Abstract

Let 𝑅 be a commutative ring with identity, and let 𝑀 be an 𝑅 -module. A proper submodule 𝑁 of 𝑀 is said to be weakly primary if 0 𝑟 𝑚 𝑁 for 𝑟 𝑅 and 𝑚 𝑀 , which implies that either 𝑚 𝑁 or 𝑟 𝑛 𝑀 𝑁 for some positive integer 𝑛 . In this paper, we study weakly primary submodules, and we investigate the union of weakly primary submodules of 𝑅 -modules.

1. Introduction

Let 𝑅 be a commutative ring with identity, and let 𝑀 be a unital 𝑅 -module. A commutative ring 𝑅 is called a quasilocal ring if it has a unique maximal ideal 𝑃 and denoted by ( 𝑅 , 𝑃 ) . Let 𝑁 be a submodule of 𝑀 and the ideal { 𝑟 𝑅 𝑟 𝑀 𝑁 } denoted by ( 𝑁 𝑀 ) . Let 𝐼 be an ideal of 𝑅 and the radical of 𝐼 denoted by 𝐼 and defined 𝐼 = { 𝑟 𝑅 𝑟 𝑛 𝐼 f o r s o m e p o s i t i v e i n t e g e r 𝑛 } . A proper submodule 𝑁 of 𝑀 is said to be prime (weakly prime) if 𝑟 𝑚 𝑁 ( 0 𝑟 𝑚 𝑁 ), then either 𝑚 𝑁 or 𝑟 𝑀 𝑁 (either 𝑚 𝑁 or 𝑟 𝑀 𝑁 ), where 𝑟 𝑅 and 𝑚 𝑀 . A proper submodule 𝑁 of 𝑀 is said to be primary (weakly primary) if 𝑟 𝑚 𝑁 ( 0 𝑟 𝑚 𝑁 ), then either 𝑚 𝑁 or 𝑟 𝑛 𝑀 𝑁 for some positive integer 𝑛 (either 𝑚 𝑁 or 𝑟 𝑛 𝑀 𝑁 for some positive integer 𝑛 ), where 𝑟 𝑅 and 𝑚 𝑀 . It is clear that every primary submodule is weakly primary. However, since 0 is always weakly primary (by definition), so a weakly primary submodule need not be primary. A proper submodule 𝑁 of an 𝑅 -module 𝑀 said to be maximal if there is no submodule 𝐾 of 𝑀 such that 𝑁 ̸ ̸ 𝐾 𝑀 . A submodule 𝑁 of 𝑀 is called 𝑢 -submodule of 𝑀 , provided that 𝑁 contained in a finite union of submodules must be contained one of those submodules. 𝑀 is called 𝑢 -module if every submodule of 𝑀 is a 𝑢 -submodule. A submodule 𝑁 of 𝑀 is called 𝑢 𝑝 𝑟 -submodule of 𝑀 , provided that 𝑁 contained in a finite union of primary submodules that must be contained in one of those primary submodules. 𝑀 is called 𝑢 𝑝 𝑟 -module if every submodule of 𝑀 is a 𝑢 𝑝 𝑟 -submodule. A submodule 𝑁 of 𝑀 is called 𝑢 𝑚 -submodule of 𝑀 , provided that 𝑁 contained in a finite union of maximal submodules that must be contained in one of those submodules. 𝑀 is called 𝑢 𝑚 -module if every maximal submodule of 𝑀 is a 𝑢 𝑚 -submodule. An 𝑅 -module 𝑀 is called a multiplication module, provided that for each submodule 𝑁 of 𝑀 , there exists an ideal 𝐼 of 𝑅 such that 𝑁 = 𝐼 𝑀 . If 𝑅 is a ring and 𝑀 an 𝑅 -module, the subset 𝑇 ( 𝑀 ) of 𝑀 is defined by 𝑇 ( 𝑀 ) = { 𝑚 𝑀 𝑟 𝑚 = 0 f o r s o m e 0 𝑟 𝑅 } . Obviously, if 𝑅 is an integral domain, then 𝑇 ( 𝑀 ) is a submodule of 𝑀 . In this paper, we investigate finite unions of weakly primary submodules of 𝑅 -modules.

2. On Weakly Primary Submodules

It is clear that every primary submodule is a weakly primary submodule. However, since 0 is always weakly primary (by definition), a weakly primary submodule need not be primary, but we have the following results.

