I have a question about an argument from a proof in Hideyuki Matsumura's "Commutative Ring Theory" on page 9, Theorem 2.5:

Let $(A,\mathfrak{m})$ be a local ring; then a finitely generated projective module $M$ over $A$ is free.

The proof works as follows:

Choose a minimal $A$-basis $\omega_1,...,\omega_n$ of $M$. Take into account that "minimal" means that there cannot exist another system of generators $b_1,..., b_m \in M$ with $m < n$ and $M = \sum_{i=1} ^m A b_i$.

Define a surjective map $\varphi:F \to M$ from the free module $F = Ae_1 \oplus \cdots \oplus Ae_n$ to $M$ by $\varphi(\sum a_i e_i) = \sum a_i\omega_i$. If we set $K = \operatorname{Ker}(\varphi)$ then, from the minimal basis property

$$\sum a_i \omega_i =0 \Rightarrow a_i \in \mathfrak{m} \text{ for all } i. $$

Thus $K \subset \mathfrak{m}F$. Because $M$ is projective, there exists $\psi: M \to F$ such that $F = \psi(M)\oplus K$, and it follows that $K = \mathfrak{m}K$. (???)

On the other hand, $K$ is a quotient of $F$, therefore finite over $A$, so that $K = 0$ by Nakayama and $F = M$.

Question: why the fact that $F = \psi(M)\oplus K$ implies that $K = \mathfrak{m}K$?

$K \subset \mathfrak{m}F$, then we can deduce $K=K∩mF$$K∩(m \psi(M)\oplus mK)$ But why the last is equal to $mK$?

Thank you for your kind help.


2 Answers 2


You have $K=K\cap \mathfrak mF=K\cap(\mathfrak m\psi(M)\oplus \mathfrak mK)$. Since $K\cap \psi(M)=0$ ($\because F=\psi(M)\oplus K$), you get $K\cap\mathfrak m\psi(M)=0$. Thus $K=\mathfrak mK$ as required.


Let $k=A/\mathfrak{m}$ be the residue field of $A$. Tensor the exact sequence $G \rightarrow F\rightarrow M\rightarrow 0$ with $k$ (where $G$ is a free module minimally projecting on $K$) to get $F/\mathfrak{m}F\simeq M/\mathfrak{m}M$ ($K\subseteq \mathfrak{m}F$). Now tensor the equality $F=\psi(M)\oplus K$ with $k$ and count vector space dimension.

  • $\begingroup$ We cannot deduce the consequence without using tensor product? Tensor product appears much later in this text. $\endgroup$ Dec 15, 2020 at 7:20
  • $\begingroup$ Sure we can, tensor products are unnecessary, $\mathfrak{m}F=\mathfrak{m}\psi(M)\oplus\mathfrak{m}K$, so just use the definition of a direct sum. $\endgroup$
    – Phoenix
    Dec 15, 2020 at 7:43
  • $\begingroup$ How do we use the definition of a direct sum? We know K⊂mψ(M)×mK.From here, why we can say K⊂mK? $\endgroup$ Dec 15, 2020 at 7:51
  • $\begingroup$ Every element $f\in F$ has a unique representation $f=a+b$, $a\in \psi(M)$, $b\in K$. If $k\in K$ is arbitrary, then $k$ as an element of $F$ (it is a submodule) can be written in the above form as $0+k$. On the other hand every element of $\mathfrak{m}F$ has a unique decomposition $x+y$, $x\in \mathfrak{m}\psi(M)\subseteq \psi(M)$, $y\in \mathfrak{m}K\subseteq K$. So $k$ as an element of $\mathfrak{m}F$ and hence $F$ can be written uniquely as $x+y$, $x\in \mathfrak{m}\psi(M)$, $y\in \mathfrak{m}K$. Therefore $x=0$, $y=k$, so $K=\mathfrak{m}K$. $\endgroup$
    – Phoenix
    Dec 15, 2020 at 17:00

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .