Nakayama’s Lemma in the Noncommutative Setting

Q: Does the lemma hold for right modules?

Yes, verbatim with *left* replaced by *right* throughout. This is because rad R is left-right symmetric, so the hypothesis J ⊆ rad R means the same thing on either side. Note that J itself must then be a right ideal and MJ the relevant product.

Q: What is the relationship between Nakayama's Lemma and projective covers?

A projective cover of M is an epimorphism P → M with P projective and kernel superfluous in P. Nakayama supplies the standard source of superfluous submodules: for finitely generated P the submodule (rad R)P is superfluous, so a lift of a minimal generating set of the top produces a cover. Existence in general needs R semiperfect.

Q: Can I use the lemma to show that a surjection of finitely generated modules is an isomorphism?

Not directly. It gives surjectivity of f : M → N from surjectivity of the induced map on tops. Injectivity is a separate matter: it follows if M and N have the same finite length, or if M is finitely generated over a commutative ring and f : M → M is surjective, but not from Nakayama alone.

Q: Why is the lemma stated for a left ideal J rather than a two-sided ideal?

Because nothing in the argument uses two-sidedness: JM is a submodule as soon as J is a left ideal, and the containment J ⊆ rad R is the only structural hypothesis. Stating it for left ideals is strictly more general, and rad R is two-sided anyway.

Executive Summary

Nakayama's Lemma is the working tool of radical theory. It says that the radical cannot generate a finitely generated module out of nothing: if $J \subseteq rad R$ and $J M = M$ with $M$ finitely generated, then $M$ was zero to begin with. Equivalently, $J M$ is a superfluous submodule of $M$ .

The practical payoff is the reverse reading. To generate $M$ it is enough to generate the much simpler quotient $M / (rad R) M$ , a module over the radical-free ring $R / rad R$ . Every construction that lifts data across $R \to R / rad R$ — projective covers, minimal free resolutions, Wedderburn decompositions of finite-dimensional algebras — rests on this one lemma.

(4.22)Lam's numbering

3Equivalent forms

f.g.Indispensable hypothesis

c. 1951Azumaya–Nakayama

Overview

Let $R$ be a ring with identity and $M$ a left $R$ -module. For a left ideal $J \subseteq R$ , write $J M$ for the subgroup of finite sums $\sum_{i} j_{i} m_{i}$ with $j_{i} \in J$ , $m_{i} \in M$ ; it is a submodule of $M$ . The question Nakayama answers is when $J M$ can exhaust $M$ .

J M = M with M finitely generated and J \subseteq rad R ⟹ M = 0

(4.22)

Both hypotheses are needed. Drop finite generation and $ℚ$ over $ℤ_{(p)}$ is a counterexample; drop $J \subseteq rad R$ and $ℤ / 3$ over $ℤ$ with $J = (2)$ is one.

The reason such a statement is available at all is the unit characterisation of the radical established on Jacobson Radical Definition and Characterisations: elements of $rad R$ annihilate every simple module. A finitely generated nonzero module has a simple quotient; the radical kills it; so the radical cannot reach the top of the module.

Nothing here requires commutativity, and nothing requires $R$ to be local — only that the ideal you divide by sits inside the radical. That is the precise noncommutative generalisation of the classical local-ring statement.

Learning Objectives

State the three equivalent forms of $(4.22)$ for a left ideal $J \subseteq R$ .
Prove $(1) \Rightarrow (2) \Rightarrow (3) \Rightarrow (1)$ and locate the single use of finite generation.
Deduce the generation criterion: generators of $M / J M$ lift to generators of $M$ .
Show that over a local ring all minimal generating sets of a finitely generated module have the same cardinality.
Give counterexamples when $M$ is not finitely generated and when $J not \subseteq rad R$ .
Compute $M / (rad R) M$ for the natural module over upper triangular matrices and read off a generator.