Proposition 2.1. Let 𝑀 be an 𝑅 -module with 𝑇 ( 𝑀 ) = 0 . Then, every weakly primary submodule of 𝑀 is primary.

Proof. Let 𝑁 be a weakly primary submodule of 𝑀 . Suppose that 𝑟 𝑚 𝑁 , where 𝑟 𝑅 , 𝑚 𝑀 . If 0 𝑟 𝑚 𝑁 , 𝑁 weakly primary gives 𝑚 𝑁 or 𝑟 𝑛 𝑀 𝑁 for some positive integer 𝑛 . If 𝑟 𝑚 = 0 , then 𝑟 = 0 or 𝑚 = 0 , since 𝑇 ( 𝑀 ) = 0 . So, 𝑁 is primary.

Proposition 2.2. Let 𝑀 be a module over a quasilocal ring ( 𝑅 , 𝑃 ) with 𝑃 𝑀 = 0 . Then, every proper submodule of 𝑀 is weakly primary.

Proof. Let 𝑁 be a proper submodule of 𝑀 and 0 𝑟 𝑚 𝑁 , where 𝑟 𝑅 and 𝑚 𝑀 . If 𝑟 is a unite, then 𝑚 𝑁 . Let 𝑟 is not a unite, so 𝑟 𝑚 𝑃 𝑀 = 0 , a contradiction. Hence, 𝑁 is weakly primary.

Lemma 2.3. Let 𝑀 be an 𝑅 -module. Assume that 𝑁 and 𝐾 are submodules of 𝑀 such that 𝐾 𝑁 with 𝑁 𝑀 . Then, the following hold:(i)if 𝑁 is a weakly primary submodule of 𝑀 , then 𝑁 / 𝐾 is a weakly primary submodule of 𝑀 / 𝐾 , (ii)if 𝐾 and 𝑁 / 𝐾 are weakly primary submodules, then 𝑁 is weakly primary.

Proof. (i) Let 0 𝑟 ( 𝑚 + 𝐾 ) = 𝑟 𝑚 + 𝐾 𝑁 / 𝐾 , where 𝑟 𝑅 and 𝑚 𝑀 . If 𝑟 𝑚 = 0 , then 𝑟 ( 𝑚 + 𝐾 ) = 0 , which is a contradiction. If 𝑟 𝑚 0 , 𝑁 weakly primary gives either 𝑚 𝑁 or 𝑟 𝑛 ( 𝑁 𝑅 𝑀 ) for some positive integer 𝑛 , hence either 𝑚 + 𝐾 𝑁 / 𝐾 or 𝑟 𝑛 ( 𝑁 / 𝐾 𝑅 𝑀 / 𝐾 ) (since we have ( 𝑁 𝑅 𝑀 ) = ( 𝑁 / 𝐾 𝑅 𝑀 / 𝐾 ) ), as required.
(ii) Let 0 𝑟 𝑚 𝑁 , where 𝑟 𝑅 and 𝑚 𝑀 , so 𝑟 ( 𝑚 + 𝐾 ) = 𝑟 𝑚 + 𝐾 𝑁 / 𝐾 . If 𝑟 𝑚 𝐾 , then 𝐾 weakly primary gives either 𝑚 𝐾 𝑁 or 𝑟 𝑛 ( 𝐾 𝑅 𝑀 ) ( 𝑁 𝑅 𝑀 ) . So, we may assume that 𝑟 𝑚 𝐾 . Then, 0 𝑟 ( 𝑚 + 𝐾 ) 𝑁 / 𝐾 . Since 𝑁 / 𝐾 is weakly primary, we get either 𝑚 + 𝑁 𝑁 / 𝐾 or 𝑟 𝑛 ( 𝑁 / 𝐾 𝑅 𝑀 / 𝐾 ) = ( 𝑁 𝑀 ) for some positive integer 𝑛 . Thus, 𝑚 𝑁 or 𝑟 𝑛 ( 𝑁 𝑅 𝑀 ) for some positive integer 𝑛 , as required.

Theorem 2.4. Let 𝑀 be a secondary 𝑅 -module and 𝑁 a nonzero weakly primary 𝑅 -submodule of 𝑀 . Then, 𝑁 is secondary.

Proof. Let 𝑟 𝑅 . If 𝑟 𝑛 𝑀 = 0 for some 𝑛 𝑁 . Then, 𝑟 𝑛 𝑁 𝑟 𝑛 𝑀 = 0 , so 𝑟 is nilpotent on 𝑁 . Suppose that 𝑟 𝑀 = 𝑀 ; we show that 𝑟 divides 𝑁 . Assume that 𝑛 𝑁 . So, 𝑛 = 𝑟 𝑚 for some 𝑚 𝑀 . We may assume that 0 𝑟 𝑚 . Hence, 0 𝑟 𝑚 𝑁 and 𝑟 𝑛 𝑀 ̸ 𝑁 for any positive integer 𝑛 (since 𝑟 𝑀 = 𝑀 ), then 𝑁 weakly primary gives 𝑚 𝑁 . Thus, 𝑟 𝑁 = 𝑁 , as needed.

Theorem 2.5. Let 𝑀 be an 𝑅 -module, 𝑁 a secondary 𝑅 -submodule of 𝑀 , and 𝐾 a weakly primary submodule of 𝑀 . Then, 𝑁 𝐾 is secondary.

Proof. The proof is straightforward.

Proposition 2.6. Let 𝑀 be a module over a commutative ring 𝑅 and 𝑆 a multiplicatively closed subset of 𝑅 . Let 𝑁 be a weakly primary submodule of 𝑀 such that ( 𝑁 𝑀 ) 𝑆 = . Then, 𝑆 1 𝑁 is a weakly primary submodule of 𝑆 1 𝑅 -module 𝑆 1 𝑀 .

Proof. Let 0 / 1 𝑟 / 𝑠 𝑚 / 𝑡 𝑆 1 𝑁 , where 𝑟 / 𝑠 𝑆 1 𝑅 and 𝑚 / 𝑡 𝑆 1 𝑀 . So, 0 / 1 𝑟 𝑚 / 𝑠 𝑡 = 𝑛 / 𝑡 for some 𝑛 𝑁 and 𝑡 𝑆 , hence there exists 𝑠 𝑆 such that 0 𝑠 𝑡 𝑟 𝑚 = 𝑠 𝑠 𝑡 𝑛 𝑁 (because if 𝑠 𝑡 𝑟 𝑚 = 0 , 𝑟 𝑚 / 𝑠 𝑡 = 𝑠 𝑡 𝑟 𝑚 / 𝑠 𝑡 𝑠 𝑡 = 0 / 1 , a contradiction) and 𝑠 𝑡 ( 𝑁 𝑀 ) , so 𝑁 weakly primary gives 0 𝑟 𝑚 𝑁 . Hence, 𝑚 𝑁 or 𝑟 𝑛 ( 𝑁 𝑀 ) , thus ( 𝑟 / 𝑠 ) 𝑛 𝑆 1 ( 𝑁 𝑅 𝑀 ) ( 𝑆 1 𝑁 𝑆 1 𝑅 𝑆 1 𝑀 ) or 𝑚 / 𝑡 𝑆 1 𝑁 , as needed.

Lemma 2.7. Let 𝑀 be a module over a quasilocal ring ( 𝑅 , 𝑃 ) and 𝑁 a weakly primary submodule of 𝑀 . Then, ( 𝑁 𝑅 𝑀 ) 𝑃 = ( 𝑁 𝑃 𝑅 𝑃 𝑀 𝑃 ) .

Proof. Let 𝑟 / 𝑠 ( 𝑁 𝑃 𝑅 𝑃 𝑀 𝑃 ) and 𝑚 𝑀 . We show that 𝑟 𝑚 𝑁 . We may assume that 𝑟 𝑚 0 . We have 𝑟 / 𝑠 𝑚 / 1 𝑆 1 𝑁 , so 𝑟 𝑚 / 𝑠 = 𝑛 / 𝑡 for some 𝑡 𝑆 and 𝑛 𝑁 . There exists 𝑡 𝑆 such that 𝑡 𝑡 𝑟 𝑚 = 𝑡 𝑠 𝑛 𝑁 . If 𝑡 𝑡 𝑟 𝑚 = 0 , then 𝑡 𝑡 ( 0 𝑟 𝑚 ) 𝑆 𝑃 𝑆 = , a contradiction. So, 0 𝑡 𝑡 𝑟 𝑚 𝑁 and 𝑡 𝑡 ( 𝑁 𝑀 ) , then 𝑟 𝑚 𝑁 . Thus, ( 𝑁 𝑃 𝑅 𝑃 𝑀 𝑃 ) ( 𝑁 𝑅 𝑀 ) 𝑃 . Clearly, ( 𝑁 𝑅 𝑀 ) 𝑃 ( 𝑁 𝑃 𝑅 𝑃 𝑀 𝑃 ) , so the proof is complete.

