Processing math: 2%

Sunday, July 12, 2015

Primary Ideals and Decompositions (15.2.28,31-32)

Dummit and Foote Abstract Algebra, section 15.2, exercises 28, 31-32:

MathJax TeX Test Page 28. Prove that each of the following rings have infinitely many minimal prime ideals, and that (0) is not the intersection of any finite number of these (so (0) does not have a primary decomposition in these rings):
(a) The infinite direct product ring ℤ/2ℤ×ℤ/2ℤ×...
(b) k[x_1,x_2,...]/(x_1x_2,x_3x_4,...)
31-32. Let Q_1 and Q_2 be P-primary ideals in a commutative ring with identity R. Show that Q_1∩Q_2 is also a P-primary ideal, and prove that if P is a maximal ideal, then Q_1Q_2 and Q_1+Q_2 are also P-primary ideals.

Proof: 28. (a) Note that since every element of ℤ/2ℤ×ℤ/2ℤ×... is multiplicatively idempotent, it follows that the quotient of ℤ/2ℤ×ℤ/2ℤ×... with any of its prime ideals results in an integral domain with two elements, 1 and 0. Therefore, every prime ideal in this ring is a maximal ideal, and in particular is minimal. If we let I_n denote the ideal of elements whose n^{\text{th}} coordinate is zero, then we easily observe an infinite collection of prime/maximal ideals. Now, we shall prove that given any ideal I with the property that it contains an element x which has infinitely many nonzero coordinates, then P∩I also has this property when P is a prime ideal; inductively, this will prove particularly that any finite intersection of prime ideals is nonzero. To wit, write x=y+z for yz=0 and each of y,z having infinitely many nonzero coordinates. Then since yz∈P we must have at least one of y,z∈P, say y∈P. Then since y=yx we also have y∈I, hence y∈P∩I as claimed.

(b) For any prime ideal P⊆R=k[x_1,x_2,...]/(x_1x_2,x_3x_4,...), since \overline{x_1}\overline{x_2}=0 we must have \{\overline{x_{2n-1}},\overline{x_{2n}}\}∩P≠ø for each n∈ℕ. Conversely, if S⊆ℕ is such that exactly one of 2n-1 or 2n is in S for each n∈ℕ, and P_S=(x_i~|~i∈S)⊆k[x_1,x_2,...], then k[x_1,x_2,...]/P_S≅R/\overline{P_S} is an integral domain and hence \overline{P_S} is a prime ideal. It follows that the minimal prime ideals of R are exactly \overline{P_S} for such sets S, of which there are (uncountably) infinitely many. Now, let P_{S_1},...,P_{S_n} be any finite collection of such ideals. For each 1≤i≤n let α_i∈\{2i-1,2i\} be such that \overline{x_{α_i}}∈P_{S_i}. Then since provided each m_1,m_2... is monomial, for polynomials p∈(m_1,m_2,...) if and only if each monomial term of p is divisible by some m_j, we see β=x_{α_1}x_{α_2}...x_{α_n}∉(x_1x_2,x_3x_4,...) while β∈P_{S_1}∩...∩P_{S_n}, so that \overline{P_{S_1}}∩...∩\overline{P_{S_n}} is nonzero.

31-32. Observe: \text{rad }(Q_1Q_2)=\text{rad }(Q_1∩Q_2)=(\text{rad }Q_1)∩(\text{rad }Q_2)=P∩P=P \text{rad }(Q_1+Q_2)=\text{rad }(\text{rad }Q_1+\text{rad }Q_2)=\text{rad }(P+P)=\text{rad }P=P Therefore it suffices to prove Q_1∩Q_2 is a primary ideal, and that when P is maximal, so too are Q_1Q_2 and Q_1+Q_2.

Suppose xy∈Q_1∩Q_2 where y∉Q_1∩Q_2, say y∉Q_1. Then x^n∈Q_1⊆P=\text{rad }Q_2 for some n, and therefore x^{nm}∈Q_1∩Q_2 for some further m.

Now suppose P is maximal. Now Q_1Q_2 and Q_1+Q_2 are primary by Proposition 19.4.~\square