Definitions

Definition—Product of an ideal and a module

For a left ideal $J \subseteq R$ and a left $R$ -module $M$ , $J M$ denotes the set of all finite sums $\sum_{i = 1}^{t} j_{i} m_{i}$ with $j_{i} \in J$ and $m_{i} \in M$ . Because $J$ is a left ideal, $J M$ is a submodule of $M$ .

Definition—Superfluous submodule

A submodule $N \subseteq M$ is superfluous (or small), written $N ≪ M$ , if for every submodule $L \subseteq M$ the equality $N + L = M$ forces $L = M$ . Nakayama's Lemma says exactly that $(rad R) M ≪ M$ whenever $M$ is finitely generated.

$rad M$: The radical of a module: the intersection of the maximal submodules of $M$ , taken to be $M$ itself if there are none. Always $(rad R) M \subseteq rad M$ .
Top of $M$: The quotient $M / (rad R) M$ , a module over $R / rad R$ . When $R$ is semilocal this is a semisimple module.
Finitely generated: $M = R x_{1} + \dots + R x_{n}$ for some finite list of elements. Equivalently $M$ is a quotient of a finite free module $R^{n}$ .
Local ring: A ring in which the non-units form an ideal; equivalently $R / rad R$ is a division ring. Not assumed commutative here.
Minimal generating set: A generating set no proper subset of which generates. Over a general ring these can have different sizes; over a local ring they cannot.

Rings have an identity and modules are unital. $J$ is a left ideal throughout — it is not assumed two-sided, although $rad R$ itself is.

Core Concepts

Finite generation buys a maximal submodule

The whole content of the lemma is the existence of a simple quotient. If $M \neq 0$ is finitely generated, say $M = R x_{1} + \dots + R x_{n}$ , then the poset of proper submodules of $M$ has maximal elements. Zorn's Lemma applies because the union $N$ of a chain of proper submodules is again proper: if $N = M$ then each $x_{j}$ lies in some member of the chain, and the largest of those $n$ members already contains every $x_{j}$ , hence equals $M$ — contradicting properness.

The radical kills every simple quotient

Given a maximal submodule $M' ⊊ M$ , the quotient $M / M'$ is a simple left $R$ -module, so $(rad R) (M / M') = 0$ . Since $J \subseteq rad R$ , this gives $J M \subseteq M' ⊊ M$ . The radical is therefore confined below the top of $M$ , which is the geometric content of the statement.

M \neq 0

f.g.maximal

M' ⊊ M

M / M'

simple

J M \subseteq M'

J M \neq M

Superfluity and the top

Form $\bar{M} = M / (rad R) M$ . This is a module over $\bar{R} = R / rad R$ , and when $R$ is semilocal — that is, $\bar{R}$ is semisimple — $\bar{M}$ is a semisimple module, so it decomposes into simples and its generators are visible. Nakayama says that lifting a generating set of $\bar{M}$ back to $M$ loses nothing.

The containment $(rad R) M \subseteq rad M$ holds for every module, since the radical annihilates each simple quotient $M / M'$ . Equality holds when $M$ is projective and when $R$ is semilocal, but not in general: for $R = ℤ$ and $M = ℚ$ the left side is $0$ and the right side is $ℚ$ .

Key Results

Theorem(4.22)Nakayama's Lemma

Let $R$ be a ring with identity and let $J \subseteq R$ be a left ideal. The following are equivalent.

$J \subseteq rad R$ .
For every finitely generated left $R$ -module $M$ , $J M = M$ implies $M = 0$ .
For all left $R$ -modules $N \subseteq M$ with $M / N$ finitely generated, $N + J M = M$ implies $N = M$ .

Proof

**(1) $\Rightarrow$ (2).** Suppose $M$ is finitely generated and $M \neq 0$ . By the Zorn argument above, $M$ has a maximal submodule $M'$ . Then $M / M'$ is simple, so $(rad R) (M / M') = 0$ , and since $J \subseteq rad R$ we get $J M \subseteq M'$ . In particular $J M \neq M$ . Contrapositively, $J M = M$ forces $M = 0$ .

