Arxiv:2106.10441V2 [Math.AC] 2 Jul 2021 Iuto Ae Tdﬃutt Characterize to Diﬃcult It Makes Situation Oteinrnsuigmlilctv Ust.Snete,Tenotion the Then, Since Subsets

arXiv:2106.10441v2 [math.AC] 2 Jul 2021 iuto ae tdffiutt characterize to difficult it makes situation oteinrnsuigmlilctv ust.Snete,tenotion the then, Since subsets. multiplicative using rings Noetherian sI hr xssafiiegnrtdsubideal generated finite a exists there s 2 ,5 ,1,1,15]. e 12, studied 10, and 9, introduced 5, are 4, domains, [2, Mori strong a rings, and coherent domains rings, GCD artinian as such rings, well-known other of Characterizing ,s S, 1 nrdcdtento of notion the introduced [1] lasi utpiaiesbe of subset multiplicative a is always ∈ o e’ obc otedfiiinof definition the to back go let’s Now hogotti article, this Throughout .Dprmn fBscCuss hnd eoatcPolytechnic Aeronautic Chengdu Courses, Basic of Department a. ⊆ 2 .Sho fMteaia cecs iha omlUiest,C University, Normal Sichuan Sciences, Mathematical of School b. S ∈ modules, 00MteaisSbetClassification: Subject Mathematics 2010 ie hteeyfree every that vided called oue sas nrdcd ring A introduced. also is modules e Words: Key Hom isof ties une0 quence in of tions S K poetv modules, -projective uhthat such S Let i.e., , Let . R ( S R ,B P, S poetv rvddta h nue eune0 sequence induced the that provided -projective S poetv oue r band h oinof notion The obtained. are modules -projective earn and ring a be S smsml ig r rvddb using by provided are rings -semisimple S I/K → eamlilctv ustof subset multiplicative a be ijciemodules. -injective ) sI S → A smsml rings, -semisimple suniformly is ⊆ S → Hom poetv oue and modules -projective K B R sdcddb h ideal the by decided is S R → S R ( ,C P, ijciemdlsand modules -injective S ioe Zhang Xiaolei mdl is -module utpiaiesbe of subset multiplicative a -al [email protected] E-mail: Nehra rings -Noetherian lasi omttv igwt dniyand identity with ring commutative a is always C yuniformity by 1. ) S → Introduction → trinb 1] hsi elgnrto of generation well a is This [18]. by -torsion R Abstract .Sm hrceiain n proper- and characterizations Some 0. hti,1 is, that , S is 0 S -split R Nehra ig.Ntc htteelement the that Notice rings. -Noetherian K 1 S S smsml.Svrlcharacteriza- Several -semisimple. scle an called is a S of Nehra ig rmteperspective the from rings -Noetherian e Qi Wei , eatfrany for -exact R S n20,Adro n Dumitrescu and Anderson 2002, In . I 64,16D60. 16D40, eatsequences, -exact n nelement an and R ∈ I S nwihfrayideal any for which in b S for -split S S and R smsml modules, -semisimple smsml igpro- ring -semisimple S S An . Nehra ig.This rings. -Noetherian hnd 110 China 610100, Chengdu , S smsml rings -semisimple S s egu606,China 610068, hengdu eatsequences. -exact → 1 soteatse- exact -short s R 2 ms efc rings, perfect lmost Hom S -module ∈ S -semisimple s -projective of s S ∈ R o any for ( tnieyin xtensively S S ,A P, -analogues uhthat such P ) is I → of s 1 R ∈ S of module-theoretic viewpoint. In order to overcome this difficulty, Qi and Kim et al. [13] recently introduced the notion of uniformly S-Noetherian rings R for which there exists an element s ∈ S such that for any ideal I of R, sI ⊆ K for some finitely generated sub-ideal K of I. They also introduced notion of S-injective modules and finally showed that a ring R is uniformly S-Noetherian if and only if any direct sum of injective modules is S-injective in the case that S is composed of non-zero-divisors. Another “uniform” case is that of S-von Neumann regular rings introduced by the author of this paper (see [18]). The author in [18] first introduced S-flat modules using uniformly S-torsion modules, and then gave the notion of S- von Neumann regular rings extending von Neumann regular rings with uniformity on the multiplicative subset S. Finally, he characterized S-von Neumann regular rings by using S-flat modules. The main motivation of this paper is to introduce and study the S-versions of projective modules and semisimple rings. In Section 2 of this article, we first introduce the notions of S-split S-exact sequences (see Definition 2.1). Dual to S-injective modules, we introduce the notion of S-projective modules and show that an R- 1 module P is S-projective if and only if ExtR(P, M) is uniformly S-torsion for any R-module M, if and only if any S-exact sequence ending at P is S-split (see Theo- rem 2.5). We also give a local characterization of projective modules in Proposition 2.10. In Section 3 of this article, we first give the notion of S-semisimple modules M for which any S-shortly exact sequence with middle term M is S-split. And then we introduced the notion of S-semisimple rings over which every free module is S-semisimple. We prove that a ring R is an S-semisimple ring if and only if every R-module is S-semisimple, if and only if every S-short exact sequence is S-split, if and only if every R-module is S-projective, if and only if every R-module is S- injective, if and only if R is uniformly S-Noetherian and S-von Neumann regular (see Theorem 3.5). We also show that if the multiplicative subset S is composed of nonzero divisors, then a ring R is an S-semisimple ring if and only if R is semisimple in Proposition 3.7. A non-trivial example of a S-semisimple ring which is not S-semisimple is given in Example 3.8. Finally, we give a local characterization of semisimple rings in Proposition 3.9.