Theorem 2.8. Let 𝑀 be a module over a quasilocal ring ( 𝑅 , 𝑃 ) . Then, there exists a one to one correspondence between the weakly primary submodules of 𝑀 and the weakly primary submodules of 𝑅 𝑃 -module 𝑀 𝑃 .

Proof. Let 𝐾 be a weakly primary submodule of 𝑀 𝑃 . So, 𝐾 = 𝑁 𝑃 for some submodule 𝑁 of 𝑀 . We show that 𝑁 is weakly primary submodule of 𝑀 . Let 0 𝑟 𝑚 𝑁 , so 0 / 1 𝑟 𝑚 / 1 𝑁 𝑃 (if 𝑟 𝑚 / 1 = 0 / 1 , then 𝑠 𝑟 𝑚 = 0 for some 𝑠 𝑆 , 𝑠 ( 0 𝑟 𝑚 ) 𝑆 𝑃 𝑆 = , a contradiction). Hence, ( 𝑟 / 1 ) 𝑛 ( 𝑁 𝑃 𝑅 𝑃 𝑀 𝑃 ) ( 𝑁 𝑃 𝑀 ) 𝑃 for some positive integer 𝑛 by Lemma 2.7 or 𝑚 / 1 𝑁 𝑃 , since 𝑁 𝑃 is weakly primary. Thus, 𝑟 𝑛 ( 𝑁 𝑀 ) for some positive integer 𝑛 or 𝑚 𝑁 , as required. Let 𝑁 be a weakly primary submodule of 𝑀 . Then, by Proposition 2.6, 𝑁 𝑃 is weakly primary submodule of 𝑀 𝑃 .

3. Unions of Weakly Primary Submodules

Definition 3.1. Let 𝑀 be a module over a commutative ring 𝑅 and 𝑁 a submodule of 𝑀 ; 𝑁 is called a 𝑢 𝑤 𝑝 𝑟 -submodule of 𝑀 , provided that 𝑁 contained in a finite weakly primary submodules of 𝑀 must be contained in one of those weakly primary submodules. 𝑀 is called a 𝑢 𝑤 𝑝 𝑟 -module if every submodule of 𝑀 is a 𝑢 𝑤 𝑝 𝑟 -submodule.

Clearly, every 𝑢 -module is 𝑢 𝑤 𝑝 𝑟 -module, and every 𝑢 𝑤 𝑝 𝑟 -module is 𝑢 𝑝 𝑟 -module and also, every 𝑢 𝑝 𝑟 -module is 𝑢 𝑚 -module.

Theorem 3.2. Let 𝑀 be a module over a commutative ring 𝑅 and 𝐾 a submodule of 𝑀 . Then,(i)if 𝑁 is a 𝑢 𝑤 𝑝 𝑟 -submodule of 𝑀 and 𝐾 is a weakly primary submodule of 𝑀 such that 𝐾 𝑁 , then 𝑁 / 𝐾 is a 𝑢 𝑤 𝑝 𝑟 -submodule of 𝑅 -module 𝑀 / 𝐾 ,(ii)if 𝑁 / 𝐾 is a 𝑢 𝑤 𝑝 𝑟 -submodule of 𝑀 / 𝐾 , then 𝑁 is a 𝑢 𝑤 𝑝 𝑟 -submodule of 𝑀 .

