In the category of $\mathbb{Z}$-modules, there exists a module $A$---for instance $\bigoplus_{k=2}^\infty \mathbb{Z}/k\mathbb{Z}$---such that a $\mathbb{Z}$-module $B$ is injective iff $\operatorname{Ext}^1_\mathbb{Z}(A,B)=0$.

Does this hold if $\mathbb{Z}$ is replaced with an arbitrary (unital) ring $R$? If not, what are some sufficient conditions on $R$ for this to hold?

As a comment, note that this is equivalent to requiring the existance of a family $\{A_j\}_{j\in J}$ of $R$-modules indexed by some set $J$ such that an $R$-module $B$ is injective iff $\operatorname{Ext}^1_R(A_j,B)=0$ for all $j\in J$: This is clearly implied by the original statement. To get the other direction, set $A:=\bigoplus_{j\in J} A_j$. Then $$\operatorname{Ext}^1_R(A,B)=\prod_{j\in J} \operatorname{Ext}^1_R(A_j, B)=0$$ since each component is $0$.


Yes, such a module exists for each unital ring $R$: Take $A= \bigoplus_I R/I$ where $I$ runs through the left ideals of $R$. This is because $B$ is injective iff $\text{Ext}^1_R(R/I,B)=0$ for each $I$ (Weibel, Lemma 4.1.11). Also note that this is a straightforward generalization of your example $R=\mathbb{Z}$.

Added: If $R$ is a Noetherian commutative ring, you can also take $A=\bigoplus_P R/P$ where $P$ runs through the prime ideals of $R$ (Bruns, Herzog: Cohen-Macaulay rings, 3.1.12).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.