Characteristic Tilting Module

• Characteristic tilting modules are canonical tilting–cotilting objects defined by Ringel’s condition, possessing both Δ- and ∇-filtrations in quasi-hereditary algebras.
• They are characterized via IS-tilting module criteria and cotorsion pair kernels, linking intrinsic algebraic structures to homological properties.
• In modular representation theory, these modules play a key role in analyzing tilting behavior over fields of positive characteristic, adhering to Weyl and dual Weyl filtration conditions.

A characteristic tilting module is, in its standard and most precise sense, the canonical basic tilting–cotilting module attached to a quasi-hereditary algebra $(A,\unlhd)$. It is characterized by Ringel’s condition

$$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$

equivalently by being the direct sum of all Ext-injective objects in the category $\mathcal F(\Delta)$ of modules admitting a standard filtration. In current usage, however, the phrase also appears in broader homological settings: a tilting module determined by the kernel of a cotorsion pair, a canonical tilting object singled out by dominant dimension, or, more loosely, tilting modules for algebraic groups over fields of positive characteristic. The common feature is that the tilting object is extracted from intrinsic structure rather than chosen ad hoc (Adachi et al., 2019, Adachi et al., 22 Jul 2025).

1. Standard meaning in quasi-hereditary theory

Let $A$ be an artin algebra with simple modules $\{S(\lambda)\mid \lambda\in\Lambda\}$, projective covers $P(\lambda)$, injective hulls $I(\lambda)$, and a partial order $\le$ on $\Lambda$. The standard module $\Delta(\lambda)$ is the maximal factor of $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$0 whose composition factors are $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$1 with $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$2, and the costandard module $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$3 is defined dually as a submodule of $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$4 with the same composition-factor restriction. The pair $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$5 is quasi-hereditary when these modules satisfy the usual heredity conditions, in particular the existence of exact sequences

$$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$6

with $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$7 and filtration factors indexed strictly above $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$8 (Adachi et al., 2019).

In this setting, the characteristic tilting module $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$9 is the direct sum of all Ext-injective objects in $\mathcal F(\Delta)$0. It is a basic tilting module and also a cotilting module. Equivalently, it has both a $\mathcal F(\Delta)$1-filtration and a $\mathcal F(\Delta)$2-filtration. This makes it the canonical tilting object encoding the highest-weight structure itself. For a quasi-hereditary structure $\mathcal F(\Delta)$3, one writes

$\mathcal F(\Delta)$4

with each indecomposable summand satisfying

$\mathcal F(\Delta)$5

The characteristic tilting module determines the Ringel dual via $\mathcal F(\Delta)$6, and in graded highest-weight settings one likewise has

$\mathcal F(\Delta)$7

one indecomposable tilting summand for each weight (Adachi et al., 22 Jul 2025, Hazi et al., 2018).

2. Recognition of characteristic tilting modules

A basic problem is to decide which tilting modules arise as characteristic tilting modules of some quasi-hereditary structure. A recent answer is given in terms of IS-tilting modules. An IS-tilting module is a pair $\mathcal F(\Delta)$8 consisting of a tilting module and a partial order on its indecomposable summands such that iterative idempotent truncation along $\mathcal F(\Delta)$9 always reveals a simple direct summand. For tilting modules, the decisive condition reduces to the existence of a decomposition $A$0 with

$A$1

In that case, $A$2 is IS-tilting, $A$3 is quasi-hereditary, and

$A$4

Equivalently, a tilting module is characteristic if and only if there exists some partial order making it IS-tilting; this yields a bijection

$A$5

The same analysis shows that all tilting modules are characteristic precisely for quadratic linear Nakayama algebras. In that case, the set of tilting modules decomposes recursively, and the corresponding quasi-hereditary structures admit descriptions via nodal gluing and binary tree sequences; the counting formulas involve Catalan numbers (Adachi et al., 22 Jul 2025).

