Fact-checked by Grok 2 weeks ago

Projective module

In mathematics, particularly in abstract algebra, a projective module over a ring R is an R-module P that is a direct summand of a free R-module, meaning there exists another R-module Q such that P \oplus Q is free.^[1] This structure generalizes free modules, which are projective by definition, and captures properties analogous to those of vector spaces in linear algebra.^[2] Projective modules satisfy the projective lifting property: for any surjective homomorphism g: M \to N of R-modules and any homomorphism f: P \to N, there exists a homomorphism h: P \to M such that g \circ h = f.^[1] Equivalently, the functor \Hom_R(P, -) is exact, preserving surjections.^[1] Over a field, every module is free and hence projective, but over more general rings, projective modules may not be free; for example, \mathbb{Z}/2\mathbb{Z} and \mathbb{Z}/3\mathbb{Z} are projective but not free \mathbb{Z}/6\mathbb{Z}-modules.^[2] Projective modules play a central role in homological algebra, where they form projective resolutions to compute derived functors like \Ext and \Tor, with \Ext^1_R(P, M) = 0 for all M characterizing projectivity.^[1] In algebraic K-theory, they are the fundamental objects defining K-groups, such as K_0(R), the Grothendieck group of isomorphism classes of finitely generated projective R-modules.^[3] Moreover, finitely generated projective modules over a commutative ring R correspond to vector bundles over the affine scheme \Spec(R), providing a bridge between algebra and geometry. This is known as the algebraic Serre–Swan theorem.^[3]^[4]

Definitions and Characterizations

Lifting Property

A module P over a ring R is projective if, for every surjective R-module homomorphism f: M \to N and every R-module homomorphism g: P \to N, there exists an R-module homomorphism h: P \to M such that f \circ h = g.^[5] This condition, known as the lifting property, ensures that homomorphisms from P can always be "lifted" through surjections without obstruction. The lifting property is illustrated by the following commutative diagram, where the existence of the map h (often depicted as a dashed arrow) makes the square commute:

\begin{array}{ccc} P & \xrightarrow{h} & M \\ \downarrow g & & \downarrow f \\ N & \xrightarrow{\id} & N \end{array}

Here, the bottom row is exact (with f surjective), and the right column is the identity on N.^[5] This diagram captures the intuitive notion of projectivity: P behaves like a "free" or "projectable" object, allowing mappings to bypass extensions or quotients seamlessly, much as vector spaces over fields permit arbitrary projections onto subspaces. The concept originated in the 1956 book Homological Algebra by Henri Cartan and Samuel Eilenberg, where it was introduced as a generalization of the projection properties inherent to free modules and vector spaces.^[5] In this framework, the lifting property avoids cohomological obstructions in extensions, enabling P to split short exact sequences ending at it. A brief sketch of why this property implies P is a direct summand of a free module proceeds as follows: since every module is a quotient of a free module, let F be free with a surjection \pi: F \twoheadrightarrow P; applying the lifting property to the identity map \id_P: P \to P yields a section s: P \to F such that \pi \circ s = \id_P, so F \cong P \oplus \ker \pi.^[5] This equivalence to the direct summand characterization is explored further in subsequent sections.

Direct Summand Characterization

A module P over a ring R is projective if it is a direct summand of a free R-module F, that is, there exists another R-module Q such that F \cong P \oplus Q.^[6] This structural characterization emphasizes the "projective" nature of P as a complemented submodule within a free module, reflecting its ability to be "projected" onto without loss of information.^[3] This definition is equivalent to the lifting property for projective modules. To see that the lifting property implies the summand characterization, consider a surjection \pi: F \to P where F is free; projectivity ensures a section s: P \to F exists such that \pi \circ s = \mathrm{id}_P, so F \cong P \oplus \ker(\pi). Conversely, if P is a direct summand of a free module F = P \oplus Q, then the projection \pi: F \to P splits, and maps to P lift through surjections by composing with the inclusion of P into F and using the projectivity of free modules.^[6]^[3] For a free module F = R^n, the splitting is realized by a projection \pi: F \to P and its section s: P \to F satisfying \pi \circ s = \mathrm{id}_P, with \ker(\pi) \cong Q. This idempotent correspondence arises from endomorphisms of F, where the image of s is P and the kernel of \pi is Q.^[3] Every free module is projective, as one may take Q = 0. However, the converse fails: for example, over the ring R = M_n(F) of n \times n matrices over a field F, the module V of row vectors is projective but not free, as it lacks a basis generating R as an ideal.^[3] In the category of R-modules, direct summands are unique up to canonical isomorphism: if F \cong P \oplus Q \cong P \oplus Q', then Q \cong Q'. This follows from the universal property of direct sums in abelian categories.^[3]

