrix ring is

$3$1

so the off-diagonal coefficient bimodule is $3$2 (Bennis et al., 2021).

A further realization identifies upper triangular rings with trivial extensions. If $3$3 are rings with identity and $3$4 is a unitary $3$5-bimodule, then

$3$6

is canonically isomorphic to the trivial extension $3$7 by the map

$3$8

where $3$9 is viewed as an $\mathcal T_3(R_i;M_{ij})$0-bimodule via $\mathcal T_3(R_i;M_{ij})$1 and $\mathcal T_3(R_i;M_{ij})$2 (Cheraghpour et al., 2024).

2. Block decomposition, center, and standard assumptions

For a triangular $\mathcal T_3(R_i;M_{ij})$3-matrix ring $\mathcal T_3(R_i;M_{ij})$4, there are canonical idempotents

$\mathcal T_3(R_i;M_{ij})$5

with $\mathcal T_3(R_i;M_{ij})$6, $\mathcal T_3(R_i;M_{ij})$7 for $\mathcal T_3(R_i;M_{ij})$8, and $\mathcal T_3(R_i;M_{ij})$9. Writing

$\hat H_n(R)$0

one has $\hat H_n(R)$1, $\hat H_n(R)$2 for $\hat H_n(R)$3, $\hat H_n(R)$4 for $\hat H_n(R)$5, and

$\hat H_n(R)$6

with $\hat H_n(R)$7 when $\hat H_n(R)$8 (Chen et al., 2020). This block decomposition is the basic device behind both structural and homological arguments.

The center is explicit: $\hat H_n(R)$9 Hence every central element is diagonal, and the diagonal entries must act compatibly on the bimodules from the left and right (Chen et al., 2020).

A recurring hypothesis is the standard assumption

$R_1,R_2,R_3$0

Since $R_1,R_2,R_3$1, this is the requirement that the center of $R_1,R_2,R_3$2 project onto the center of each diagonal corner (Chen et al., 2020). In the $R_1,R_2,R_3$3 literature this is the analogous hypothesis used in triangular rings and generalized matrix algebras, and in the $R_1,R_2,R_3$4-block situation it controls center-valued parts of Lie-type maps.

The standard assumption is not vacuous. Example 2.1 in the multiplicative Lie derivation paper constructs a subring $R_1,R_2,R_3$5 that is a triangular $R_1,R_2,R_3$6-matrix ring but cannot be realized as a $R_1,R_2,R_3$7 triangular ring; its center is

$R_1,R_2,R_3$8

and the diagonal corners satisfy

$R_1,R_2,R_3$9

(Chen et al., 2020).

3. Lie-, Jordan-, and clean-structure phenomena

A multiplicative Lie derivation on an associative ring $1\le i<j\le 3$0 is a map $1\le i<j\le 3$1, not assumed additive, such that

$1\le i<j\le 3$2

For triangular $1\le i<j\le 3$3-matrix rings, the main structural theorem states that if

$1\le i<j\le 3$4

then every multiplicative Lie derivation $1\le i<j\le 3$5 has the standard form

$1\le i<j\le 3$6

where $1\le i<j\le 3$7 is a derivation and $1\le i<j\le 3$8 is center-valued with $1\le i<j\le 3$9 for all $M_{ij}$0 (Chen et al., 2020). The proof first produces a decomposition

$M_{ij}$1

with $M_{ij}$2 taking values in singular subspaces $M_{ij}$3 or $M_{ij}$4, and then shows under the full standard assumption that this residual term is absorbed into a derivation (Chen et al., 2020). A common misconception is that non-additivity forces essentially new Lie behavior; in this setting the non-derivation part is forced into a center-valued map invisible on commutators.

For upper triangular matrix rings viewed as trivial extensions, Jordan superderivations exhibit an analogous rigidity. If

$M_{ij}$5

with $M_{ij}$6 $M_{ij}$7-torsion free, then every Jordan superderivation decomposes into an even part

$M_{ij}$8

and an odd part

$M_{ij}$9

where $(R_i,R_j)$0 are Jordan derivations, $(R_i,R_j)$1 satisfies the bimodule compatibility relation

$(R_i,R_j)$2

and $(R_i,R_j)$3 (Cheraghpour et al., 2024). In particular, every Jordan superderivation of $(R_i,R_j)$4 is a Jordan derivation, every degree-$(R_i,R_j)$5 Jordan superderivation is inner, and if $(R_i,R_j)$6 is faithful as a left $(R_i,R_j)$7-module and right $(R_i,R_j)$8-module, then every Jordan superderivation is a derivation (Cheraghpour et al., 2024).