Proof. (i) Let 𝑁 / 𝐾 𝐿 1 𝐿 2 𝐿 𝑛 , where 𝐿 𝑖 ’s are weakly primary submodules of 𝑀 / 𝐾 . Then, by Lemma 2.3, there exists weakly primary submodules 𝑁 𝑖 of 𝑀 such that 𝐿 𝑖 = 𝑁 𝑖 / 𝐾 for 𝑖 = 1 , 2 , , 𝑛 . So, 𝑁 / 𝐾 𝑁 1 / 𝐾 𝑁 𝑛 / 𝐾 = 𝑁 1 𝑁 𝑛 / 𝐾 , so 𝑁 𝑁 1 𝑁 2 𝑁 𝑛 . Hence, 𝑁 𝑁 𝑘 for some 𝑘 , since 𝑁 is 𝑢 𝑤 𝑝 𝑟 -submodule. Thus 𝑁 / 𝐾 𝑁 𝑘 / 𝐾 , for some 𝑘 , as needed.
(ii) Let 𝑁 𝑁 1 𝑁 2 𝑁 𝑛 , where 𝑁 𝑖 ’s are weakly primary submodule of 𝑀 . So, 𝑁 / 𝐾 𝑁 1 𝑁 𝑛 / 𝐾 = 𝑁 1 / 𝐾 𝑁 𝑛 / 𝐾 . Therefore, 𝑁 𝑖 / 𝐾 ’s are weakly primary submodules of 𝑀 / 𝐾 , so N / 𝐾 𝑁 𝑘 / 𝐾 for some 𝑘 . Hence, 𝑁 𝑁 𝑘 for some 𝑘 , as required.

Theorem 3.3. Let 𝑀 be an 𝑅 -module, 𝑁 a weakly primary submodule of 𝑀 , and 𝑀 a 𝑢 𝑤 𝑝 𝑟 -module. Then, 𝑀 / 𝑁 is a 𝑢 𝑤 𝑝 𝑟 -module.

Proof. By Theorem 3.2.

Theorem 3.4. Let 𝑀 be a module over quasi local ring ( 𝑅 , 𝑃 ) . Then, 𝑀 is a 𝑢 𝑤 𝑝 𝑟 -module if and only if 𝑅 𝑃 module 𝑀 𝑃 is a 𝑢 𝑤 𝑝 𝑟 -module.

Proof. Let 𝑀 be a 𝑢 𝑤 𝑝 𝑟 -module, and let 𝐾 be a submodule of 𝑅 𝑃 -module 𝑀 𝑃 such that 𝐾 𝐾 1 𝐾 𝑛 , where 𝐾 𝑖 ’s are weakly primary submodules of 𝑀 𝑃 . So, 𝐾 = 𝑁 𝑃 for some submodule 𝑁 of 𝑀 and 𝐾 𝑖 = ( 𝑁 𝑖 ) 𝑃 for some weakly primary submodules 𝑁 i of 𝑀 by Theorem 2.8. Hence, 𝑁 𝑃 ( 𝑁 1 ) 𝑃 ( 𝑁 𝑛 ) 𝑃 , so 𝑁 𝑃 ( 𝑁 1 𝑁 𝑛 ) 𝑃 , thus 𝑁 𝑁 1 𝑁 𝑛 by [1, Theorem  2.8]. Therefore, by hypothesis 𝑁 𝑁 𝑘 for some 1 𝑘 𝑛 . So 𝑁 𝑃 ( 𝑁 𝑘 ) 𝑃 for some 1 𝑘 𝑛 as needed.
Conversely, let 𝑅 𝑃 -module 𝑀 𝑃 is a 𝑢 𝑤 𝑝 𝑟 -module, and let 𝑁 be a submodule of 𝑀 such that 𝑁 𝑁 1 𝑁 𝑛 , where 𝑁 𝑖 ’s are weakly primary submodules of 𝑀 . So, 𝑁 𝑃 ( 𝑁 1 𝑁 𝑛 ) 𝑃 = ( 𝑁 1 ) 𝑃 ( 𝑁 𝑛 ) 𝑃 . Thus, 𝑁 𝑃 ( 𝑁 𝑘 ) 𝑃 for some 1 𝑘 𝑛 by hypothesis. Then, 𝑁 𝑁 𝑘 for some 1 𝑘 𝑛 . So 𝑀 is a 𝑢 𝑤 𝑝 𝑟 -module.

Theorem 3.5. Let 𝑀 be a finitely generated 𝑅 -module. Then, 𝑀 is a 𝑢 𝑚 -module if and only if every submodule 𝑁 in 𝑀 such that 𝑁 𝑛 𝑖 = 1 𝑃 𝑖 , where 𝑃 𝑖 ’s are weakly primary submodules implies that 𝑁 + 𝑃 𝑖 𝑀 for some 𝑖 .

Proof. Let 𝑀 be a finitely generated 𝑢 𝑚 -module. Suppose that 𝑁 be a submodule of 𝑀 such that 𝑁 𝑛 𝑖 = 1 𝑃 𝑖 , where 𝑃 𝑖 ’s are weakly primary submodules of 𝑀 . For each 𝑃 𝑖 , 𝑀 𝑖 is a maximal submodule containing 𝑃 𝑖 . Then, 𝑁 𝑛 𝑖 = 1 𝑀 𝑖 , and so, 𝑁 𝑀 𝑖 for some 𝑖 by hypothesis. Since 𝑃 𝑖 𝑀 𝑖 , we have 𝑁 + 𝑃 𝑖 𝑀 𝑖 𝑀 .
Conversely, let 𝑁 be a submodule of 𝑀 such that 𝑁 𝑛 𝑖 = 1 𝑀 𝑖 , where 𝑀 𝑖 ’s are maximal submodules of 𝑀 . Since every maximal submodule is weakly primary submodule, then 𝑁 + 𝑀 𝑖 𝑀 for some 𝑖 by hypothesis. Therefore, since 𝑀 𝑖 𝑁 + 𝑀 𝑖 / 𝑀 , then 𝑁 + 𝑀 𝑖 = 𝑀 𝑖 , so 𝑁 𝑀 𝑖 for some 𝑖 . The proof is complete.

Definition 3.6. By a chain of weakly primary submodules of an 𝑅 -module 𝑀 , we mean a finite strictly increasing sequence 𝑃 1 𝑃 𝑛 ; the weakly primary dimension of this chain is 𝑛 . We define the weakly primary dimension of 𝑀 to be the supremum of the lengths of all chains of weakly primary submodules in 𝑀 .

Theorem 3.7. Let 𝑀 be a finitely generated 𝑅 -module with weakly primary dimension 1. Then, 𝑀 is a 𝑢 𝑤 𝑝 𝑟 -module if and only if 𝑀 is a 𝑢 𝑚 -module.

Proof. Let 𝑀 be a 𝑢 𝑤 𝑝 𝑟 -module. Since every 𝑢 𝑤 𝑝 𝑟 -module is a 𝑢 𝑚 -module, so 𝑀 is a 𝑢 𝑚 -module.
Conversely, let 𝑀 be a 𝑢 𝑚 -module. Let 𝑁 be a nonzero submodule of 𝑀 such that 𝑁 𝑃 1 𝑃 2 𝑃 𝑛 , where 𝑃 𝑖 ’s are weakly primary submodules of 𝑀 . We may assume that 𝑃 𝑖 0 for all 𝑖 { 1 , 2 , , 𝑛 } . By Theorem 3.5, 𝑁 + 𝑃 𝑖 𝑀 for some 𝑖 . There exists a maximal submodule 𝑃 𝑖 of 𝑀 such that 𝑁 + 𝑃 𝑖 𝑀 𝑖 . Since 0 is a weakly primary submodule of 𝑀 , so we have 0 / 𝑁 𝑖 𝑀 𝑖 . Hence, since weakly primary dimension of 𝑀 is 1; 𝑃 𝑖 = 𝑀 𝑖 . Consequently, 𝑁 𝑃 𝑖 for some 𝑖 , as needed.

Theorem 3.8. Let 𝑀 be an 𝑅 -module with 𝑇 ( 𝑀 ) = 0 . Then, 𝑀 is a 𝑢 𝑝 𝑟 -module if and only if 𝑀 is a 𝑢 𝑤 𝑝 𝑟 -module.

Proof. Let 𝑀 be a 𝑢 𝑤 𝑝 𝑟 -module. Then, 𝑀 is a 𝑢 𝑝 𝑟 -module since every primary submodule is weakly primary. Let 𝑀 be a 𝑢 𝑝 -module and 𝑁 a submodule of 𝑀 such that 𝑁 𝑃 1 𝑃 2 𝑃 𝑛 where 𝑃 𝑖 ’s are weakly primary submodules of 𝑀 . By Proposition 2.1, 𝑃 𝑖 ’s are primary submodule, so 𝑁 𝑃 𝑖 for some 𝑖 , as needed.