Proposition

Let RR be the ring Z[x1,x2,]\mathbb{Z}[x_1,x_2,\ldots] and let I=(x1,x2,)I = (x_1, x_2, \ldots) be the ideal of RR generated by all indeterminates. Then II is not finitely generated as an RR-module.

Proof

Suppose II is indeed finitely generated, that is, there exists a surjective RR-module homomorphism π:RnI\pi : R^{\oplus n} \to I for some nN.n \in \mathbb{N}.

Since π\pi has a finitely generated free module as its domain, we can describe it fully by specifying where it sends each basis element. This gives us a finite list of polynomials. Let xξx_{\xi} be the lowest index indeterminate not present in any of these polynomials.

Since π\pi is surjective, it must send some RR-linear combination of basis elements r1j1+r2j2++rnjnr_1j_1 + r_2j_2 + \dots + r_n j_n to xξ.x_\xi. This gives us a relation that holds in RR:

xξ=r1π(j1)+r2π(j2)++rnπ(jn).x_\xi = r_1 \pi(j_1) + r_2 \pi(j_2) + \dots + r_n \pi(j_n).

Recall that RR is itself a free object on the set {x1,x2,}\{x_1, x_2, \ldots\} in Ring.\mathbf{Ring}. Consider the morphism φ:RZ\varphi : R \to \mathbb{Z} sending xξx_\xi to 11 and all other indeterminates to 0.0. Note that φ(π(jk))=0\varphi(\pi(j_k)) = 0 for all k.k. This gives us 1=01 = 0 (as integers), a contradiction.

\blacksquare