2. S-split S-exact sequences and S-projective modules Let R be a ring and S a multiplicative subset of R. Recall from [17, Deﬁnition 1.6.10] that an R-module T is called a uniformly S-torsion module provided that there exists an element s ∈ S such that sT = 0. Suppose M, N and L are R- modules. 2 (1) An R-homomorphism f : M → N is called an S-monomorphism (resp., S-epimorphism) provided that Ker(f) (resp., Coker(f)) is a uniformly S- torsion module. (2) An R-homomorphism f : M → N is called an S-isomorphism provided that f is both an S-monomorphism and an S-epimorphism. f g (3) An R-sequence M −→ N −→ L is called S-exact provided that there is an element s ∈ S such that sKer(g) ⊆ Im(f) and sIm(f) ⊆ Ker(g). (4) An S-exact sequence 0 → A → B → C → 0 is called a short S-exact sequence. It is easy to verify that f : M → N is an S-monomorphism (resp., S-epimorphism) f f if and only if 0 → M −→ N (resp., M −→ N → 0 ) is S-exact. f g Recall that an exact sequence 0 → A −→ B −→ C → 0 is split provided that there ′ ′ is an R-homomorphism f : B → A such that f ◦ f = IdA.

f g Deﬁnition 2.1. Let ξ :0 → A −→ B −→ C → 0 be an S-short exact sequence. Then ξ is said to be S-split provided that there is s ∈ S and R-homomorphism f ′ : B → A ′ ′ such that f (f(a)) = sa for any a ∈ A, that is, f ◦ f = sIdA.

f Obviously, split exact sequence is S-split. Certainly, an exact sequence 0 → A −→ g B −→ C → 0 if and only if there exists an R-homomorphism g′ : C → B such that ′ g ◦ g = IdC .

Lemma 2.2. Let R be a ring and S a multiplicative subset of R. An S-short exact f g sequence ξ : 0 → A −→ B −→ C → 0 is S-split if and only if there is s ∈ S and R-homomorphism g′ : C → B such that g ◦ g′(c) = sc for any c ∈ C, that is, ′ g ◦ g = sIdC for some s ∈ S.