3. Strongly quasi-hereditary structures and Auslander-type comparison

The characteristic tilting module also serves as a diagnostic for stronger homological forms of quasi-heredity. A quasi-hereditary algebra $A$6 with characteristic tilting module $A$7 is right-strongly quasi-hereditary exactly when $A$8, equivalently when every standard module has projective dimension at most $A$9. Dually, it is left-strongly quasi-hereditary exactly when $\{S(\lambda)\mid \lambda\in\Lambda\}$0. If both hold, the algebra is strongly quasi-hereditary, and then $\{S(\lambda)\mid \lambda\in\Lambda\}$1 (Adachi et al., 2019).

This interacts closely with dominant-dimension constructions. For an almost $\{S(\lambda)\mid \lambda\in\Lambda\}$2-Auslander algebra, one has a distinguished relative tilting module $\{S(\lambda)\mid \lambda\in\Lambda\}$3, defined as the unique basic $\{S(\lambda)\mid \lambda\in\Lambda\}$4-tilting module in

$\{S(\lambda)\mid \lambda\in\Lambda\}$5

where $\{S(\lambda)\mid \lambda\in\Lambda\}$6 is the basic injective module collecting precisely the injectives of projective dimension at most $\{S(\lambda)\mid \lambda\in\Lambda\}$7. The characteristic tilting module $\{S(\lambda)\mid \lambda\in\Lambda\}$8 of a right-strongly quasi-hereditary structure can then be compared with $\{S(\lambda)\mid \lambda\in\Lambda\}$9. The key implication chain is

$P(\lambda)$0

If $P(\lambda)$1 is projective, these conditions become equivalent; for Auslander algebras of global dimension $P(\lambda)$2, they are also equivalent to $P(\lambda)$3 being a Nakayama algebra (Adachi et al., 2019).

4. Characteristic tilting modules from cotorsion-pair kernels

A broader homological use of the expression arises from complete hereditary cotorsion pairs $P(\lambda)$4 in $P(\lambda)$5. Here the kernel is

$P(\lambda)$6

Although the paper does not formally introduce the term, it explicitly identifies as natural the notion of a characteristic tilting module attached to $P(\lambda)$7: a tilting module $P(\lambda)$8 such that

$P(\lambda)$9

The main structural theorem gives equivalent criteria for this to happen. For a complete hereditary cotorsion pair and $I(\lambda)$0, the condition

$I(\lambda)$1

is equivalent to

$I(\lambda)$2

together with closure of $I(\lambda)$3 under arbitrary direct sums; it is also equivalent to a description of the right class of the form

$I(\lambda)$4

for a class $I(\lambda)$5 of strongly Gorenstein projective modules. The resulting $I(\lambda)$6 is unique up to additive equivalence (Wang et al., 2018).

In this framework, several classical questions become kernel-detection problems. For a left Noetherian ring,

$I(\lambda)$7

is equivalent to the kernel of the cotorsion pair built from $I(\lambda)$8, and equivalently the one built from $I(\lambda)$9, being of the form $\le$0 for some tilting module. A Wakamatsu tilting module $\le$1 of finite projective dimension is tilting exactly when the kernel of an associated cotorsion pair is $\le$2 for some tilting $\le$3. For commutative rings, Gorensteinness is characterized by the kernels of $\le$4 or $\le$5 being additive closures of tilting modules. In all these cases, the kernel acts as the carrier of a canonical tilting object (Wang et al., 2018).

5. Positive characteristic and representation-theoretic usage

In modular representation theory the same phrase can mean something different: tilting modules considered specifically over a field of prime characteristic. For $\le$6 with $\le$7, a finite-dimensional rational $\le$8-module is tilting when it admits both a Weyl filtration and a dual Weyl filtration. The category $\le$9 is additive, Karoubian, rigid, and monoidal, with indecomposables $\Lambda$0 indexed by dominant weights $\Lambda$1. In this literature, “characteristic tilting module” often means precisely a tilting module in positive characteristic, as opposed to a canonical Ringel object of a quasi-hereditary algebra (Tubbenhauer et al., 2020).

