Fulton 2 - Lemmas

Posted on November 15, 2017

Lemma 1. If N is a submodule of a finitely generated R-module M, then M/N is finitely generated.

Let \{s_1,\ldots,s_n\} be a generating set for M. Then \{s_1+N,\ldots,s_n+N\} generates M/N.

Lemma 2. Suppose N is a submodule of an R-module M and N,M/N are finitely generated over R. Then M is finitely generated over R.

Let N=\sum Ra_i, M/N=\sum R(b_i+N). We will show that M=\sum Ra_i+\sum Rb_i. If x\in M then clearly x+N\in M/N. Therefore x+N=\sum_{i=1}^mr_ib_i+N for some r_i\in R. Therefore x-\sum_{i=1}^mr_ib_i=y\in N. Moreover, y=\sum_{i=1}^ns_ia_i for some s_i\in R. Therefore x=\sum_{i=1}^mr_ib_i+\sum_{i=1}^ns_ia_i.

Definition 3. A module is Noetherian if every submodule is finitely generated.

Lemma 4. Suppose N is a submodule of an R-module M. Then M is Noetherian if and only if N,M/N are Noetherian. Equivalently if N,M,P are R-modules and      0\longrightarrow N\longrightarrow M\longrightarrow P\longrightarrow 0 is exact, then M is Noetherian if and only if N,P are Noetherian.

Suppose M is Noetherian. Then N is clearly Noetherian since submodules of N are submodules of M. Moreover, any submodule L of M/N is R-module isomorphic to L&39;/N, where L&39; is some submodule of M containing N. Since L&39; is finitely generated, L is finitely generated by Lemma 1. Hence M/N is Noetherian as well.

Suppose conversely that N,M/N are Noetherian. Let L be a submodule of M. Then L\cap N is a submodule of N and hence is finitely generated. Moreover, L/(L\cap N) is isomorphic to (L+N)/N. Since (L+N)/N is a submodule of M/N, (L+N)/N is finitely generated. Thus L\cap N and L/(L\cap N) are both finitely generated, and hence L is finitely generated by Lemma 2.

The result may be reinterpreted in terms of the exact sequence above by noting that P is R-module isomorphic to M/N.

Lemma 5. If R is a Noetherian ring, then R^n is Noetherian as an R-module for all n\ge 1.

If n=1 then R is Noetherian as an R-modules since it is a Noetherian ring (the R-submodules of R are precisely the ideals of R).

Suppose that R^n is a Noetherian R-module for some n\ge 1. Observe that R^{n+1}/R is R-module isomorphic to R^n, where R is identified with the (n+1)th copy of R in R^{n+1}. Since R^n, R are each Noetherian R-modules, R^{n+1} is Noetherian by Lemma 4.

Lemma 6. If R is a Noetherian ring and M is an R-module, M is Noetherian if and only if it is finitely generated over R.

If M is Noetherian, then it is finitely generated by definition. Suppose M is generated by the finite set \{s_1,\ldots,s_n\}. Note that there exists a natural surjective R-module homomorphism \varphi:R^n\to M mapping 1 in the kth copy of R to s_k. Thus M is R-module isomorphic to R^n/\ker\varphi. By Lemma 5, R^n is Noetherian. Hence M is Noetherian by Lemma 4.

Lemma 7. Let K be a field and R\supset K an integral domain. Let L be the field of fractions of R. Then R is module finite over K if and only if L is a finite field extension of K.

Suppose R is module finite over K and \{v_1,\ldots,v_n\} a basis for R. Note that since R is module finite over K, v_1,\ldots,v_n are each algebraic over K. So K(v_1,\ldots,v_n) is a finite extension of K. Since R\subset K(v_1,\ldots,v_n), L\subset K(v_1,\ldots,v_n). Thus L is a finite extension of K.

Suppose L is a finite extension of K. Since fields are Noetherian, L is Noetherian as a K-module by Lemma 6. Thus since K\subset R\subset L, R is a finitely generated K-module.

Lemma 8. Let F\in k[X_1,\ldots,X_n] be a polynomial of degree 1. Then there exists an affine change of coordinates T:\mathbb{A}^n\to\mathbb{A}^n such that F^T=X_m.

Suppose that F=\sum_{i=1}^n a_iX_i+a_0 with a_1,\ldots,a_n\in k\setminus\{0\}. We will define an affine change of coordinates T=T&39;\circ T&39;&39; with T&39; a linear map and T&39;&39; a translation. Let T_i&39;&39;=X_i for i\ne n and T_n&39;&39;=X_n-a_0. Define T&39; by setting T_i&39;=X_i for i<d, T_d&39;=a_d^{-1}X_d, and T_i&39;=a_i^{-1}X_i-a_i^{-1}X_{i-1} for i>d. Observe that T&39; is defined by the following matrix in k^{n\times n},      T&39;=     \begin{bmatrix}         1 & 0 & 0 & \ldots & \ldots & 0\\         0 & a_1^{-1} & 0 & \ldots & \ldots & 0\\         0 & -a_{2}^{-1} & a_{2}^{-1} & \ldots & \ldots & 0\\         \vdots & \ddots & \ddots & \ddots & \ddots & \vdots\\         0 & \ldots & \ldots & -a_{n-1}^{-1} & a_{n-1}^{-1} & 0\\         0 & \ldots & \ldots & 0 & -a_n^{-1} & a_n^{-1}     \end{bmatrix} Note that T&39; is upper triangular, and therefore invertible. So T=T&39;\circ T&39;&39; is an affine change of coordinates. Observe that  \begin{aligned}[t] F^T &= \left(\sum_{i=1}^n a_i(T_i)\right)+a_0\\ &= \left(\sum_{i=1}^{n-1} a_i(T_i&39;(X_i)\right)+a_n(T_n&39;(X_n-a_0)+a_0\\ &=  a_1(a_1^{-1}X_1)+\left(\sum_{i=2}^{n-1} a_i(a_i^{-1}X_i-a_i^{-1}X_{i-1})\right)+a_n(a_n^{-1}(X_n-a_0)-a_n^{-1}X_{n-1})+a_0\\ &= X_n. \end{aligned}

Lemma 9. Let T be an affine change of coordinates on \mathbb{A}^n and S\subset k[X_1,\ldots,X_n]. If I(S)\ne I(S\setminus\{F_i\}) for some i\in\{1,\ldots,r\}, then I(S)^T\ne I(S\setminus\{F_i\})^T.

Suppose G\in I(S). Observe that if \widetilde{T}(G)\in I(S\setminus\{F_i\})^T, then \widetilde{T}^{-1}(\widetilde{T}(G))=G\in I(S\setminus\{F_i\}).

Definition 10. Let R be a ring, A\subset R, and I=I(A). We say A is a minimal generating set for I if for all B\subset I such that I=I(B), |B|\ge|A|.

Lemma 11. Suppose V=V(F_1,\ldots,F_r) is a linear subvariety of \mathbb{A}^n. If \{F_1,\ldots,F_r\} is a minimal generating set for I(V), then V has dimension n-r.

By Problem 2.14 there exists an affine change of coordinates T on \mathbb{A}^n such that V^T=V(X_{m+1},\ldots,X_n). Also, V^T=V(I(V)^T)=V(F_1^T,\ldots,F_r^T). Since \{F_1,\ldots,F_r\} is a minamal generating set for I(V), by Lemma 2.9, \{F_1^T,\ldots,F_r^T\} is a minimal generating set for I(V)^T. Therefore n-(m+1)+1=r and m=n-r.

Lemma 12. Let V be an algebraic set in \mathbb{A}^n, and L a linear subvariety of dimension 1. Then either L\subset V or V\cap L is finite.

Suppose L=V(X_2,\ldots,X_n). Let F\in I(V). Note that V\subset V(F). Observe that the set V(F)\cap L is the set of solutions to F(X_1,0,\ldots,0), a polynomial of one variable. Thus either F(X_1,0,\ldots,0)=0 and F\in I(L), or V(F)\cap L is finite.

Suppose L is any linear subvariety of dimension 1. By 2.14(b), there exists an affine change of coordinates T such that L^T=V(X_2,\ldots,X_n). So by the result above, either L^T\subset V^T or V^T\cap L^T is finite. Thus L\subset V or V\cap L is finite.

Note. The lemma above also has a proof using Corollary 4 of the Nullstellensatz, by showing that k[X_1,\ldots,X_n]/I(V\cap L) is a finite dimensional vector space over k.

Lemma 13. Suppose W is a subvariety of a variety V, and P is a point in W. Then the natural homomorphism \pi:\Gamma(V)\to\Gamma(W) extends uniquely to a homomorphism \mathcal{O}_{P}(V)\to\mathcal{O}_{P}(W).

Note that \pi:\Gamma(V)\to\Gamma(W) is the homomorphism induced by the inclusion polynomial map W\hookrightarrow V. Thus by Problem 2.21, \pi extends uniquely to a homomorphism \mathcal{O}_{P}(V)\to\mathcal{O}_{P}(W).

Lemma 14. Let V\subset\mathbb{A}^n be an algebraic set. Then \dim_k(k[X_1,\ldots,X_n]/I(V)) is finite if and only if V is finite.

Let us decompose V(I) into irreducible algebraic sets V(I)=\bigcup_{i=1}^mV_i.

Suppose V is finite, that is, each V_i is a point. Therefore the ideals I(V_1),\ldots,I(V_n) are distinct maximal ideals. In particular, they are pairwise comaximal. Thus, by the Chinese Remainder Theorem,      k[X_1,\ldots,X_n]/I(V)\cong k[X_1,\ldots,X_n]/I(V_1)\times         \ldots\times k[X_1,\ldots,X_n]/I(V_m)\cong k^m Conversely, suppose \dim_k(k[X_1,\ldots,X_n]/I(V))=d<\infty. Since I(V_i)\supset I(V) for each i, \dim_k(k[X_1,\ldots,X_n]/I(V_i)) is finite. Thus, by Problem 2.4, k[X_1,\ldots,X_n]/I(V_i)\cong k for each i and the result follows as above. Moreover, d=m.