f g Proof. Suppose the S-short exact sequence ξ : 0 → A −→ B −→ C → 0 is S-split. ′ ′ Then there is s1 ∈ S and R-homomorphism f : B → A such that f ◦ f = s1IdA. Set A′ = Ker(g). Then there is an S-isomorphism t : A → A′ such that f = i ◦ t where i : A′ → B is the natural embedding map. By [19, Proposition 1.1], there ′ ′ ′ ′ exists an S-isomorphism t : A → A satisfying t ◦ t = s2IdA′ and t ◦ t = s2IdA for ′ ′ ′ ′ ′ ′ ′ some s2 ∈ S. Set i = s2t ◦ f , then i ◦ i = i ◦ s2t ◦ f = t ◦ f ◦ i ◦ t ◦ t = s1s2IdA′ . Setting C′ = Coker(i) = Im(g), we have the following commutative diagram with rows exact: i π 0 / A′ / B / C′ / 0 i′ ④ ④ s1s2 ④ s1s2 ④ s1s2 } ④ ④ ′ ④} ④} π 0 / A′ / B / C′ / 0, i π 3 where the vertical maps are all multiplications by s1s2. By [17, Exercise 1.60], The ′ ′ ′ ′ ′ existence of i : B → A such that i ◦ i = s1s2IdA′ implies that of π : C → B such ′ ′ that π ◦ π = s1s2IdC′ . Note that C is S-isomorphic to C . By [19, Proposition 1.1], ′ ′ ′ ′ there exist an S-isomorphism l : C → C such that l ◦ l = s3IdC and l ◦ l = s3IdC′ ′ ′ ′ ′ for some s3 ∈ S where l : C → C is the embedding map. Set g = l1 ◦ π and ′ ′ ′ s = s1s2s3 ∈ S. Then g ◦ g = l ◦ π ◦ π ◦ l = sIdC . So the necessity holds. The suﬃciency can be proved similarly. Recall from [13, Deﬁnition 4.1] that an R-module E is called S-injective provided that the induced sequence

0 → HomR(C, E) → HomR(B, E) → HomR(A, E) → 0 is S-exact for any S-exact sequence 0 → A → B → C → 0. By [13, Theorem 4.3], an R-module E is S-injective, if and only if for any short exact sequence f g g∗ f ∗ 0 → A −→ B −→ C → 0, the induced sequence 0 → HomR(C, E) −→ HomR(B, E) −→ 1 HomR(A, E) → 0 is S-exact, if and only if ExtR(M, E) is uniformly S-torsion for any R-module M. We can characterize S-injective modules using S-exact sequences.

Proposition 2.3. Let R be a ring, S a multiplicative subset of R and E an R- module. Then the following assertions are equivalent: (1) E is S-injective; f (2) for any S-monomorphism A −→ B and any R-homomorphism h : A → E, there exists an R-homomorphism g : B → E and s ∈ S such that sh = g ◦ f; (3) any S-short exact sequence 0 → E → B → C → 0 beginning at E is S-split; (4) any short exact sequence 0 → E → B → C → 0 beginning at E is S-split.

Proof. (1) ⇒ (2): Considering the short exact sequence 0 → Im(f) → B →

Coker(f) → 0, we have the following S-exact sequence 0 → HomR(Coker(f), E) → 1 HomR(B, E) → HomR(Im(f), E) → ExtR(Coker(f), E). Since E is S-injective, 1 ExtR(Coker(f), E) is uniformly S-torsion. Since the natural R-homomorphism π A −→ Im(f) is an S-isomorphism, then there is an S-isomorphism α : Im(f) → A

such that π ◦ α = s1IdIm(f) and α ◦ π = s1IdA for some s1 ∈ S by [18, Proposition

1.1]. Thus there is an R-homomorphism g : B → E such that s2h ◦ α = g ◦ π for

some s2 ∈ S. So s2h ◦ α ◦ π = g ◦ i ◦ π where i : Im(f) → B is the natural embedding

map. Setting s = s1s2 ∈ S, we have sh = g ◦ f. i (2) ⇒ (1): Let M be an R-module and 0 → N −→ P → M → 0 a short exact sequence of R-modules with P projective. Then we have a long exact sequence i∗ 1 0 → HomR(M, E) → HomR(P, E) −→ HomR(N, E) → ExtR(M, E) → 0. By (2), i∗ 1 is an S-epimorphism. Thus ExtR(M, E) is uniformly S-torsion. So E is S-injective. 4 f g (1) ⇒ (3): Let 0 → E −→ B −→ C → 0be an S-short exact sequence of R-modules.

Then 0 → HomR(C, E) → HomR(B, E) → HomR(E, E) → 0 is S-exact. Thus ′ there exists an element s ∈ S such that sIdE = f ◦ f for some R-homomorphism f ′ : B → E. So0 → E → B → C → 0 is S-split. (3) ⇒ (4): Obvious. f g (4) ⇒ (1): Let 0 → E −→ B −→ C → 0 be a short exact sequence of R-modules. ′ ′ Then there is an R-homomorphism f : B → E satisfying f ◦ f = sIdE. Note g∗ f ∗ that there is an exact sequence 0 → HomR(C, E) −→ HomR(B, E) −→ HomR(E, E). ′ ′ Let g be R-homomorphism in HomR(E, E). Then g ◦ f ◦ f = sg with g ◦ f ∈ g∗ f ∗ HomR(B, E). Thus 0 → HomR(C, E) −→ HomR(B, E) −→ HomR(E, E) → 0 is S-exact. So E is S-injective.

