1 Definition and Basic Examples
1.1 Annihilator of a subset in a module
Let \(R\) be a ring and \(M\) a (left) \(R\)-module. For a subset \(S \subseteq M\), the annihilator of \(S\) is the set of ring elements that act as zero on every element of \(S\): \[ \operatorname{Ann}_R(S)=\{\,r\in R : r s=0 \text{ for all } s\in S\,\}. \] When \(S\) is a single element \(\{m\}\), one writes \(\operatorname{Ann}_R(m)\). When \(S\) is a submodule \(N\le M\), the same definition is used.
1.2 Annihilator ideal and annihilator submodule
If \(M\) is a left \(R\)-module, then \(\operatorname{Ann}_R(S)\) is a left ideal of \(R\). In commutative settings, it is an ideal in the usual sense. Dually, one can define an annihilator inside a module by taking ring elements that annihilate a given set and viewing them through module actions; however, the most common “submodule annihilator” language is: for a submodule \(N\le M\), elements of \(M\) that kill \(N\) (under an appropriate bilinear action) form an annihilating submodule. In many standard treatments (e.g., modules over a ring), the primary focus is the annihilator ideal \(\operatorname{Ann}_R(N)\).
1.3 Connections to the notion of “zero action”
The defining property \(rs=0\) captures a precise form of “zero action”: the ring element \(r\) has no effect on the subset \(S\) through the module structure. This is different from requiring \(rs=0\) for only one element of \(S\); the annihilator imposes the same vanishing simultaneously for all members of the subset.
1.4 Examples from rings and modules
- Regular module over itself. Take \(M=R\) with its natural left action. For a subset \(S\subseteq R\),
\[ \operatorname{Ann}_R(S)=\{r\in R: rS=0\}. \] For \(S=\{a\}\), this is the left annihilator of \(a\), often denoted \(\operatorname{Ann}_\ell(a)\).
- \(\mathbb{Z}\)-modules. If \(R=\mathbb{Z}\) and \(M=\mathbb{Z}\) as a module over itself, then for an element \(m\in\mathbb{Z}\),
\[ \operatorname{Ann}_\mathbb{Z}(m)=\{n\in\mathbb{Z}: nm=0\}. \] Over the integers, this equals \(0\) unless \(m=0\), reflecting that \(\mathbb{Z}\) has no nonzero zero divisors.
- Modules with torsion. Over a ring with zero divisors (or in modules containing torsion), annihilators can be nontrivial even when the subset contains nonzero elements. For instance, if \(R=k[x]/(x^2)\) and \(M=R\), then the class of \(x\) annihilates the ideal \((x)\), giving a concrete nonzero annihilator.
2 Annihilators in Module Theory
2.1 Properties of annihilators
2.1.1 Inclusion-reversing behavior
Annihilators reverse inclusion. If \(S\subseteq T\subseteq M\), then \[ \operatorname{Ann}_R(T)\subseteq \operatorname{Ann}_R(S). \] Reason: an element of \(R\) that kills all of \(T\) certainly kills all of \(S\).
2.1.2 Relation to kernels of homomorphisms
Annihilators often appear as kernels. For example, given a submodule \(N\le M\), define a map \[ R\to \operatorname{Hom}_\mathbb{Z}(N,M),\quad r\mapsto (n\mapsto rn). \] Then \(\operatorname{Ann}_R(N)\) is the kernel of the action map restricted to \(N\). In contexts where the module action is encoded by a homomorphism, annihilators become “the elements acting trivially,” i.e., the kernel of the induced action.
2.1.3 Behavior under sums and intersections
A basic guiding principle is: annihilators of larger sets are smaller, and annihilators interact with algebraic operations on submodules. For submodules \(N_1,N_2\le M\),
- \(\operatorname{Ann}_R(N_1+N_2)\subseteq \operatorname{Ann}_R(N_1)\cap \operatorname{Ann}_R(N_2)\) because killing the sum implies killing each part.
- \(\operatorname{Ann}_R(N_1)\cup \operatorname{Ann}_R(N_2)\) is not typically an annihilator, but \(\operatorname{Ann}_R(N_1\cap N_2)\) contains both \(\operatorname{Ann}_R(N_1)\) and \(\operatorname{Ann}_R(N_2)\) (since killing a larger submodule ensures killing the intersection).
Exact equalities depend on additional structure; the inclusions are universal and follow directly from the definition.
2.2 Double annihilators
2.2.1 Inclusion between a subset and its double annihilator
For module annihilators, there is a standard “double annihilator” relationship in many settings: elements that annihilate an annihilator tend to contain the original set. Concretely, when one has a natural pairing or when the module is viewed through its action, one obtains inclusions of the form \[ N \subseteq \operatorname{Ann}_M(\operatorname{Ann}_R(N)), \] where \(\operatorname{Ann}_M(J)\) denotes the submodule of \(M\) killed by the ideal \(J\subseteq R\). The exact formulation depends on whether one treats annihilators in \(R\) or in \(M\), but the overarching theme is that “killing what kills \(N\)” does not shrink \(N\) in typical module contexts.