A related but different rigidity problem concerns strong cleanness in skew triangular matrix rings $(R_i,R_j)$9, where multiplication in upper entries is twisted by an endomorphism $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$00. For a local ring $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$01, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$02 is strongly clean if and only if $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$03 is surjective for any $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$04, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$05, and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$06 is strongly clean under surjectivity of

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$07

for any $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$08, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$09 (Chen et al., 2013). The paper also proves necessary conditions for $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$10 and identifies the $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$11 case as open (Chen et al., 2013).

4. Module categories and homological algebra

For the formal triangular ring

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$12

the category $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$13-Mod is equivalent to the category of triples $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$14, where $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$15 is a left $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$16-module, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$17 is a left $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$18-module, and

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$19

is $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$20-linear (Bennis et al., 2021). A short exact sequence of left $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$21-modules is exact if and only if it is exact in each component. Projective modules are characterized by: $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$22 projective over $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$23, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$24 projective over $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$25, and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$26 injective; injective modules are characterized dually through $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$27, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$28, and surjectivity of the adjoint map $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$29 (Bennis et al., 2021).

This triple description supports several layers of relative homological algebra. In the relative Gorenstein framework of $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$30, if $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$31 is $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$32-compatible, then a left $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$33-module

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$34

is $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$35-projective if and only if $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$36 is injective, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$37 is $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$38-projective over $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$39, and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$40 is $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$41-projective over $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$42 (Bennis et al., 2021). Under the same compatibility hypotheses, the relative global dimension $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$43 is bounded between

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$44

and

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$45

and for the standard extension $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$46 one has

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$47

(Bennis et al., 2021).

Classical Gorenstein dimensions admit an analogous componentwise control. Under finite flat/projective dimension hypotheses on $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$48, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$49, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$50, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$51, and the strongly Gorenstein versions $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$52, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$53, are characterized by the corresponding dimensions of $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$54, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$55, or $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$56, together with injectivity or surjectivity of the structural maps at suitable syzygies or cosyzygies (Zhu et al., 2014). In the same spirit, Ding theory over formal triangular matrix rings shows that if $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$57 and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$58 have finite flat dimensions, then a left $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$59-module $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$60 is Ding projective if and only if $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$61 and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$62 are Ding projective and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$63 is a monomorphism, while right Ding injective modules are described via $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$64, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$65, and surjectivity of $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$66 under coherence and finiteness assumptions on $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$67 (Mao, 2019).

The Gorenstein projective category over

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$68

also admits a clean description: if $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$69 and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$70, then

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$71

is Gorenstein projective if and only if $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$72-GProj and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$73 is injective with Gorenstein projective cokernel in $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$74-Mod (Li et al., 2019). The same paper proves that the standard recollement of $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$75 restricts to subcategories of complexes of finite Gorenstein projective dimension and yields recollements of the stable category $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$76-GProj and of the Gorenstein defect category (Li et al., 2019).

More recently, the same formal triangular framework has been used to transport hereditary left and right $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$77-cotorsion pairs from $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$78-Mod and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$79-Mod to $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$80-Mod by means of special classes $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$81, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$82, and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$83, under $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$84- and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$85-vanishing assumptions involving $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$86 (Long et al., 14 Feb 2025).

5. Polynomial, combinatorial, and graph-theoretic perspectives

Triangular matrix rings also admit combinatorial models. For a nonsingular lower triangular matrix $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$87 over a ring, one can associate a weighted bipartite graph $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$88 whose perfect matching is given by the diagonal edges. If $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$89, then for $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$90,

$$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$91

where $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$92 is the set of alternating paths from $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$93 to $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$94 in $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$95, $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$96, and $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$97 is the ordered alternating product of edge weights and inverse weights along $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$98 (Bapat et al., 2013). In the real nonnegative case, if $$\begin{pmatrix}A&0\ U&B\end{pmatrix}$$99 is bipartite, then $3\times3$00 and $3\times3$01 have the same zero–nonzero pattern if and only if $3\times3$02 is a corona of a bipartite graph (Bapat et al., 2013).