Recall that an R-module P is said to be projective provided that the induced

sequence 0 → HomR(P, A) → HomR(P, B) → HomR(P,C) → 0 is exact for any exact sequence 0 → A → B → C → 0. Now we give an S-analogue of projective modules.

Deﬁnition 2.4. Let R be a ring and S a multiplicative subset of R. An R-module P is called S-projective provided that the induced sequence

0 → HomR(P, A) → HomR(P, B) → HomR(P,C) → 0

is S-exact for any S-exact sequence 0 → A → B → C → 0.

In common with the classical cases, we have the following characterizations of S- projective modules. Since the proof is very similar with that of characterizations of S-injective modules (see Proposition 2.3 and [13, Theorem 4.3]), we omit the proof.

Theorem 2.5. Let R be a ring, S a multiplicative subset of R and P an R-module. Then the following assertions are equivalent: (1) P is S-projective; f g (2) for any short exact sequence 0 → A −→ B −→ C → 0, the induced sequence f∗ g∗ 0 → HomR(P, A) −→ HomR(P, B) −→ HomR(P,C) → 0 is S-exact; 1 (3) ExtR(P, M) is uniformly S-torsion for any R-module M; n (4) ExtR(P, M) is uniformly S-torsion for any R-module M and n ≥ 1; g (5) for any S-epimorphism B −→ C and any R-homomorphism h : P → C, there exists an R-homomorphism α : P → B and s ∈ S such that sh = g ◦ α; f g (6) every S-short exact sequence 0 → A −→ B −→ P → 0 is S-split; f g (7) every short exact sequence 0 → A −→ B −→ P → 0 is S-split. 5 By Theorem 2.5, projective modules are S-projective. Moreover, uniformly S- torsion modules are S-projective by [13, Lemma 4.2].

Corollary 2.6. Let R be a ring and S a multiplicative subset of R. Suppose P is a uniformly S-torsion R-module or a projective R-module. Then P is S-projective.

R Recall from [18] that an R-module F is S-ﬂat if and only if Tor1 (M, F ) is uniformly S-torsion for any R-module M.

Proposition 2.7. Let R be a ring and S a multiplicative subset of R. If P is an S-projective R-module, then P is S-ﬂat.

Proof. Let P be an S-projective R-module, M an R-module and E an injective 1 cogenerator. Then there is an element s ∈ S such that sExtR(P, HomR(M, E))=0 R by Theorem 2.5. By [3, Lemma 2.16(b)], we have sHomR(Tor1 (P, M), E)) = 0. Let R f : sTor1 (P, M) → E be an R-homomorphism. Since E is injective, there is an R- R R →֒ (homomorphism g : Tor1 (P, M) → E such that f = g ◦ i where i : sTor1 (P, M R R Tor1 (P, M) is the embedding map. Since sHomR(Tor1 (P, M), E)) = 0, we have R R f(sx)= g(sx)= sg(x) = 0forany x ∈ Tor1 (P, M). Thus HomR(sTor1 (P, M), E)= R 0. So sTor1 (P, M) = 0 since E is an injective cogenerator. Consequently, P is S- ﬂat.

Proposition 2.8. Let R be a ring and S a multiplicative subset of R. Then the following assertions hold. (1) Any ﬁnite direct sum of S-projective modules is S-projectve. f g (2) Suppose 0 → A −→ B −→ C → 0 is an S-exact sequence. If A and C are S-projective modules, so is B. (3) Suppose A → B is an S-isomorphism. Then A is S-projective if and only if B is S-projective. f g (4) Suppose 0 → A −→ B −→ C → 0 is an S-exact sequence. If B and C are S-projective, then A is S-projective. f g (5) Suppose 0 → A −→ B −→ C → 0 is an S-split S-exact sequence. If B is S-projective, then A and C are S-projective.