2.2.2 Criteria for equality in special settings
Equality in double annihilator statements can fail in general. It can hold under additional hypotheses such as reflexivity-like properties, faithfulness constraints, or when the module action is sufficiently nondegenerate. In commutative algebra, analogous phenomena often depend on finiteness conditions and on how support and torsion control the annihilator behavior.
2.3 Annilhilators and quotient modules
2.3.1 Annihilator of a quotient
If \(N\le M\), then the annihilator of the quotient \(M/N\) consists of ring elements that act trivially modulo \(N\): \[ \operatorname{Ann}_R(M/N)=\{\,r\in R : r m\in N\text{ for all }m\in M\,\}. \] This links annihilators directly to how the submodule \(N\) absorbs the action of \(r\).
2.3.2 Exact sequences and annihilator bounds
Annihilators interact with exact sequences through containment relationships. If \[ 0\to A\to B\to C\to 0 \] is exact, then annihilators of the outer terms control that of the middle term, and vice versa, often via inclusions reflecting how elements act through the maps. For example, an element of \(R\) that kills \(B\) must kill both \(A\) and the image of \(B\) in \(C\), forcing \[ \operatorname{Ann}_R(B)\subseteq \operatorname{Ann}_R(A)\cap \operatorname{Ann}_R(C). \] Refined statements can be given using Tor/Ext or depth-like invariants, but the basic idea is that annihilation is compatible with the module structure along exact sequences.
3 Annihilators in Ring Theory
3.1 Annihilator ideals of elements
For \(a\in R\), the left annihilator of \(a\) is \[ \operatorname{Ann}_\ell(a)=\{\,r\in R : ra=0\,\}, \] and the right annihilator is \[ \operatorname{Ann}_r(a)=\{\,r\in R : ar=0\,\}. \] In commutative rings, these coincide and give the ordinary annihilator ideal \(\operatorname{Ann}(a)\).
3.2 Left, right, and two-sided annihilators
3.2.1 Differences in noncommutative settings
In noncommutative rings, left and right annihilators can differ because multiplication is not symmetric. For an element \(a\),
- \(\operatorname{Ann}_\ell(a)\) collects multipliers on the left that force \(a\) to vanish.
- \(\operatorname{Ann}_r(a)\) collects multipliers on the right that force \(a\) to vanish.
Their intersection is a two-sided annihilator: \[ \operatorname{Ann}_{2\text{-sided}}(a)=\operatorname{Ann}_\ell(a)\cap \operatorname{Ann}_r(a). \] This distinction is central when studying modules that are not commutative with respect to scalars.
3.3 Annihilators and zero divisors
3.3.1 Right/left zero divisors via annihilators
An element \(a\) is a left zero divisor if there exists a nonzero \(r\in R\) with \(ra=0\), which is equivalent to \(\operatorname{Ann}_\ell(a)\neq 0\). Similarly, \(a\) is a right zero divisor precisely when \(\operatorname{Ann}_r(a)\neq 0\). This makes annihilators a structural tool for detecting zero divisor behavior.
3.4 Annihilators of ideals
3.4.1 Product ideals and annihilator containment
For an ideal \(I\lhd R\), the annihilator \[ \operatorname{Ann}_R(I)=\{\,r\in R : rI=0\,\} \] is an ideal (left ideal in general, two-sided in commutative settings). Containments such as \[ \operatorname{Ann}_R(IJ)\supseteq \operatorname{Ann}_R(I)\cap \operatorname{Ann}_R(J) \] reflect how multiplication interacts with “killing.” In commutative rings, one often rewrites annihilator conditions using colon ideals and relates them to product decompositions of ideals.
4 Annihilators in Linear Algebra
4.1 Vector spaces as modules
A vector space \(V\) over a field \(k\) is a \(k\)-module. Since \(k\) is a field, annihilators are typically simple: for a nonzero vector \(v\), the only scalar \(a\in k\) with \(av=0\) is \(a=0\). Thus over fields, annihilators do not capture much internal structure. The conceptual bridge is clearer when one views more general linear algebra objects—such as modules over algebras, representations, or endomorphism rings—where scalar actions are replaced by richer algebraic actions.
4.2 Annihilators of vectors and subspaces
If \(V\) is a module over a ring \(R\), one can still define \(\operatorname{Ann}_R(v)\) for a vector \(v\) and \(\operatorname{Ann}_R(W)\) for a subspace \(W\le V\). In representation-theoretic linear algebra, \(R\) may be an algebra acting on \(V\) via linear transformations. Then \(\operatorname{Ann}_R(W)\) is the set of operators that send every vector in \(W\) to zero.