Polynomial equations over upper triangular rings exhibit similarly rigid behavior. For an associative ring $3\times3$03 with identity and a quadratic

$3\times3$04

having two different roots $3\times3$05 such that $3\times3$06 is not a right zero divisor, a matrix $3\times3$07 or $3\times3$08 satisfies

$3\times3$09

if and only if every diagonal entry lies in $3\times3$10, every $3\times3$11 with $3\times3$12 is arbitrary, and whenever $3\times3$13 the entry $3\times3$14 is determined recursively by

$3\times3$15

for $3\times3$16, while $3\times3$17 (Gargate et al., 2020). In the finite case, the number of such $3\times3$18 matrices is

$3\times3$19

(Gargate et al., 2020).

For polynomial functions on upper triangular matrix algebras over a commutative ring, matrix-coefficient polynomials are reduced to scalar-coefficient data through the isomorphism

$3\times3$20

The entries of a matrix-coefficient polynomial are constrained by scalar integer-valued polynomial rings on smaller triangular algebras, and the resulting sets of right and left integer-valued polynomials with coefficients in $3\times3$21 are subrings of $3\times3$22; similarly, right and left null-polynomials with matrix coefficients form ideals of $3\times3$23 (Frisch, 2016).

Graph invariants of triangular rings have also been computed explicitly. For the unitary Cayley graph of $3\times3$24, with $3\times3$25, the graph is isomorphic to the semistrong product

$3\times3$26

and if $3\times3$27, it has $3\times3$28 connected components, each isomorphic to $3\times3$29 (Hołubowski et al., 2024). For commuting graphs of the triangular ring

$3\times3$30

the finite-field case yields a decomposition into $3\times3$31 cliques of size $3\times3$32, whereas for $3\times3$33 not a field, the commuting graph is connected with diameter $3\times3$34, and its clique number is $3\times3$35 (Cheraghpour et al., 2024).

6. Alternative usages, transfer properties, and scope

A distinct recent usage introduces the triangular coefficient matrix ring

$3\times3$36

as the set of all upper triangular $3\times3$37 matrices over $3\times3$38 with usual addition but modified multiplication

$3\times3$39

Its constant-diagonal subring

$3\times3$40

is naturally isomorphic to the truncated Hurwitz polynomial ring

$3\times3$41

via

$3\times3$42

(Danchev et al., 20 Jul 2025). In this setting the Jacobson radical, Brown–McCoy radical, prime radical, upper nilradical, Wedderburn radical, and Levitzki radical are all computed explicitly from the corresponding radicals of $3\times3$43; for example,

$3\times3$44

and hence

$3\times3$45

(Danchev et al., 20 Jul 2025). The same paper proves transfer equivalences for many ring-theoretic properties: $3\times3$46 is local, semi-local, matrix local, semi-primary, Jacobson, $3\times3$47-primal, of left stable range one, Dedekind finite, weakly symmetric, $3\times3$48-symmetric, abelian, clean, nil-clean, or $3\times3$49-good if and only if $3\times3$50 has the same property (Danchev et al., 20 Jul 2025).

At a more order-theoretic extreme, structural matrix rings

$3\times3$51

identify upper and lower triangular matrix rings with incidence algebras of linear orders, while more general preorders yield structural matrix rings, and ideal-valued matrices $3\times3$52 satisfying $3\times3$53 classify all subrings of $3\times3$54 containing the diagonal ring (Foldes et al., 2010). In this sense, triangular coefficient constructions sit inside a wider incidence-algebra framework.

A recurrent source of confusion is therefore terminological rather than structural. In the available literature, “triangular coefficient matrix ring” may refer to a formal triangular matrix ring $3\times3$55, a higher triangular $3\times3$56-matrix ring $3\times3$57, a trivial extension model of an upper triangular ring, or the Hurwitz-type ring $3\times3$58 with binomially weighted multiplication (Bennis et al., 2021). What unifies these usages is the replacement of homogeneous scalar entries by heterogeneous coefficient data—rings, bimodules, or coefficient sequences—and the consequent block-sensitive behavior of centers, derivations, ideals, radicals, and homological invariants. A plausible implication of the current literature is that the $3\times3$59 and $3\times3$60 cases are comparatively well understood, while genuinely higher-rank triangular coefficient systems remain technically harder; for example, the multiplicative Lie derivation methods developed for $3\times3$61-matrix rings are stated not to extend to $3\times3$62 (Chen et al., 2020).