This usage emphasizes characteristic-dependent structure. For $\Lambda$2, the center of the tilting category in characteristic $\Lambda$3 is described by

$\Lambda$4

where the generators are nilpotent central natural transformations built from transported quiver loops. The block decomposition is controlled by $\Lambda$5-adic combinatorics, and indecomposable tilting modules decompose according to Donkin’s tensor product theorem (Tubbenhauer et al., 2020).

The dependence on $\Lambda$6-adic expansions is also explicit in tensor-product criteria. For $\Lambda$7 over an algebraically closed field of characteristic $\Lambda$8, the module

$\Lambda$9

is tilting if and only if, for the primitive pair $\Delta(\lambda)$0 of $\Delta(\lambda)$1, one has either

$\Delta(\lambda)$2

or the symmetric condition with $\Delta(\lambda)$3 and $\Delta(\lambda)$4 interchanged (Martin, 2017).

This modular picture is not purely affirmative. For $\Delta(\lambda)$5 in characteristic $\Delta(\lambda)$6, there exists a projective indecomposable $\Delta(\lambda)$7-module that is not the restriction of any indecomposable tilting $\Delta(\lambda)$8-module, yielding a counterexample to Donkin’s Tilting Module Conjecture; the same paper constructs a Weyl module without a $\Delta(\lambda)$9-Weyl filtration, giving a counterexample to the $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$00-Filtration Conjecture (Bendel et al., 2019). Complementarily, the Steinberg quotient

$$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$01

for $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$02 has a controlled orbit-sum expansion

$$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$03

with $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$04 exactly when $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$05, and $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$06 whenever $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$07 (Sobaje, 2019).

6. Broader homological avatars

Several adjacent developments isolate characteristic-like tilting objects without using the standard terminology. For algebras of positive dominant dimension, there are unique $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$08-special $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$09-tilting and $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$10-cotilting modules $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$11 and $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$12, called shifted and coshifted modules, where $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$13 is a maximal projective-injective summand. Their endomorphism algebras have global dimension bounded above by that of the original algebra, and a non-selfinjective algebra is minimal $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$14-Auslander–Gorenstein exactly when

$$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$15

These modules play a canonical role closely analogous to characteristic tilting modules in quasi-hereditary theory (Pressland et al., 2017).

Base change provides another characteristic mechanism. If $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$16 is a central non-unit non-zero-divisor on a ring $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$17, then any classical $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$18-tilting module $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$19 over $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$20 yields classical $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$21-tilting modules $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$22 over $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$23 and $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$24 over $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$25. Conversely, under noetherianity and regularity hypotheses, tilting over both $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$26 and $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$27 reflects back to tilting over $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$28. Since characteristic tilting modules of quasi-hereditary algebras are classical tilting modules with extra $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$29-filtration structure, this gives a base-change framework for transporting them (Moradifar et al., 2017).

A categorical variant appears in the study of the $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$30-structure induced by an $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$31-tilting module $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$32. The heart $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$33 is a Grothendieck category if and only if $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$34 is pure projective, and for $$\operatorname{add} T_{\unlhd}= \mathcal F(\Delta)\cap \mathcal F(\nabla),$$35-tilting modules the HRS heart is Grothendieck if and only if the tilting module is pure projective, while it is equivalent to a module category exactly when the tilting module is classical (Bazzoni, 2016, Bazzoni et al., 2017). These are not redefinitions of characteristic tilting modules, but they identify categorical signatures of tilting objects singled out by especially rigid behavior.

Taken together, these developments show that “characteristic tilting module” has a stable core and several controlled extensions. The core notion remains Ringel’s canonical tilting–cotilting object of a quasi-hereditary structure. Around it lie closely related usages in cotorsion theory, dominant-dimension methods, and modular representation theory, each preserving the same essential idea: a tilting module determined by the intrinsic architecture of the category or ring under study.