Noetherian polynomial rings and finite-type algebras
Component notice and licence. Adapted from the exact 00FN statement and proof, with the elementary completions identified below. All rings here are commutative with a unit; homomorphisms preserve it. Natural-number induction and ordinary set-theoretic choice are the foundational conventions.
The statement
If every ideal of a ring is finitely generated, the same is true of , every finite-type -algebra, and every localization . Thus every ideal of is finitely generated. This includes the zero ideal and the zero ring.
Elementary completion: the ring constructions
The polynomial ring consists of finite coefficient sequences, with coefficientwise addition and the convolution product. Finite distributivity proves its ring laws. Over a domain, the product of two nonzero one-variable polynomials has degree equal to the sum of the degrees, because its leading coefficient is the product of two nonzero coefficients. Iterating this observation shows that a polynomial ring in finitely many variables over a domain is a domain. Evaluation at commuting elements preserves sums and products.
For an ideal , the quotient consists of additive cosets. If representatives are changed by elements of , their sums and products change by elements of : for products expand . Thus the quotient operations are well-defined and inherit the ring laws. The kernel of a ring homomorphism is an ideal; a surjective homomorphism induces a bijection from the quotient by its kernel to its target, preserving both operations. Indeed equal images mean precisely that the representatives differ by a kernel element. A proper ideal is prime exactly when its quotient is a nonzero domain, by the definition that forces or .
Here is the localization construction, including rings with zero divisors. Let be a multiplicatively closed subset containing . On pairs set
Reflexivity and symmetry are immediate. For transitivity, multiply by witnesses annihilating the two terms on the right. The resulting witness for is their product times , which belongs to . Write for the class and define
These formulas respect representatives. If , the cross-multiplied difference for adding to and is ; for multiplication it is . The same witness annihilates both. Replace the other argument in turn. Negation respects the relation as well. Passing finitely many terms to a common denominator proves the ring laws by those of ; and are its zero and unit. The map is a homomorphism, and has inverse for every . If , all classes coincide and this gives the zero ring, as required.
If is a nonzero domain and , the relation reduces to . The map is injective. A nonzero fraction has and inverse , so this is a field, the fraction field. Any field containing contains these fractions with the same operations; hence the fraction field is the smallest such field. This supplies the fraction fields used in the critical-values argument.
Elementary completion: finite generation and ascending chains
An ideal is an additive subgroup closed under multiplication by arbitrary ring elements. It is finitely generated if it consists of the finite sums for some fixed .
If every ideal is finitely generated and , their union is an ideal. Its finitely many generators all lie in one , so and the chain stabilizes. Conversely, if an ideal has no finite generating set, choose and recursively choose . The resulting strictly ascending chain contradicts the ascending chain condition (ACC). Consequently the two definitions of Noetherian ring agree.
Elementary completion of the source's hint
Every infinite sequence of natural numbers has an infinite nondecreasing subsequence. If some value appears infinitely often, take that constant subsequence. Otherwise every bounded set of values appears only finitely often; recursively choose later terms larger than the last one. Apply this first to the first coordinates of a sequence of distinct pairs, and then to the second coordinates of the resulting subsequence. This gives an infinite subsequence nondecreasing in both coordinates. In particular every infinite subset of contains an infinite increasing sequence of distinct pairs. Therefore a family of ideals increasing in both indices cannot assume infinitely many distinct values in a Noetherian ring: choosing one pair for each of infinitely many different ideals and applying this argument would produce a strictly ascending chain of ideals.
The uniform stabilization needed below also has the following direct proof. For ideals increasing in both indices, the diagonal chain stabilizes, say at . For , sandwich between and , so it equals . For each of the finitely many , the chain in stabilizes. Choose greater than all those stabilization indices. Then for every and every .
Polynomial-ring proof
Let be ideals of . Define to be the coefficients of in polynomials of of degree at most . This includes zero and is an ideal: addition and scalar multiplication keep degree at most . Equivalently its nonzero elements are the leading coefficients of degree- polynomials in . Multiplying a polynomial by shows , and the chain in gives the other monotonicity. The preceding argument supplies a single with for all .
For , , induct on its degree. The zero polynomial lies in . If has degree , its leading coefficient lies in , so choose of degree at most with that coefficient; since it is nonzero, has degree . Then has smaller degree, hence lies in by induction. Thus , and the chain stabilizes. ACC equivalence proves that is Noetherian.
Quotients, finite type, and localization
If is the quotient map and is an ideal of , then is an ideal of ; images of a finite generating set generate . Any algebra generated by finitely many elements is a quotient of by the evaluation homomorphism . Iterating the one-variable result and applying the quotient result proves finite-type permanence.
For an ideal , put . If , then ; conversely implies . Hence and images of finitely many generators of generate .
Finally, a field is Noetherian because its ideals are zero and the whole field: a nonzero element of an ideal is invertible. Apply this to and then iterate the polynomial-ring result. No algebraic-closure theorem, algebraic-geometric dimension theorem, or external closed textbook proof is an input to this argument.