4.3 Orthogonality vs. annihilation (duality perspective)
Annihilation involves the module action producing zero. Orthogonality in inner-product spaces involves a bilinear form pairing vectors to zero. These notions are related when one uses dual spaces and adjoint operators: annihilators in representation contexts correspond to kernels of induced maps on duals or to orthogonality with respect to a pairing. However, annihilation and orthogonality are not identical in general; the relationship depends on having an appropriate nondegenerate pairing or identifying the dual action.
4.4 Applications to linear maps
4.4.1 Annihilator of the image and the kernel (conceptual bridge)
Given a linear map \(T: V\to W\), the kernel \(\ker(T)\) is the set of vectors killed by \(T\). When \(T\) comes from an algebra action, such as \(T(v)=a\cdot v\) for some operator \(a\), then annihilator language reappears: \(\ker(T)=\operatorname{Ann}_{\langle a\rangle}(V)\) in a suitable cyclic-action sense. More generally, for families of maps (or an algebra of endomorphisms), annihilators describe vectors that are killed simultaneously by all operators in a chosen set, while quotients and kernels control how much action remains.
5 Computation and Techniques
5.1 Using generators to compute annihilators
When \(S\subseteq M\) is finitely generated, the annihilator can often be computed from generators. If \(S=\langle s_1,\dots,s_t\rangle\) is the submodule generated by \(s_1,\dots,s_t\), then \[ \operatorname{Ann}_R(S)=\bigcap_{i=1}^t \operatorname{Ann}_R(s_i), \] because killing each generator forces killing every element in the generated submodule.
5.2 Gröbner-basis style viewpoints (conceptual)
For modules presented by generators and relations, annihilators correspond to solving equations \(r s=0\) inside \(M\). In algorithmic settings (especially commutative algebra), computational methods akin to Gröbner bases can be used to compute these solutions systematically. The idea is to transform annihilation constraints into membership and elimination problems in polynomial rings, where standard polynomial techniques apply.
5.3 Decompositions and annihilator behavior
5.3.1 Primary decomposition connections (high-level)
In commutative noetherian settings, annihilators are closely tied to ideal decompositions. When modules decompose into primary components (or when ideals admit primary decompositions), annihilators often split correspondingly or admit descriptions in terms of radicals of annihilating ideals. This connection underlies many structural theorems relating annihilators, support, and decomposition of modules into simpler pieces.
5.4 Practical examples over common rings
- Quotient rings. In \(R=k[x]/(f)\), annihilators of ideals correspond to divisibility relations among polynomials, since multiplication by classes is governed by the factor structure of \(f\).
- Finite-dimensional algebras. If \(R\) acts on a finite-dimensional vector space \(V\) via matrices, then annihilators correspond to matrices (elements of \(R\)) that satisfy linear constraints “acting as zero” on a chosen subspace. These constraints can be solved by standard linear algebra over the base field.
6 Further Properties and Generalizations
6.1 Annihilator under scalar extension
Given a ring homomorphism \(R\to R'\) and an \(R\)-module \(M\), one can extend scalars to form \(R'\otimes_R M\). Annihilators behave compatibly in many cases: elements of \(R\) that annihilate a subset typically lead to elements in \(R'\) that annihilate the extended subset. Exact formulas may require additional hypotheses (such as flatness of \(R'\) over \(R\)), but the guiding principle is that scalar extension preserves “zero action” in a controlled manner.
6.2 Annihilators in graded modules
If \(R\) is a graded ring and \(M\) a graded module, annihilators can be studied in graded form. One often considers the homogeneous components of elements that annihilate a graded submodule. This refinement is useful because the grading can turn global annihilation questions into degree-by-degree constraints.
6.3 Compatibility with localization (high-level)
Localization in commutative algebra replaces the ring by one where certain elements become invertible. Annihilators are compatible with localization: when localizing at a multiplicative set, annihilating behavior can be checked after localization, and annihilator ideals localize accordingly under appropriate assumptions. Conceptually, localizing isolates the algebraic behavior “near” prime ideals, making annihilator computations more local and tractable.
6.4 Annihilators in commutative algebra settings
6.4.1 Support and annihilator relationships (overview)
In commutative algebra, the support of a module is the set of prime ideals where the module does not vanish after localization. Annihilators influence support because if an ideal \(I\subseteq R\) annihilates a module \(M\), then the support of \(M\) lies inside the variety (or prime set) determined by \(I\). Conversely, the radicals of annihilators often encode where a module lives. In this way, annihilators provide a bridge between algebraic “killing” and geometric/topological information expressed through primes.