Proof. We only prove (5) since the proof of (1)-(4) is dual to that of [13, Proposition 4.7]. f g (5): Suppose 0 → A −→ B −→ C → 0 is an S-split S-exact sequence where B is S-projective. Let h : M → N be an S-epimorphism. Let β : C → N be an 6 R-homomorphism. Consider the following diagram where both rows are S-exact.

g B / C / 0 ✤ α β ✤ h M / N / 0 There is an R-homomorphism α : B → M and s ∈ S such that sβ ◦ g = h ◦ α. Since g is an S-split S-epimorphism. Then there is an R-homomorphism g′ : C → B ′ ′ ′ such that g ◦ g = tIdC for some t ∈ S. So sβ = sβ ◦ g ◦ g = h ◦ α ◦ g . So C is S-projective. Let γ : A → N be an R-homomorphism. Consider the following diagram where both rows are S-exact.

g 0 / A / B ❊❊ γ ❊❊ ❊❊ " h ❊" M / N / 0 ′ ′ Let g : B → A be an R-homomorphism such that g ◦ g = sIdA for some s ∈ S. Since B is S-projective, there exists an R-homomorphism δ : B → M such that h ◦ δ = s′γ ◦ g′ for some s′ ∈ S. Thus h ◦ δ ◦ g = s′γ ◦ g′ ◦ g = ss′γ. So A is S-projective.

It is well-known that any direct sum of projective modules is projective. However, the following example shows that any direct sum of S-projective modules is not S- projective.

Example 2.9. Let R = Z be the ring of integers, p a prime in Z and S = {pn|n ∈ n N}. Let Mn = Z/hp i for each n ≥ 1. Then Mn is uniformly S-torsion and thus ∞ 1 n m min{m,n} S-projective. Set N = L Mn. Note that ExtZ(Z/hp i, Z/hp i) ∼= Z/hp i. n=1 1 1 n m min{m,n} We have ExtZ(N, N) ∼= Q ( L ExtZ(Z/hp i, Z/hp i))) ∼= Q ( L Z/hp i). n∈N m∈N n∈N m∈N Since the abelian group Q ( L Z/hpmin{m,n}i) contains a subgroup Q Z/hpni which n∈N m∈N n∈N 1 is not uniformly S-torsion. So ExtZ(N, N) is not uniformly S-torsion. Consequently N is not S-projective.

Let p be a prime ideal of R. We say an R-module P is p-projective shortly provided that P is (R − p)-projective.

Proposition 2.10. Let R be a ring and P an R-module. Then the following statements are equivalent: (1) P is projective; (2) P is p-projective for any p ∈ Spec(R); 7 (3) P is m-projective for any m ∈ Max(R).

Proof. (1) ⇒ (2) ⇒ (3) : Trivial. 1 (3) ⇒ (1) : Let M be an R-module. Then ExtR(P, M) is (R − m)-torsion. Thus 1 for any m ∈ Max(R), there exists sm ∈ R − m such that smExtR(P, M) = 0. Since 1 the ideal generated by {sm | m ∈ Max(R)} is R, we have ExtR(P, M)=0. So P is projective.

3. S-semisimple modules and S-semisimple rings Let R be a ring. Recall from [14] that an R-module M is semisimple provided that it is a direct sum of simple modules. By [14, Proposition 4.1] an R-module M is semisimple if and only if every submodule is a direct summand of M. So M is semisimple if and only if any short exact sequence 0 → A → M → C → 0 is split. Utilizing this characterization, we introduce an S-analogue of semisimple modules.

Deﬁnition 3.1. Let R be a ring and S a multiplicative subset of R. An R-module M is called S-semisimple provided that any S-short exact sequence 0 → A → M → C → 0 is S-split.

Obviously, uniformly S-torsion modules are S-semisimple. Certainly, the class of S-semisimple modules is closed under S-isomorphisms. We can deduce that semisimple modules are also S-semisimple from the following lemma.

Lemma 3.2. An R-module M is S-semisimple if and only if any short exact sequence 0 → L → M → N → 0 is S-split.