**(2) $\Rightarrow$ (3).** Let $N \subseteq M$ with $M / N$ finitely generated and $N + J M = M$ . Apply (2) to $\bar{M} = M / N$ : the image of $J M$ in $\bar{M}$ is $J \bar{M}$ , and $N + J M = M$ says precisely $J \bar{M} = \bar{M}$ . Since $\bar{M}$ is finitely generated, $\bar{M} = 0$ , i.e. $N = M$ .

**(3) $\Rightarrow$ (1).** Suppose some $y \in J$ fails to lie in $rad R$ . Then $y \notin 𝔪$ for some maximal left ideal $𝔪$ , so the left ideal $𝔪 + J$ strictly contains $𝔪$ and hence equals $R$ . Take $M = R$ and $N = 𝔪$ ; the quotient $R / 𝔪$ is cyclic, hence finitely generated, and $N + J M = 𝔪 + J R \supseteq 𝔪 + J = R$ . Condition (3) now yields $𝔪 = R$ , contradicting maximality. Hence $J \subseteq rad R$ .

Corollary(4.22a)Generators lift from the top

Let $J \subseteq rad R$ be a left ideal, let $M$ be a finitely generated left $R$ -module, and let $x_{1}, \dots, x_{n} \in M$ . If the images ${\bar{x}}_{1}, \dots, {\bar{x}}_{n}$ generate $M / J M$ , then $x_{1}, \dots, x_{n}$ generate $M$ .

Proof

Put $N = R x_{1} + \dots + R x_{n}$ . The hypothesis says $(N + J M) / J M = M / J M$ , that is, $N + J M = M$ . The quotient $M / N$ is finitely generated, being a quotient of the finitely generated module $M$ , so form (3) of $(4.22)$ applies and gives $N = M$ .

Corollary(4.22b)Minimal generating sets over a local ring

Let $R$ be a local ring with $𝔪 = rad R$ and residue division ring $D = R / 𝔪$ , and let $M$ be a finitely generated left $R$ -module. Then $M / 𝔪 M$ is a finite-dimensional left $D$ -vector space, elements of $M$ generate $M$ if and only if their images span $M / 𝔪 M$ , and every minimal generating set of $M$ has exactly $\dim_{D} M / 𝔪 M$ elements.

Proof

$M / 𝔪 M$ is annihilated by $𝔪$ , hence is a $D$ -vector space, finite-dimensional because $M$ is finitely generated. One direction of the generation criterion is trivial and the other is $(4.22 a)$ . If $x_{1}, \dots, x_{n}$ is a minimal generating set and the images were $D$ -linearly dependent, some ${\bar{x}}_{i}$ would lie in the span of the others; the remaining $n - 1$ images would still span, so by $(4.22 a)$ the remaining $n - 1$ elements would generate $M$ , contradicting minimality. Hence the images form a basis and $n = \dim_{D} M / 𝔪 M$ .

Corollary(4.22c)Superfluity

For any finitely generated left $R$ -module $M$ , the submodule $(rad R) M$ is superfluous in $M$ . More generally, if $M / N$ is finitely generated then $N + (rad R) M = M$ implies $N = M$ .

Remark—Why no determinant trick

Over a commutative ring the lemma has a sharper form proved by the Cayley–Hamilton or determinant trick: if $M$ is finitely generated and $I M = M$ for an ideal $I$ , then $a M = 0$ for some $a \equiv 1 (\mod I)$ — no hypothesis on $I$ at all. That argument needs determinants of matrices over $R$ and has no honest noncommutative analogue, which is why $(4.22)$ trades it for the hypothesis $J \subseteq rad R$ and a Zorn argument.

Proof Techniques and Method

How the proof works, and the reusable move.

Move 1

Finite generation gives a maximal submodule

A chain of proper submodules of a finitely generated module has proper union, because finitely many generators cannot be spread over an unbounded chain. This is the only place the hypothesis is used.