Dual Basis Lemma

The dual basis lemma provides a constructive characterization of projective modules in terms of bases and dual functionals, bridging the abstract lifting property with explicit algebraic structures. For a left R-module P, where R is a unital ring, P is projective if and only if it admits a dual basis. That is, there exist families \{p_i\}_{i \in I} \subseteq P and \{\varphi_i\}_{i \in I} \subseteq \Hom_R(P, R) such that for every x \in P,

x = \sum_{i \in I} \varphi_i(x) \, p_i,

where the sum is finite (i.e., \varphi_i(x) = 0 for all but finitely many i). The proof proceeds by first assuming P is a direct summand of a free module F = \bigoplus_{i \in I} R e_i, with inclusion j: P \hookrightarrow F and retraction \pi: F \twoheadrightarrow P. For each basis element e_i of F, define p_i = \pi(e_i) \in P and \varphi_i = \ev_{e_i} \circ j \in \Hom_R(P, R), where \ev_{e_i} is the evaluation at the i-th coordinate. Then the identity map on P decomposes as required. Conversely, given a dual basis, one constructs a free module on the p_i and defines a retraction using the functionals to split the surjection onto P. This relies on the finite support to ensure well-defined maps. A key corollary arises over local rings: if R is local, then every projective R-module is free. This follows from the dual basis lemma, as the maximal ideal annihilates non-basis elements in the finite case, forcing a basis without relations; the general case extends by Kaplansky's global-to-local argument. This lemma, introduced by Irving Kaplansky in the mid-20th century, generalizes the finite-dimensional dual basis from linear algebra to arbitrary rings, facilitating computations and connections to invertible ideals.

Equivalent Conditions via Exactness

One key characterization of projective modules involves split-exact sequences. Consider a short exact sequence of the form $0 \to K \to F \to P \to 0, where F is a free R-module. This sequence is split-exact if there exists a homomorphism s: P \to F such that the composition F \to P \xrightarrow{s} F is the identity on P. In this case, P is isomorphic to a direct summand of F, and conversely, if P is a direct summand of some free module, then every such presentation splits.^[7] More generally, an R-module P is projective if and only if every short exact sequence $0 \to A \to B \to P \to 0 splits, meaning there exists a retraction B \to P making B \cong A \oplus P. This condition extends the lifting property, where for any surjection \epsilon: F \twoheadrightarrow M with F free and any homomorphism f: P \to M, there exists a lift g: P \to F such that \epsilon \circ g = f; applying this to the presentation of P itself yields the splitting.^[7] Projectivity can also be characterized via the exactness of the Hom functor. Specifically, P is projective if and only if the covariant functor \Hom_R(P, -): \Mod_R \to \Ab is exact, meaning it preserves both kernels and cokernels, or equivalently, maps short exact sequences to short exact sequences. Since \Hom_R(P, -) is always left exact, this condition is equivalent to it being right exact as well. For instance, if $0 \to A \to B \to C \to 0 is short exact, then $0 \to \Hom_R(P, A) \to \Hom_R(P, B) \to \Hom_R(P, C) \to 0 is exact precisely when P is projective.^[8] In the context of surjections from free modules, projectivity ensures splitting under certain images. Given a surjection \epsilon: F \twoheadrightarrow M with F free, if there is a homomorphism \phi: P \to M such that the image of \phi is P (identifying P with a submodule of M), then projectivity of P implies the existence of a splitting \sigma: P \to F with \epsilon \circ \sigma = \phi, making the induced sequence split. This follows directly from the lifting property applied to \phi.^[7] This exactness of \Hom_R(P, -) distinguishes projective modules from flat modules. While every projective module is flat (as direct summands of free modules preserve exactness under tensor products), the converse does not hold; flatness means the functor -\otimes_R Q is exact for a flat Q, preserving exact sequences via tensor product, whereas projectivity requires the stronger condition that \Hom_R(P, -) is fully exact, not merely left exact.^[8]

Basic Examples and Properties

Elementary Examples

Free modules provide the most straightforward examples of projective modules. Over any ring R, a free R-module R^n (for n \geq 0) is projective, as it is a direct summand of itself via the identity map.https://www.math.lsu.edu/\sim adkins/m7211/AWchap3.pdf This follows from the characterization of projective modules as direct summands of free modules. Over principal ideal domains (PIDs) such as \mathbb{Z} or k (where k is a field), all projective modules are free.https://sites.math.rutgers.edu/\sim weibel/Kbook/Kbook.I.pdf In particular, principal ideals in these rings are free of rank 1 and hence projective; for instance, the ideal $2\mathbb{Z} in \mathbb{Z} is isomorphic to \mathbb{Z} as a \mathbb{Z}-module. Finite-dimensional vector spaces over a field k are free k-modules of finite rank and thus projective.https://www.math.lsu.edu/\sim adkins/m7211/AWchap3.pdf More generally, over a division ring D, every module (regarded as a "vector space" over D) admits a basis and is therefore free, making all such modules projective.https://faculty.etsu.edu/gardnerr/5410/notes/IV-2.pdf A classic non-example is \mathbb{Z}/2\mathbb{Z} as a \mathbb{Z}-module, which is not projective. The surjection \mathbb{Z} \to \mathbb{Z}/2\mathbb{Z} does not admit a section (splitting map), violating the lifting property: while there are two distinct endomorphisms of \mathbb{Z}/2\mathbb{Z}, there is only one \mathbb{Z}-module homomorphism from \mathbb{Z}/2\mathbb{Z} to \mathbb{Z}, so no lift exists for the identity map on \mathbb{Z}/2\mathbb{Z}.https://sites.math.rutgers.edu/courses/551/551-f06/proj.pdf Over the polynomial ring R = k[x,y] (with k a field), the module R itself is projective as the free module of rank 1.https://sites.math.rutgers.edu/\sim weibel/Kbook/Kbook.I.pdf However, non-principal ideals like (x,y) are not projective; this ideal is finitely generated but not free (tensoring with k = R/(x,y) yields a relation showing it lacks a basis), and by the Quillen-Suslin theorem, all finitely generated projective R-modules are free.