Proof. Suppose M is S-semisimple, then obviously any short exact sequence 0 → f g L → M → N → 0 is S-split. Now let 0 → A −→ M −→ C → 0 be an S-short exact g1 sequence. Consider the natural exact sequence 0 → Ker(g) → M −→ Im(g) → 0. ′ ′ Then there is an R-homomorphism g1 : Im(g) → M and s ∈ S such that g1 ◦ g1 = sIdIm(g). Let i : Im(g) → C be the embedding map. Then by [19, Proposition ′ 1.1] there exists an S-isomorphism j : C → Im(g) such that i ◦ j = s IdC for some ′ ′ ′ ′ ′ s ∈ S. setting g = g1 ◦ j, we have g ◦ g = ss IdC . So the S-short exact sequence f g 0 → A −→ M −→ C → 0 is S-split by Lemma 2.2.

Proposition 3.3. Let 0 → A → B → C → 0 be an S-short exact sequence. If B is S-semisimple then A and C are S-semisimple.

Proof. Since the class of S-semisimple modules is closed under S-isomorphisms, we f g can assume that 0 → A −→ B −→ C → 0 is a short exact sequence. Suppose B is 8 a b S-semisimple. First, we will show A is S-semisimple. Let 0 → X −→ A −→ Y → 0 be an exact sequence. Consider the following pullback-pushout diagram:

0 0

X X

a f g 0 / A / B / C / 0

b d c 0 / Y / Z / C / 0

0 0

′ ′ Then there exists an R-homomorphism d : Z → B such that d ◦ d = sIdZ for some s ∈ S. Since d ◦ d′ ◦ c = sc where s means the multiplication of Y by s. Thus there ′ ′ is an R-homomorphism b : Y → A such the b ◦ b = sIdY by the universal property of pullbacks. So A is S-semisimple by Lemma 3.2. Next, we will show C also is i j S-semisimple. Let 0 → M −→ C −→ K → 0 be an exact sequence. We have the following pullback-pushout diagram:

0 0

0 / A / N / M / 0

i f g 0 / A / B / C / 0 π j K K

0 0

′ ′ ′ Then there exists an R-homomorphism π : K → B such that π ◦ π = s IdK for ′ ′ ′ ′ some s ∈ S. Thus j ◦ g ◦ π = π ◦ π = s IdK. So C also is S-semisimple by Lemma 3.2.

Recall that a ring R is semisimple provided that R is semisimple as an R-module. Note that a ring R is semisimple if and only if every free R-module is semisimple by [14, Proposition 4.5]. To give a “uniform” version of semisimple rings, we deﬁne S-semisimple rings by considering all free R-modules. 9 Deﬁnition 3.4. Let R be a ring and S a multiplicative subset of R. R is called an S-semisimple ring provided that any free R-module is S-semisimple.

Trivially, if 0 ∈ S then all rings are S-semisimple. Obviously, all semisimple rings are S-semisimple for any multiplicative subset S of R. The next result gives various of characterizations of S-semisimple rings.

Theorem 3.5. Let R be a ring and S a multiplicative subset of R. Then the following assertions are equivalent: (1) R is an S-semisimple ring; (2) every R-module is S-semisimple; (3) every S-short exact sequence is S-split; (4) every short exact sequence is S-split; 1 (5) for any R-modules M and N, ExtR(M, N)=0 is uniformly S-torsion; (6) every R-module is S-projective; (7) every R-module is S-injective; (8) R is uniformly S-Noetherian and S-von Neumann regular; (9) there exists an element s ∈ S such that for any ideal I of R there is an

R-homomorphism fI : R → I satisfying fI (i)= si for any i ∈ I.

Proof. (1) ⇒ (2): Let M be an R-module. There exists an exact sequence 0 → K → F → M → 0 with F free R-module. By Proposition 3.3, M is S-semisimple. (2) ⇒ (3): Let ξ :0 → A → B → C → 0 be an S-short exact sequence. Since B is S-semisimple, the S-short exact sequence ξ is S-split. (3) ⇒ (2): Let M be an R-module and 0 → A → M → B → 0 an S-short exact sequence. By (3), 0 → A → M → B → 0 is S-split. So M is S-semisimple. (2) ⇒ (1) and (3) ⇒ (4): Trivial.