Move 2

Test against a simple quotient

Once a maximal submodule $M'$ exists, $M / M'$ is simple and the radical annihilates it. Any statement of the form the radical cannot do X reduces to this.

Move 3

Quotient to reduce relative to absolute

Form (3) with $N$ present follows from form (2) by passing to $M / N$ . Whenever a lemma has an absolute and a relative version, check whether the relative one is just the absolute one applied to a quotient.

The converse direction, $(3) \Rightarrow (1)$ , is the one that shows the hypothesis is not merely convenient: taking $M = R$ and $N$ a maximal left ideal turns the module statement back into a statement about maximal left ideals, which is the definition of the radical. So $rad R$ is not just a left ideal for which Nakayama holds — it is the largest one.

Worked Example

The natural module over upper triangular matrices

Let $k$ be a field and $R = T_{2} (k)$ the ring of upper triangular $2 \times 2$ matrices. As computed on Jacobson Radical Definition and Characterisations, $J : = rad R$ is the set of strictly upper triangular matrices, and $R / J ≅ k \times k$ .

Take $M = k^{2}$ , columns, with the usual matrix action. Write $e_{1}, e_{2}$ for the standard basis. Then

J M = {(\begin{matrix} 0 & b \\ 0 & 0 \end{matrix}) (\begin{matrix} v_{1} \\ v_{2} \end{matrix}) : b, v_{1}, v_{2} \in k} = {(\begin{matrix} b v_{2} \\ 0 \end{matrix})} = k e_{1},

(E.1)

so $M / J M$ is one-dimensional, spanned by the image of $e_{2}$ . Corollary $(4.22 a)$ predicts that $e_{2}$ alone generates $M$ . Check it directly:

R e_{2} = {(\begin{matrix} a & b \\ 0 & c \end{matrix}) (\begin{matrix} 0 \\ 1 \end{matrix})} = {(\begin{matrix} b \\ c \end{matrix}) : b, c \in k} = M .

(E.2)

$M$ is cyclic, generated by $e_{2}$ , even though $M$ has $k$ -dimension $2$ .

Contrast $e_{1}$ , whose image in $M / J M$ is zero: $R e_{1} = k e_{1} ⊊ M$ . The top of the module, not the module itself, is what governs generation.

Both hypotheses are sharp

Finite generation cannot be dropped. Let $R = ℤ_{(p)}$ , the localisation of $ℤ$ at a prime $p$ ; it is local with $rad R = p R$ . Take $M = ℚ$ . Then $p M = ℚ = M$ but $M \neq 0$ . Of course $ℚ$ is not finitely generated over $ℤ_{(p)}$ .
** $J \subseteq rad R$ cannot be dropped.** Let $R = ℤ$ , so $rad R = 0$ , and take $J = (2)$ , $M = ℤ / 3$ . Then $M$ is finitely generated and $J M = 2 \cdot (ℤ / 3) = ℤ / 3 = M$ , yet $M \neq 0$ .
The two are independent. Neither counterexample can be repaired by strengthening the other hypothesis; $(3) \Rightarrow (1)$ shows the radical is exactly the boundary case.

Process and Workflow

Can I apply Nakayama here?

$M$ f.g., $J \subseteq rad R$ Yes. Use form (2) to conclude

M = 0

, or form (3) with

N = \sum R x_{i}

to conclude a candidate list generates.

$M$ not f.g., but $M / N$ isYes — form (3) only asks the quotient to be finitely generated. This is the version to reach for when

M

is large but the cokernel is small.

$J not \subseteq rad R$ No. If

R

is commutative, fall back on the determinant trick:

I M = M

with

M

f.g. gives

a \in 1 + I

with

a M = 0

. Noncommutatively there is no substitute.

Nothing is finitely generatedNo. Look for a grading bounded below or a complete filtration; otherwise the conclusion is simply false.

