我自己写了个proof,你看看成不⋯⋯懒得切换中英文,所以我直接用英文写了。 Proof: Suppose A is free Let e_i be the element of A whose i-th coordinate is 1, other coordinate is 0. Consider the homomorphism h: Hom(A,Z)->A, such that every homomorphism f in Hom(A,Z), we map f to the element (f(e_n)) in A. I claim that h is an isomorphism 1,Surjectivity Suppose A is free. Pick any basis S of A, and without loss of generality we may assume that e_n is in S for all n. Then every element (a_n) in A, by the universal property of free module, there should be a homomorphism f mapping each e_n to a_n. So h is surjective. 2,Injectivity Suppose f is a homomorphism mapping all e_n to 0. By the lemma below, f is 0. So h is injective. So, we see that h is an isomorphism. In particular, every homomorphism in Hom(A,Z) is uniquely determined by its value on e_n. Suppose the basis S contains element other than these e_n, say a. Then there are no homomorphism f satisfying f(e_n)=0 and f(a)=0. This is a contradiction with the universal property of free modules. So S must only contain these e_n. But then A is a free abelian group with a countable basis, which it clearly is not. So our initial assumption is false, A is not free.
Lemma: If f(e_n)=o for all n, then f=0. Proof: Consider any sequence of the form a=(a_n*2^n)=(2a_1,4a_2,8a_3,….). Then I claim that f(a)=0. To see this, for any positive integer k, we have a=x+y, where x=(2a_1,4a_2,…, a_k*2^k,0,0,0,0,….), and y=2^k*(0,0,….,0,2a_(k+1),4a_(k+2),8a_(k+3),…). Now, f(a)=f(x)+f(y). As x is clearly a finite linear combination of those e_n, we have f(x)=0. On the other hand, as f is Z-linear, f(y) must be a multiple of 2^k. As k is arbitrary, f(a)=f(y) is a multiple of 2^k for all k. So we can only have f(a)=0. Done. For simplicity, let us give the set of all such elements name A2. Now consider any sequence of the form a=(a_n*3^n)=(3a_1,9a_2,27a_3,….). By similar arguments, we can see that f(a)=0 for all such a too. Let us call the set of all such elements A3. Finally, I claim that A2+A3=A, which will trivially imply the lemma. The claim is actually trivial. Any sequence (a_n) in A, for each coordinate n, we know 2^n and 3^n are caprice. So exist b_n and c_n such than a_n=b_n*2^n+c_n*3^n. Then we obtained the desired decomposition.
谢谢你的意见!不善英语的我只是小结一下看懂的和没有懂的. 首先如果假设可数直积 A 是 自由的,那么 A 同构于可数直和 F (=⊕n≥1 {Z}),并且同构于(同态的群) Hom(A,Z). 当然也同构于 Hom(Z,A)(这个不需要自由性的假设). 这些都是不难说明的. 请问你的证明,包括引理,都是为了说明 A 同构于 Hom(A,Z) 吗?.... 而我想这里是关键: >>>...then A is a free abelian group with a countable basis, which it clearly is not. 想问一下为什么A不是可数秩自由Z模?是不是你已经证明而我看漏了? 还有些细节问题根据必要再讨论吧.我英语不好,基本上只看得懂法国数学家写的相对简洁的英文文献.