(2) ⇒ (9): Let Γ := {I}I✂R be the set of all ideals of R. Considering the i π natural short exact sequence 0 → L I −→ L R −→ L R/I → 0, we have an R- I∈Γ Γ I∈Γ ′ ′ homomorphism i : L R → L I such that i ◦ i = sId L I for some s ∈ S. For any Γ I∈Γ I∈Γ ′ I ✂ R, set fI = iI ◦ i ◦ πI where iI : R → L R is the I-the component natural Γ embedding map and πI : L I → I is the I-the component natural projective map. I∈Γ Then we have fI (i)= si for any i ∈ I. (4) ⇒ (6): Let M be an R-module and 0 → K → P → M → 0 be a short exact sequence with P projective. Then M is S-projective by Proposition 2.8. (5) ⇔ (6): Let M be an R-module and 0 → K → P → M → 0 be an exact sequence with P projective. Let N be an R-module. Then there is a long exact 10 sequence

1 0 → HomR(M, N) → HomR(P, N) → HomR(K, N) → ExtR(M, N) → 0.

1 Thus ExtR(M, N) = 0 is uniformly S-torsion for any R-module N if and only if M 1 is S-projective. So ExtR(M, N) = 0 is uniformly S-torsion for any R-modules M and N if and only if every R-module is S-projective. (5) ⇔ (7): Similar to (5) ⇔ (6). (6) ⇒ (3): Let 0 → N → K → M → 0 be an S-short exact sequence. Since M is an S-projective module, then 0 → N → K → M → 0 is S-split by Theorem 2.5.

(4) + (6) ⇒ (8): Let Γ := {I}I✂R be the set of all ideals of R. Considering i π the natural short exact sequence 0 → L I −→ L R −→ L R/I → 0, we have an I∈Γ Γ I∈Γ ′ ′ R-homomorphism i : L R → L I such that i ◦ i = sId L I for some s ∈ S. Γ I∈Γ I∈Γ So the natural embedding map Im(i′) ֒→ L I is an S-isomorphism. Thus the set I∈Γ Γ := {I}I✂R is uniformly S-ﬁnite (see [13] for example) since the I-th component of Im(i′) is ﬁnitely generated for any ideal I of R. So R is a uniformly S-Noetherian ring. Since every R-module is S-projective, we have R is S-von Neumann regular by Proposition 2.7.

(8) ⇒ (9) : Let Γ := {I}I✂R be the set of all ideals of R. Since R is uniformly S-Noetherian, there exists an element s ∈ S such that for any ideal I ∈ Γ there is a ﬁnitely generated sub-ideal K of I satisfying sI ⊆ K. Since R is S-von Neumann regular, there is an element s′ ∈ S such that for any ﬁnitely generated ideal K of R there is an idempotent e ∈ K such that s′(K/hei) = 0 by [18, Theorem 3.13]. Let f : R → I be the R-homomorphism satisfying f(1) = e. Then we have f(i) = ss′i for any i ∈ I.

Certainly, if R is an S-semisimple ring then R is S-semisimple as an R-module. However, the following example shows that the converse does not hold in general.

Example 3.6. Let R = Z be the ring of all integers and the multiplicative subset i S = Z −{0}. Let hni be an ideal of Z and consider the exact sequence 0 →hni −→ Z → Z/hni → 0. Set i′ : Z → hni to be the Z-homomorphism satisfying i′(1) = n. Then i′ ◦ i(m)= nm for any m ∈hni. Thus Z is an S-semisimple Z-module . Since R is not S-von Neumann regular by [18, Example 3.15], R is not an S-semisimple ring by Theorem 3.5.

Moreover, the following result shows that every S-semisimple ring is in fact a semisimple ring in the case that S is composed of non-zero-divisors.

11 Proposition 3.7. Let R be a ring and S a multiplicative subset of R. Suppose S is composed of non-zero-divisors. Then R is an S-semisimple ring if and only if R is a semisimple ring.

Proof. Let R is an S-semisimple ring where each element in S is a non-zero-divisor. There exists an element s ∈ S such that for any ideal I of R there is an R- homomorphism f : R → I satisfying f(i) = si for any i ∈ I by Theorem 3.5. Set I = hs2i. Then s2f(1) = f(s2) = s3. Let f(1) = s2r ∈ I for some r ∈ R. Then s4r = s3. Since s is a non-zero-divisor, we have sr = 1 and thus s is a unit. Consequently, R is a semisimple ring.

The following nontrivial example shows that the condition that “S is composed of non-zero-divisors” in Corollary 3.7 cannot be removed.