Pass to the topCompute

\bar{M} = M / (rad R) M

as a module over

\bar{R} = R / rad R

Solve the easy problemFind a generating set of

\bar{M}

. If

R

is semilocal,

\bar{M}

is semisimple and this is linear algebra over division rings.

LiftChoose any preimages

x_{1}, \dots, x_{n} \in M

. By

(4.22 a)

they generate

M

Check minimalityOver a local ring the lift is minimal exactly when the images form a basis of

\bar{M}

; then

n

is an invariant of

M

Comparison and Classification

Versions of Nakayama's Lemma across settings
Setting	Statement	What replaces finite generation
Commutative local $(R, 𝔪)$	$M$ f.g., $𝔪 M = M \Rightarrow M = 0$	nothing — $M$ must be f.g.
Commutative, arbitrary ideal $I$	$M$ f.g., $I M = M \Rightarrow a M = 0$ for some $a \in 1 + I$	the determinant trick removes the hypothesis on $I$
Noncommutative, $J \subseteq rad R$ a left ideal	$M$ f.g., $J M = M \Rightarrow M = 0$ — this is $(4.22)$	nothing — $M$ must be f.g.
Graded, $R = ⨁_{n \geq 0} R_{n}$	$M$ graded and bounded below, $R_{+} M = M \Rightarrow M = 0$	the grading: degrees are bounded below
Complete filtered rings	$M$ complete and separated for the $J$ -adic filtration	completeness replaces finiteness

Which conclusion is available under which hypotheses
	$M$ f.g.	$M$ arbitrary	$M$ graded, bounded below
$J \subseteq rad R$ : $J M = M \Rightarrow M = 0$	yes	no	partial
$J \subseteq rad R$ : $J M$ superfluous	yes	no	partial
Generators lift from $M / J M$	yes	no	yes
$(rad R) M = rad M$	partial	partial	partial

Which conclusion is available under which hypotheses

In the last row the equality is automatic for projective modules and for semilocal rings, and can fail otherwise.

Relationship Map

Nakayama sits between the elementary characterisations of the radical and the machinery of projective covers.

y \in rad R \Rightarrow 1 - x y \in U (R)

radical kills simplesNakayama

(4.22)

generators liftprojective covers

Nakayama's Lemma — $J \subseteq rad R$ , $M$ finitely generated
- Immediate corollaries
  - generators of $M / J M$ lift to generators of $M$
  - $(rad R) M$ is superfluous in $M$
  - a surjection $f : M \to N$ of f.g. modules is onto as soon as it is onto modulo the radical
- Structural consequences
  - minimal generating sets over a local ring have constant size
  - idempotent and unit lifting arguments in semiperfect rings
  - uniqueness of minimal free resolutions over local and graded rings
- Fails without
  - finite generation of $M$ (or of $M / N$ )
  - the containment $J \subseteq rad R$

Applications and Industry Use

Applications here means where this structure is used — inside mathematics and in the engineering and computing disciplines that consume it.

Module theory

Projective covers

A projective cover of $M$ is a surjection $P \to M$ with projective $P$ and superfluous kernel. Nakayama is what makes the kernel superfluous and hence what makes semiperfect and perfect rings work; see Projective Covers.

Computational algebra

Minimal generating sets

Systems that compute over local or graded rings — Macaulay2, Singular — reduce generation questions to linear algebra over the residue field, which is valid precisely by $(4.22 a)$ . A minimal generating set is read off from a basis of the top.

Representation theory

Lifting across the radical

For a finite-dimensional algebra $A$ , the projective indecomposables correspond bijectively to the simple $A / rad A$ -modules. Nakayama supplies the lifting half of that correspondence.

Commutative algebra

Fibre dimension and Krull intersection

Over a Noetherian local ring the minimal number of generators of a finitely generated module is $\dim_{D} M / 𝔪 M$ , and the Krull intersection theorem is a direct application of the lemma to the intersection of the powers of $𝔪$ .

The honest description is that Nakayama is infrastructure: it is almost never the theorem being proved, and it is used in the first three lines of a great many proofs that are.