https://people.brandeis.edu/\sim igusa/Math101aF07/Math101a_notesB6.pdf$$$$https://sites.math.rutgers.edu/\sim weibel/Kbook/Kbook.I.pdf

Fundamental Properties

A key property of projective modules is their closure under direct sums. Specifically, the direct sum of any family of projective R-modules is again projective. This follows from the characterization of projective modules as direct summands of free modules: if each P_i is a direct summand of a free module F_i, then \bigoplus P_i is a direct summand of \bigoplus F_i, which is free.^[1] Projective modules are precisely the direct summands of free modules. In particular, a finitely generated projective module over any ring R is a direct summand of a free module of finite rank. This equivalence provides a fundamental structural description, linking projectivity to the existence of a complementary submodule in a free module via inclusion and projection maps.^[1] Over a Noetherian ring R, every projective R-module is a direct sum of finitely generated projective submodules. This result, due to H. Bass, ensures that infinite projective modules decompose into countably many finitely generated components, leveraging the ascending chain condition on ideals to control generation.^[9] Over an Artinian ring R, finitely generated projective R-modules have finite length. Since Artinian rings are Noetherian and have finite length as modules over themselves, finitely generated free modules inherit this property, and finitely generated projectives as their direct summands do as well.^[10] The tensor product of a projective module with a flat module is not necessarily projective. A counterexample occurs over the ring \mathbb{Z}, where \mathbb{Z} is projective and \mathbb{Q} is flat, but \mathbb{Z} \otimes_{\mathbb{Z}} \mathbb{Q} \cong \mathbb{Q} is flat yet not projective.^[11] If P is a projective R-module, then its endomorphism ring \operatorname{End}_R(P) contains the identity as a unit idempotent, which corresponds to the projection onto P in the decomposition of a containing free module F = P \oplus Q. This idempotent lifts the summand structure algebraically within the endomorphisms.^[3]

Projective versus Free Modules

Free modules over a ring R are those isomorphic to R^{(I)} for some index set I, meaning they admit a basis consisting of elements that generate the module freely. Projective modules, on the other hand, are precisely the direct summands of free modules; that is, a module P is projective if there exists a free module F and another module Q such that F \cong P \oplus Q.^[12] This direct summand characterization links projectives to frees but allows for a broader class, as not every projective module possesses a basis.^[2] Over certain rings, the distinction vanishes. For instance, if R is a local ring, every finitely generated projective R-module is free; this follows from Nakayama's lemma applied to the structure of projective modules as direct summands of free modules, where the maximal ideal acts trivially on the basis after tensoring with the residue field.^[12] A concrete example of a non-free projective module arises over the product ring R = S \times T, where S and T are nonzero rings: the modules S \times 0 and $0 \times T are projective (as direct summands of the free module R) but neither is free, since they lack a basis spanning the full rank.^[3] A ring R is semisimple Artinian if and only if every left (or right) R-module is projective; in such rings, every module decomposes as a direct sum of simple projective modules and is also injective.^[13] However, these modules are not necessarily free, as the simple modules over a semisimple ring (which is a finite product of matrix rings over division rings) generally do not admit bases unless the ring is a direct product of division rings.^[13] The lifting properties highlight a core difference: projective modules satisfy the lifting property with respect to surjective homomorphisms (i.e., for any surjection f: M \twoheadrightarrow N and map \phi: P \to N, there exists \psi: P \to M with f \circ \psi = \phi), but free modules extend this universally via their basis, allowing explicit constructions of lifts that preserve the free structure. In algebraic K-theory, Bass's stable range condition quantifies when projectives "stabilize" to free modules: if the stable range \mathrm{sr}(R) \leq n, then any projective module P such that P \oplus R^m is free for m \geq n is itself free, connecting projective stability to the structure of the Grothendieck group K_0(R).^[14]

Projective versus Flat Modules

A module M over a ring R is called flat if the functor -\otimes_R M preserves exact sequences, meaning that for any short exact sequence $0 \to A \to B \to C \to 0, the sequence $0 \to A \otimes_R M \to B \otimes_R M \to C \otimes_R M \to 0 remains exact. In contrast, a module P is projective if the functor \mathrm{Hom}_R(P, -) preserves exact sequences, or equivalently, if every surjection onto P splits.^[1] Projective modules form a subclass of flat modules. If P is projective, then P is flat, as the lifting property of projective modules ensures that tensoring with P preserves monomorphisms, and by dimension shifting, the full exactness follows; this implication can also be derived using the tensor-hom adjunction.^[15] Specifically, the exactness of \mathrm{Hom}_R(P, -) implies the exactness of P \otimes_R - via the natural isomorphism \mathrm{Hom}_R(P \otimes_R N, K) \cong \mathrm{Hom}_R(P, \mathrm{Hom}_R(N, K)).^[6] The converse implication does not hold: there exist flat modules that are not projective. A standard example is the rationals \mathbb{Q} as a \mathbb{Z}-module, which is flat since it is torsion-free over the principal ideal domain \mathbb{Z}, but not projective because it lacks a basis over \mathbb{Z} and is not a direct summand of any free \mathbb{Z}-module.^[6] Over a principal ideal domain R, flat modules are precisely the torsion-free modules, while projective modules are precisely the free modules.^[16] This distinction highlights that torsion-freeness suffices for flatness in this setting, but projectivity requires the stronger structural property of being free. In particular, over commutative rings, every finitely generated flat module is projective.^[17] Homologically, flatness of a module F is detected by the vanishing of all higher Tor groups, i.e., \mathrm{Tor}_i^R(N, F) = 0 for all i > 0 and all R-modules N. For a projective module P, this holds, and in particular \mathrm{Tor}_1^R(R/I, P) = 0 for every ideal I, reflecting the exactness of tensoring the presentation $0 \to I \to R \to R/I \to 0 with P. However, the flatness condition requires the full Tor vanishing across all dimensions, whereas projectivity imposes the additional requirement that \mathrm{Hom}_R(P, -) is exact, distinguishing it via the dual functorial property.^[6]

Projective versus Injective Modules

In homological algebra, projective and injective modules form a dual pair of concepts. A left R-module P is projective if, for every surjective R-module homomorphism f: M \twoheadrightarrow N and every R-module homomorphism g: P \to N, there exists an R-module homomorphism h: P \to M such that f \circ h = g; this is the lifting property with respect to surjections. Dually, a left R-module I is injective if, for every injective R-module homomorphism i: K \hookrightarrow L and every R-module homomorphism j: K \to I, there exists an R-module homomorphism k: L \to I such that k \circ i = j; this is the extension property with respect to injections.^[18] The extension property for injective modules is equivalently characterized by Baer's criterion: an R-module I is injective if and only if, for every left ideal \mathfrak{a} of R and every R-module homomorphism \phi: \mathfrak{a} \to I, there exists an R-module homomorphism \psi: R \to I such that \psi|_{\mathfrak{a}} = \phi.^[19] Over general rings, projective and injective modules are distinct classes, and this distinction persists even over Noetherian rings, where they do not coincide in general. For instance, over the ring of integers \mathbb{Z}, the module \mathbb{Z} is projective (being free of rank 1) but not injective, as it fails to be divisible—for example, there is no element x \in \mathbb{Z} such that $2x = 1. Conversely, the quotient module \mathbb{Q}/\mathbb{Z} is injective (as an injective hull of \mathbb{Z}/n\mathbb{Z} for all n) but not projective, since it is torsion and hence cannot be a direct summand of a free \mathbb{Z}-module.^[18] A ring R is semisimple if and only if every R-module is both projective and injective; in this case, every module is semisimple and decomposes as a direct sum of simple modules.^[15] The duality between projective and injective modules manifests in their interaction via Hom-spaces and exact sequences: specifically, for a projective module P and an injective module I, every short exact sequence of the form

$0 \to I \to M \to P \to 0

splits, meaning M \cong I \oplus P as R-modules. This splitting follows from the projectivity of P, as the surjection M \twoheadrightarrow P admits a section given by lifting the identity map on P. Dually, it follows from the injectivity of I, as the injection I \hookrightarrow M admits a retraction.^[18] In terms of homological applications, projective modules are employed to construct projective resolutions, which approximate modules from the left in chain complexes for computing right derived functors like \operatorname{Ext}, while injective modules are used for injective resolutions, approximating from the right for left derived functors like \operatorname{Tor}. This left-right asymmetry underscores their complementary roles in homological duality.^[18]

Homological and Categorical Aspects

Projective Resolutions

A projective resolution of a module M over a ring R is a long exact sequence of R-modules

\dots \to P_1 \xrightarrow{d_1} P_0 \xrightarrow{\epsilon} M \to 0,

where each P_i is a projective R-module, the sequence is exact at each P_i (i.e., \ker d_i = \operatorname{im} d_{i+1}), and \epsilon is surjective.^[20] For finitely presented modules, a projective resolution can be constructed inductively by starting with a free cover F_0 of M generated by a minimal set of generators, then taking the first syzygy module \operatorname{Syz}_1(M) = \ker(F_0 \to M) and covering it with a free module F_1, and continuing this process with higher syzygies \operatorname{Syz}_{i+1}(M) = \ker(F_i \to F_{i-1}) = \operatorname{im}(F_{i+1} \to F_i) to build the resolution \dots \to F_2 \to F_1 \to F_0 \to M \to 0.^[21] This yields a minimal free resolution when the differentials map into the maximal ideal times the target module in a local setting.^[21] A canonical example of a projective resolution arises from the Koszul complex over a polynomial ring. For R = k[x_1, \dots, x_n] where k is a field and x_1, \dots, x_n form a regular sequence, the Koszul complex K(\mathbf{x}) is an exact sequence of free modules resolving the residue field k = R/(\mathbf{x}), with terms \wedge^i R^n and differentials defined by wedging with the x_j. In the simple case n=1, R = k, the resolution of k is

$0 \to R \xrightarrow{\cdot x} R \to k \to 0.

^[22] Projective resolutions are fundamental for computing derived functors such as \operatorname{Ext}. Specifically, for modules M and N, the groups \operatorname{Ext}^i_R(M, N) are the cohomology groups H^i(\operatorname{Hom}_R(P_\bullet, N)) of the chain complex obtained by applying \operatorname{Hom}_R(-, N) to a projective resolution P_\bullet \to M \to 0 of M (deleting M); these groups classify extensions of N by M up to equivalence in degree 1 and higher obstructions.^[23] Every R-module M admits a projective resolution, constructed inductively by successively mapping free modules onto the kernels of previous maps.^[20] The length of a projective resolution of M, or projective dimension \operatorname{pd}_R(M), is the minimal length of such a resolution (possibly infinite). The global dimension \operatorname{gl.dim}(R) of the ring R is then the supremum of \operatorname{pd}_R(M) over all R-modules M, measuring the homological complexity of the category of R-modules.^[24]

Category of Projective Modules

The category of projective modules over a ring R, denoted \mathrm{Proj}-R, is the full subcategory of the category of all R-modules, \mathrm{Mod}-R, whose objects are the projective R-modules and whose morphisms are the R-module homomorphisms between projective modules.^[25]^[26] As a full subcategory of the additive category \mathrm{Mod}-R, \mathrm{Proj}-R is itself additive, inheriting the direct sum (biproduct) structure from \mathrm{Mod}-R.^[25] However, \mathrm{Proj}-R is not abelian in general, since the kernel or image of a morphism between projective modules need not be projective.^[25] Morphisms in \mathrm{Proj}-R do admit kernels and cokernels, computed as in \mathrm{Mod}-R, but these may lie outside \mathrm{Proj}-R.^[25] The category \mathrm{Proj}-R arises as the idempotent completion of the category of free R-modules: every projective module is a direct summand of a free module, and adjoining images of idempotents to the free modules yields precisely the projectives.^[27] The inclusion functor from the category of free R-modules to \mathrm{Proj}-R is full and faithful.^[27] Since \mathrm{Mod}-R is idempotent complete, so is \mathrm{Proj}-R: every idempotent endomorphism in \mathrm{Proj}-R splits, yielding a direct sum decomposition of the underlying projective module into the images of the idempotent and its complement.^[28] As a concrete example, if R = [k](/page/K) is a field, then every [k](/page/K)-module (vector space) is free and hence projective, so \mathrm{Proj}-[k](/page/K) coincides with the category \mathrm{Vect}_k of vector spaces over [k](/page/K); this is a semisimple abelian category, in which every short exact sequence splits.^[29]

Projective Modules over Special Rings

Over Commutative Rings

When the base ring R is commutative, finitely generated projective modules exhibit symmetric behavior and possess a well-defined rank function. The rank of a finitely generated projective R-module P is given by the locally constant function \mathrm{rk}_P: \mathrm{Spec}(R) \to \mathbb{N} \cup \{0\}, where \mathrm{rk}_P(\mathfrak{p}) equals the unique integer n such that P_\mathfrak{p} \cong R_\mathfrak{p}^n as R_\mathfrak{p}-modules. This rank can be characterized algebraically via the zeroth Fitting ideal \mathrm{Fit}_0(P), which for projective P is generated by the n \times n minors of a presentation matrix when n = \mathrm{rk}(P).^[30] A fundamental local-global principle holds: a finitely generated R-module P is projective if and only if P_\mathfrak{p} is a free R_\mathfrak{p}-module for every prime ideal \mathfrak{p} \subset R. This equivalence relies on the fact that over local rings, finitely generated projective modules are free, a result provable using Nakayama's lemma,^[31] and the dual basis lemma, which states that P admits elements \{x_i\}_{i=1}^n and \{f_j\}_{j=1}^n \subset \mathrm{Hom}_R(P, R) such that \sum f_j(x_i) = \delta_{ij}.^[17] Non-free examples abound over commutative rings. For instance, over a Dedekind domain R that is not a principal ideal domain, such as R = \mathbb{Z}[\sqrt{-5}], the ideal I = (2, 1 + \sqrt{-5}) is a rank-1 projective module but not free, as it is non-principal. Serre's conjecture posited that every finitely generated projective module over the polynomial ring k[x_1, \dots, x_n], where k is a field, is free; this was affirmatively resolved independently by Quillen and Suslin. Over more general commutative rings, projective modules need not be free, though they are often stably free, meaning P \oplus R^m \cong R^{r+m} for some m, r > 0. The cancellation problem inquires whether P \oplus R \cong Q \oplus R implies P \cong Q for projective modules P, Q. In general, no, particularly in infinite-rank cases where the Eilenberg-Mazur swindle demonstrates that all countably infinite free modules are isomorphic after stabilization, rendering cancellation trivial but uninformative.^[32] For finite rank, counterexamples exist over certain commutative rings, highlighting the subtlety of isomorphism classes.

Over Polynomial Rings

Polynomial rings, such as R = k[x_1, \dots, x_n] where k is a field or more generally A for a commutative ring A, provide a key setting for studying projective modules due to their nice homological properties. A landmark result in this area is the Quillen-Suslin theorem, which asserts that every finitely generated projective module over k[x_1, \dots, x_n] is free.^[33] This theorem resolves Serre's conjecture affirmatively for polynomial rings over fields and was proved independently by Daniel Quillen and Andrei Suslin in 1976. The proof by Quillen employs algebraic K-theory, relating the freeness of projectives to the generation of the special linear group by elementary matrices and incorporating Steinberg relations in the second K-group K_2(R). Suslin's approach is more algebraic, using homological methods to establish the result. As a consequence, stably free modules over these rings—those isomorphic to a free module after direct sum with a free module—are also free, since all projectives are free. In contrast, over coordinate rings of certain real affine varieties, non-free stably free projective modules exist. For instance, the coordinate ring of the 2-sphere, \mathbb{R}[x,y,z]/(x^2 + y^2 + z^2 - 1), admits a rank-2 stably free module corresponding to the tangent bundle, which is stably trivial but not free. Cancellation for projective modules also holds over polynomial rings, as established by Bass's theorem: if P and Q are projective R-modules with P \oplus R^m \cong Q \oplus R^m for some m, and the rank of P exceeds the Krull dimension of R, then P \cong Q. This follows from the general Bass cancellation result for noetherian rings, specialized to the dimension n of k[x_1, \dots, x_n]. The Quillen-Suslin theorem has significant implications for invariant theory, particularly in addressing aspects of Hilbert's 14th problem concerning the finite generation of rings of invariants under linear algebraic group actions, by ensuring free syzygies in resolutions over polynomial rings.

Geometric and Local Interpretations

Locally Free Sheaves and Vector Bundles

In algebraic geometry, there is a fundamental correspondence between finitely generated projective modules over a commutative ring R and locally free sheaves on the affine scheme \operatorname{Spec}(R). Specifically, given a finitely generated projective R-module P, the associated quasicoherent sheaf \tilde{P} on \operatorname{Spec}(R) is locally free, with rank at each prime ideal \mathfrak{p} \in \operatorname{Spec}(R) equal to the rank of the localization P_\mathfrak{p}.^[3] Conversely, the global sections of a locally free sheaf of finite rank on \operatorname{Spec}(R) recover the original projective module.^[3] This equivalence, known as the algebraic Serre-Swan theorem, establishes a dictionary between algebraic projectivity and geometric local freeness in the affine setting.^[34] More generally, a vector bundle E on a scheme X is defined as a locally free sheaf of \mathcal{O}_X-modules of finite constant rank, meaning that on an open cover \{U_i\} of X, E|_{U_i} \cong \mathcal{O}_{U_i}^r for some fixed r, with transition functions in \operatorname{GL}_r(\mathcal{O}_{U_i \cap U_j}).^[3] Over affine open subsets U = \operatorname{Spec}(A) \subset X, the restriction E|_U corresponds to a finitely generated projective A-module given by its sections \Gamma(U, E).^[3] This local correspondence glues to yield the global structure of the vector bundle, bridging module theory with sheaf theory on schemes. A key result characterizes when the global sections of a vector bundle form a projective module: for a vector bundle E on a scheme X, the module \Gamma(X, E) is projective over \Gamma(X, \mathcal{O}_X) if and only if E is locally free of constant rank.^[3] This holds particularly for projective schemes, where the constant rank condition ensures that the projective module captures the bundle's trivializations without torsion or varying dimensions.^[3] A concrete example arises with the tangent bundle T_{\mathbb{P}^n} on projective space \mathbb{P}^n_k over an algebraically closed field k. The global sections \Gamma(\mathbb{P}^n_k, T_{\mathbb{P}^n_k}) form a finitely generated projective module over the homogeneous coordinate ring k[x_0, \dots, x_n], but this module is non-free for n \geq 2, reflecting the nontrivial topology of the bundle via the Euler sequence $0 \to \mathcal{O} \to \mathcal{O}(1)^{n+1} \to T_{\mathbb{P}^n_k} \to 0.^[3] On projective varieties, Serre's theorem provides a deeper link: a coherent sheaf \mathcal{F} on a projective variety X = \operatorname{Proj}(S), where S is a finitely generated graded algebra over a field, is locally free if and only if the associated graded S-module of global sections is projective.^[35] This equivalence extends the affine correspondence to the projective setting, allowing projective modules to model geometric vector bundles via homogenization.^[35] This connection extends to K-theory, where the Grothendieck group K^0(R) of finitely generated projective R-modules is isomorphic to the Grothendieck group of vector bundles on \operatorname{Spec}(R), with the isomorphism induced by the sheafification functor \tilde{\cdot}.^[36] In the scheme-theoretic setting, this identifies algebraic K-theory with the K-theory of coherent sheaves, emphasizing the role of projectives in both realms.^[36]

Rank and Dimension Theory

For a projective module P over an integral domain R with field of fractions K, the rank of P is defined to be \mathrm{rk}(P) = \dim_K (P \otimes_R K).^[37] This dimension is a non-negative integer, as projectives over domains are torsion-free and of constant rank.^[3] More generally, over a commutative ring R, the local rank of P at a prime ideal \mathfrak{p} \in \mathrm{Spec}(R) is the rank of the free R_\mathfrak{p}-module P_\mathfrak{p}, given by \dim_{k(\mathfrak{p})} (P \otimes_R k(\mathfrak{p})), where k(\mathfrak{p}) = \mathrm{Frac}(R/\mathfrak{p}).^[3] This defines a rank function r_P: \mathrm{Spec}(R) \to \mathbb{Z}_{\geq 0}, \mathfrak{p} \mapsto \mathrm{rk}_{R_\mathfrak{p}}(P_\mathfrak{p}), which is locally constant with respect to the Zariski topology on \mathrm{Spec}(R).^[3] Thus, r_P is constant on each connected component of \mathrm{Spec}(R); in particular, over an integral domain (where \mathrm{Spec}(R) is connected), the rank function is constant and equals the generic rank \mathrm{rk}(P).^[3] Over a Dedekind domain R, finitely generated projective modules admit a complete classification involving the rank function. Specifically, every such module P of rank n \geq 1 is isomorphic to R^{n-1} \oplus I for a unique (up to isomorphism) invertible ideal I \subseteq R.^[38] Two such modules are isomorphic if and only if they have the same rank n (i.e., the same constant rank function) and the corresponding ideals belong to the same class in the ideal class group \mathrm{Cl}(R), known as the Steinitz invariant.^[38] The Krull dimension of the endomorphism ring \mathrm{End}_R(P) for a projective P of constant rank n over a commutative ring R of Krull dimension d is d. This follows from the fact that \mathrm{End}_R(P)_\mathfrak{p} \cong M_n(R_\mathfrak{p}) locally at each prime \mathfrak{p}, and the Krull dimension of a matrix ring M_n(A) over a commutative ring A equals the Krull dimension of A. Fitting ideals provide an algebraic tool to determine the rank from a presentation of P. Given a finite presentation F_1 \to F_0 \to P \to 0 with \mathrm{rk}(F_0) = m, the i-th Fitting ideal \mathrm{Fit}_i(P) is the ideal generated by the (m - i)-minors of the presentation matrix.^[39] For a projective P of rank n, \mathrm{Fit}_{m-n}(P) = R and \mathrm{Fit}_{m-n-1}(P) = 0, so the rank is the largest k such that \mathrm{Fit}_{m-k}(P) = R.^[39] Over a discrete valuation ring (DVR), which is a local Dedekind domain with trivial class group, every finitely generated projective module is free, and thus classified up to isomorphism solely by its rank.^[38] For instance, projectives over the p-adic integers \mathbb{Z}_p are free \mathbb{Z}_p-modules of rank equal to \dim_{\mathbb{Q}_p} (P \otimes_{\mathbb{Z}_p} \mathbb{Q}_p).^[38] In algebraic K-theory, the rank of a projective module P induces its image under the rank homomorphism \mathrm{rk}: K_0(R) \to \mathbb{Z}, where [P] \mapsto n = \mathrm{rk}(P).^[3] This rank determines the zeroth Chern class in the Chern character decomposition of [P] in the rationalized K-theory ring, providing a connection to higher invariants like Todd classes in index theory.^[3] Geometrically, constant-rank projective modules correspond to vector bundles on \mathrm{Spec}(R) whose rank matches the algebraic rank function.^[3]

References

[1]
Section 10.77 (05CD): Projective modules—The Stacks project
10.77 Projective modules. Some lemmas on projective modules. Definition 10.77.1. Let R be a ring. An R-module P is projective if and only if the functor ...
[2]
[PDF] 43 Projective modules
43.2 Definition. An R-module P is a projective module if there exists an R- module Q such that P ⊕ Q is a free R-module. 43.3 Examples. 1) If R is a ring with ...
[3]
[PDF] projective modules and vector bundles
The basic objects studied in algebraic K-theory are projective modules over a ring, and vector bundles over schemes. In this first chapter we introduce the ...
[4]
[PDF] Homological Algebra
WIDDER. Page 9. HOMOLOGICAL. ALGEBRA. By. HENRI CARTAN and. SAMUEL EILENBERG. PRINCETON, NEW JERSEY. PRINCETON UNIVERSITY PRESS. 1956. Page 10. Published, 7955, ...
[5]
[PDF] Some module theory - Purdue Math
A finitely generated projective module over commutative noethe- rian local ring is free. Proof. Let (R, m) be a comm. noeth. local ring, and P a fin. gen.
[6]
[PDF] Commutative Algebra
The following are equivalent: (1) M is projective as an R-module;. (2) For any short exact sequence. 0 −→ A −→ B −→ M −→ 0, the sequence splits, i.e. there is ...
[7]
[PDF] Notes on Homological Algebra - Mathematics
Jan 15, 2008 · Hom functor is exact. Definition 1.8.1 A left R-module I is called ... of a projective module (Definition 1.7.1). Definition 4.2.1 An ...
[8]
[PDF] when every projective module is a direct sum of finitely generated ...
We characterize rings over which every projective module is a direct sum of finitely generated modules, and give various examples of rings with and without this ...<|control11|><|separator|>
[9]
Section 10.53 (00J4): Artinian rings—The Stacks project
A ring R is Artinian if and only if it has finite length as a module over itself. Any such ring R is both Artinian and Noetherian, any prime ideal of R is a ...
[10]
Flat Module -- from Wolfram MathWorld
Z -module is flat iff it is torsion-free: hence Q and the infinite direct product Z×Z×... are flat Z -modules, but they are not projective. In fact, over a ...
[11]
Section 10.78 (00NV): Finite projective modules—The Stacks project
By Lemma 10.78.2 being finite projective over a local ring is the same thing as being finite free. Suppose that M \otimes _ R S is a finite free S-module.
[12]
[PDF] NONCOMMUTATIVE RINGS 1. Semisimplicity Let A be a (not ...
In general, modules with this property are called projective. Hence, every module over a semi-simple ring is projective. Projective modules share a mapping ...
[13]
[PDF] K-theory and stable algebra - Numdam
This paper by H. Bass discusses K-theory and stable algebra, covering stable structures of linear groups, projective modules, and the functors K.
[14]
Lectures on Modules and Rings
### Summary of Projective Modules and Related Concepts from "Lectures on Modules and Rings"
[15]
Lemma 15.125.9 (0ASV)—The Stacks project
Although this is pretty trivial, it may be worthwhile saying in the statement that, in particular, every finitely generated projective module over a PID is free ...
[16]
An Introduction to Homological Algebra
The landscape of homological algebra has evolved over the last half-century into a fundamental tool for the working mathematician. This book provides a ...
[17]
Lemma 19.2.6 (05NU): Baer's criterion—The Stacks project
an open source textbook and reference work on algebraic geometry.
[18]
[PDF] A Beginner's Guide to Homological Algebra - arXiv
Sep 4, 2022 · In this paper, we seek to pro- vide an introductory guide for advanced students of mathematics and related specialties seeking to get started on ...
[19]
[PDF] 1. Introduction to Syzygies - Cornell Mathematics
Motivated by applications in. Invariant Theory, he introduced the idea of associating a free resolution to a finitely generated module in a famous paper in 1890 ...
[20]
[PDF] Notes on Koszul complexes and commutative algebra
The Hilbert syzygy theorem is concerned with free resolutions of graded modules over polynomial rings R = k[x1,...,xn]. The proof shows easily the power of the ...
[21]
[PDF] Chapter 1 uick Homological Algebra
We compute Ext¯R(M, N) by taking a projective resolution of M, applying the functor HomR ( - , N), then taking homology. First, a projective R-module P is a ...
[22]
10.109 Rings of finite global dimension - Stacks Project
Jun 13, 2014 · The ring R is said to have finite global dimension if there exists an integer n such that every R-module has a resolution by projective R- ...
[23]
[PDF] Introduction to abelian and derived categories.
If R is a ring, the category Mod R of (right) R-modules is an additive category. ... simple, the category Proj R of projective modules over R is not abelian since ...
[24]
[PDF] An Introduction to Morita Theory - Matt Booth
Write Proj-R for the subcategory of projective modules, and proj-R for the subcategory of finitely generated projective modules. Two rings R, S are defined ...
[25]
The homotopy category of complexes of projective modules
Aug 6, 2025 · Our work also relates to seminal work on the homotopy category of projective modules, first studied in [Jør05, Nee08], and from the model ...
[26]
http://mattbooth.info/notes/morita.pdf
[27]
Section 9.4 (09FM): Vector spaces—The Stacks project
Vector spaces, or k-modules over a field, are simple to study. Every k-module is free, and every exact sequence of modules over a field splits.
[28]
Fitting ideals and various notions of equivalence for modules
Jan 5, 2023 · We develop a certain notion of “equivalence of modules” over suitable commutative rings R. We establish that understanding the equivalence of R-modules is ...
[29]
10.85 Projective modules over a local ring - Stacks Project
Let R be a ring. Then every projective R-module is free if and only if every countably generated projective R-module is free. Proof. Follows immediately ...
[30]
Example 15.70.3 (0EX0)—The Stacks project
Example 15.70.3. Let R be a Dedekind domain. Then every nonzero ideal I is a finite projective module, see Lemma 15.22.11. Thus R/I has projective dimension 1.
[31]
Mazur's swindle | What's new - Terence Tao
Oct 5, 2009 · This reminds me of “Eilenberg's Trick” from ring theory: If P is projective, there is a free module F with P \oplus F = F. The idea is to ...
[32]
The cancellation problem for projective modules and related topics
The cancellation problem for projective modules and related topics ... To read the full-text of this research, you can request a copy directly from the author.
[33]
Projective modules over polynomial rings | Inventiones mathematicae
Projective modules over polynomial rings ... Article PDF. Download to read the full article text. Use our pre-submission checklist.
[34]
Serre's problem on projective modules, by T. Y. Lam, Springer ...
Jan 7, 2008 · Let v = (v1,...,vn) be a unimodular row over the polynomial ring R = k[x1,...,xn]. Is v the first row of an invertible matrix over. R?
[35]
[PDF] VECTOR BUNDLES AND PROJECTIVE MODULES - UNL Math
A number of examples of projective modules have been given using this correspondence. However, no rigorous treatment of the correspondence seems to have been ...
[36]
[PDF] FOUNDATIONS OF ALGEBRAIC GEOMETRY CLASS 30
Feb 4, 2008 · Generation by global sections, and Serre's Theorem. 3. 4. Every quasicoherent sheaf on a projective A-scheme arises from a graded module. 6.
[37]
[PDF] the grothendieck group k0
Both K0(R) and K0(X) are formed by taking the group completion of an abelian monoid—the monoid P(R) of finitely generated projective R-modules and the monoid VB ...
[38]
Rank of a module - ac.commutative algebra - MathOverflow
Jun 30, 2010 · Consider a projective module P of finite type over a commutative ring A. It corresponds to a locally free sheaf F over X=Spec(A). The rank of F ...Is a projective module of constant finite rank finitely generated?Different definitions of the rank of a module - MathOverflowMore results from mathoverflow.net
[39]
[PDF] Ideal classes and relative integers - Keith Conrad
In particular, when A is a Dedekind domain all fractional A-ideals are projective A-modules. xiyi = x. Check the A-linear map h: M → a⊕kerf given by the ...
[40]
Section 15.8 (07Z6): Fitting ideals—The Stacks project
The Fitting ideals of a finite module are the ideals determined by the construction of Lemma 15.8.2.