Example 3.8. Let T = Z2 × Z2 be a semi-simple ring and s = (1, 0) ∈ T . Let R = T [x]/hsx, x2i with x the indeterminate and S = {1,s} be a multiplicative subset of R. Then R is Noetherian and S-von Neumann regular by [18, Example 3.18]. So R is an S-semisimple ring. However, R is not a semisimple ring since R is not reduced.

Let p be a prime ideal of R. We say a ring R is a p-semisimple ring shortly provided R is an (R − p)-semisimple ring. The ﬁnal result gives a new local characterization of semisimple rings.

Proposition 3.9. Let R be a ring. Then the following statements are equivalent: (1) R is a semisimple ring; (2) R is a p-semisimple ring for any p ∈ Spec(R); (3) R is an m-semisimple ring for any m ∈ Max(R).

Proof. (1) ⇒ (2) : Let P be an R-module and p ∈ Spec(R). Then P is projective, and thus is p-projective. So R is an p-semisimple ring by Theorem 3.5. (2) ⇒ (3) : Trivial. (3) ⇒ (1) : Let M be an R-module, then M is m-projective for any m ∈ Max(R). Thus M is projective by Proposition 2.10. So R is a semisimple ring.

Acknowledgement. The author was supported by the Natural Science Foundation of Chengdu Aeronau- tic Polytechnic (No. 062026) and the National Natural Science Foundation of China (No. 12061001). 12 References

[1] D. D. Anderson, T. Dumitrescu, S-Noetherian rings, Commun. Algebra 30 (2002), 4407-4416. [2] D. D. Anderson, A. Hamed and M. Zafrullah, On S-GCD domains, J. Algebra Appl. (2019)1950067 (14 pages). [3] Gobel, R., Trlifaj, J. (2012). Approximations and endomorphism algebras of modules, De Gruyter Exp. Math., vol. 41, Berlin: Walter de Gruyter GmbH & Co. KG. [4] S. Bazzoni, L. Positselski, S-almost perfect commutative rings, J. Algebra 532 (2019), 323-356. [5] D. Bennis, M. El Hajoui, On S-coherence, J. Korean Math. Soc. 55 (2018), no. 6, 1499-1512. [6] L. Fuchs, L. Salce, Modules over Non-Noetherian Domains, Providence, AMS, 2001. [7] S. Glaz, Commutative Coherent Rings, Lecture Notes in Mathematics, vol. 1371, Spring- Verlag, Berlin, 1989. [8] J. S. Golan, Torsion Theories, Pitman Monographs and Surveys in Pure and Applied Math- ematics Series, Vol. 29, New York, Longman Scientic and Technical, 1986. [9] J. W. Lim, A Note on S-Noetherian Domains, Kyungpook Math. J. 55, (2015), 507-514. [10] J. W. Lim, D. Y. Oh, S-Noetherian properties on amalgamated algebras along an ideal,J. Pure Appl. Algebra 218, (2014), 2099-2123. [11] C. Nˇastˇasescu, C. Nita, Objects noethériens par rapport àune sous-categorie épaisse dun´ catégorie abelienne, Rev. Roum. Math. Pures et Appl. 9, (1965), 1459-1468. [12] H. Kim, M. O. Kim, J. W. Lim, On S-strong Mori domains, J. Algebra 416, (2014): 314-332. [13] W. Qi, H. Kim, F. G. Wang, M. Z. Chen, W. Zhao, Uniformly S-Noetherian rings, submitted. [14] J. Rotman, An Introduction to Homological Algebra, Second edition, Universitext, Springer, New York, (2009). [15] E. S. Sevim, U. Tekir, and S. Koc, S-artinian rings and finitely S-cogenerated rings, J. Algebra Appl.(2020) 2050051 (16 pages). [16] B. Stenström, Rings of quotients, Die Grundlehren Der Mathematischen Wissenschaften, Berlin: Springer-verlag, 1975. [17] F. G. Wang, H. Kim, Foundations of Commutative Rings and Their Modules, Singapore, Springer, 2016. [18] X. L. Zhang, Characterizing S-flat modules and S-von Neumann regular rings by uniformity, https://arxiv.org/abs/2105.07941. [19] X. L. Zhang, The S-weak global dimension of commutative rings, https://arxiv.org/abs/2106.00535.