Design Considerations

Design considerations here means the choices made when modelling a problem with these algebraic structures.

Choose the relative form. Form (3) with $M / N$ finitely generated is strictly more usable than form (2); state your lemma that way and you avoid needing $M$ itself to be small.
**Pick $J$ as large as you can.** Any left ideal inside $rad R$ works, but the strongest conclusion comes from $J = rad R$ . There is no gain from a smaller $J$ except when tracking a filtration.
Decide whether you need a local ring. The constant-size statement for minimal generating sets is genuinely local. Over a semilocal ring you still get lifting, but generating sets of different sizes can be minimal.
Watch which side. $M$ is a left module and $J$ a left ideal. Because $rad R$ is side-neutral the mirror statement holds for right modules, but $J M$ must be formed consistently.

Failure Modes and Common Mistakes

Do not conclude that a minimal generating set is a basis. Over a local ring its size is an invariant, but the relations among the generators need not vanish.
Do not assume $J M = M$ is preserved by localisation or completion without checking that finite generation survives.
Do not quote the determinant-trick form of the lemma in a noncommutative argument; there are no determinants.
Do not use the lemma to prove a surjection is an isomorphism. It gives surjectivity from surjectivity modulo the radical; injectivity needs a separate argument, usually finiteness of length.

Historical Notes and Lessons Learned

1930sKrull's ideal-theoretic formIn commutative local algebra, Krull uses the implication for ideals of a local ring, in effect the case where the module is itself an ideal.
1945The radical becomes availableJacobson's definition of the radical for arbitrary rings supplies the object that the noncommutative statement needs; before this there is no candidate for the ideal J.
c. 1951Azumaya and NakayamaThe module-theoretic formulations, including the relative version with a submodule N, are given by Azumaya and by Nakayama in their work on Frobenius and maximally central algebras.
1960Bass and projective coversBass's paper on perfect rings makes superfluous submodules and projective covers central, and the lemma becomes the standard first step in lifting arguments.
1962Nagata records the attributionIn Local Rings Nagata reports Nakayama's own suggestion that the result should be called Krull-Azumaya in the commutative case and Jacobson-Azumaya in the noncommutative one.

The name that stuck is the one Nakayama himself did not want. The lesson worth extracting is not about credit but about formulation: Krull's version was about ideals in a local ring, and it took the module-theoretic restatement to make the result apply to arbitrary rings and arbitrary finitely generated modules. Choosing the right category to state a lemma in is most of the work.

Quick Reference

Core statement

J \subseteq rad R

M

f.g.,

J M = M

\Rightarrow

M = 0

Relative form

M / N

f.g.,

N + J M = M

\Rightarrow

N = M

Working formimages generate

M / J M

\Rightarrow

elements generate

M

Superfluity

(rad R) M ≪ M

for

M

finitely generated

Local ringsminimal generating sets all have

\dim_{D} M / 𝔪 M

elements

Converseif the conclusion holds for all such

M

, then

J \subseteq rad R

Fails for

M = ℚ

over

ℤ_{(p)}

;

J = (2)

M = ℤ / 3

over

ℤ

Commutative bonusdeterminant trick:

I M = M \Rightarrow a M = 0

a \in 1 + I

Checklist before invoking the lemma
Question	Required answer	If not
Is $J$ inside $rad R$ ?	yes	no conclusion; try the determinant trick if $R$ is commutative
Is $M$ (or $M / N$ ) finitely generated?	yes	no conclusion; look for a grading or completeness
Are modules on the correct side?	$J$ left ideal, $M$ left module	mirror the whole statement
Do you need minimality?	only over a local ring	lifting still works, invariance of size does not

Frequently Asked Questions

Why does Nakayama's Lemma need finite generation when the commutative determinant-trick version seems not to?

The determinant trick also needs it — it applies to a finitely generated module $M$ and produces a matrix relation among a finite generating set. What it does not need is the hypothesis $I \subseteq rad R$ . So the two versions trade one hypothesis for another; finite generation is common to both.

