SOLUTIONS4
SOLUTIONS4
Problems 5 and 7
5) Show that for an Artinian ring A, the spectrum Spec(A) is discrete and finite.
We have seen in the lectures that (i) an Artinian ring has finitely many maximal ideals, (ii)
every prime ideal in an Artinian ring is maximal.
Thus, from (i) and (ii), we see that Spec(A) is finite. Moreover, each point m ∈ SpecA is closed,
since m = V (m). The complement of each point is also closed, since it is a finite union of closed
sets. Hence, Spec(A) has discrete topology.
Proof of (ii) is explicitly stated as Proposition 6.6 in the lecture notes. Proof of (i) appears
implicitly in the proof of Theorem 6.7 but let’s spell out a slight variation of that proof here
(which does not use the Noetherian assumption):
Consider the set of all finite intersections m1 ∩ . . . ∩ mr with mi maximal ideals in A. As this
is an Artinian ring, this set has to have a minimal element (otherwise, we can form an infinite
descending chain). Denote the minimal element as m1 ∩ . . . ∩ ms . Now, given any maximal ideal
m, we have
m ∩ m1 ∩ . . . ∩ ms = m1 ∩ . . . ∩ ms
by minimality. But, this means m ⊃ m1 ∩ . . . ∩ ms . This implies m ⊃ mi for some i, as otherwise,
we can find xi ∈ mi \ m for all i = 1, . . . , s and then the element x = x1 x2 . . . xs ∈ m1 m2 · · · ms =
m1 ∩ . . . ∩ ms but not in m since m is prime, a contradiction. Thus, m ⊃ mi for some i, but then
since mi is maximal, m = mi .
7) Suppose that A is a ring with the property that Ap has no nilpotent elements for all p ∈
Spec(A). Show that A has no nilpotent elements. If each Ap is an integral domain, must A be
an integral domain?
Suppose that A has a nilpotent element, that is x ∈ A such that xn = 0 for som n > 1. Let
ann(x) = {a ∈ A : ax = 0} be the annihilator ideal of x. In fact, it is a proper ideal since
1∈/ ann(x). Let m ⊃ ann(x) be a maximal ideal containing ann(x). Consider the localisation
Am . We claim that x1 ̸= 0 ∈ Am . Indeed, if x1 = 0 ∈ Am , there exists s ∈ A \ m such that sx = 0,
n
but such s ∈ ann(x) ⊂ m. Therefore, it follows that x1 ̸= 0 ∈ Am but ( x1 ) = x1 = 0, hence x1 is
a nilpotent element in Am , which is a contradiction to the hypothesis of the problem. Hence, A
has no nilpotent elements.
Take any field k (for example k = F2 ) and consider A = k × k. Then, (1, 0) · (0, 1) = 0 hence A
is not an integral domain, but the only prime ideals in A are k × (0) and (0) × k (since for any
ideal I ⊂ k × k, I ∩ k × {0} and I ∩ {0} × k are ideals), and their localisations are easily seen to
be isomorphic to k, hence are integral domains (in fact, they are isomorphic to fields).