Does the lemma hold for right modules?

Yes, verbatim with left replaced by right throughout. This is because $rad R$ is left-right symmetric, so the hypothesis $J \subseteq rad R$ means the same thing on either side. Note that $J$ itself must then be a right ideal and $M J$ the relevant product.

(4.22)

true if

J

is only assumed to be a nil ideal?

Yes, but only because every nil one-sided ideal is contained in $rad R$ by Lam's $(4.11)$ , so this is a special case rather than a generalisation. The same applies to nilpotent ideals and to the lower and upper nilradicals.

What is the relationship between Nakayama's Lemma and projective covers?

A projective cover of $M$ is an epimorphism $P \to M$ with $P$ projective and kernel superfluous in $P$ . Nakayama supplies the standard source of superfluous submodules: for finitely generated $P$ the submodule $(rad R) P$ is superfluous, so a lift of a minimal generating set of the top produces a cover. Existence in general needs $R$ semiperfect.

Can I use the lemma to show that a surjection of finitely generated modules is an isomorphism?

Not directly. It gives surjectivity of $f : M \to N$ from surjectivity of the induced map on tops. Injectivity is a separate matter: it follows if $M$ and $N$ have the same finite length, or if $M$ is finitely generated over a commutative ring and $f : M \to M$ is surjective, but not from Nakayama alone.

Why is the lemma stated for a left ideal

J

rather than a two-sided ideal?

Because nothing in the argument uses two-sidedness: $J M$ is a submodule as soon as $J$ is a left ideal, and the containment $J \subseteq rad R$ is the only structural hypothesis. Stating it for left ideals is strictly more general, and $rad R$ is two-sided anyway.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §4, statement (4.22).
M. Nagata, Local Rings, Interscience Tracts in Pure and Applied Mathematics 13, Interscience, 1962.
F. W. Anderson and K. R. Fuller, Rings and Categories of Modules, 2nd edition, Graduate Texts in Mathematics 13, Springer-Verlag, 1992, §15 and §17.
H. Bass, “Finitistic dimension and a homological generalization of semi-primary rings”, Transactions of the American Mathematical Society 95 (1960), 466–488.
H. Matsumura, Commutative Ring Theory, Cambridge Studies in Advanced Mathematics 8, Cambridge University Press, 1986, Chapter 1.
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964.

AI Suggested Questions

Work out the graded version of Nakayama's Lemma and show exactly why finite generation is unnecessary there.
Give a semilocal ring and a finitely generated module with two minimal generating sets of different cardinalities.
Prove the Krull intersection theorem from Nakayama's Lemma for a Noetherian commutative local ring.
How is Nakayama's Lemma used to establish the bijection between projective indecomposables and simple modules over a finite-dimensional algebra?
Show that a finitely generated projective module over a local ring is free, using the lifting corollary.
What is the correct statement of Nakayama's Lemma for rings without identity, where maximal left ideals may not exist?
Compare the determinant trick and the Zorn argument as proofs, and identify which generalises to noncommutative graded rings.

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Finite generation buys a maximal submodule

The radical kills every simple quotient

Superfluity and the top

06Key Results

07Proof Techniques and Method

Finite generation gives a maximal submodule

Test against a simple quotient

Quotient to reduce relative to absolute

08Worked Example

The natural module over upper triangular matrices

Both hypotheses are sharp

09Process and Workflow

10Comparison and Classification

11Relationship Map

12Applications and Industry Use

Projective covers

Minimal generating sets

Lifting across the radical

Fibre dimension and Krull intersection

13Design Considerations

14Failure Modes and Common Mistakes

15Historical Notes and Lessons Learned

16Quick Reference

17Frequently Asked Questions

18Related KEVOS Topics

19References

20AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Key Results

Proof Techniques and Method

Worked Example

Process and Workflow

Comparison and Classification

Relationship Map

Applications and Industry Use

Design Considerations

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions