Uploaded by Refaat Salem

generalized power series over zip and weak zip ring1

advertisement
Southeast Asian Bulletin of Mathematics (2013) 37: 259–268
Southeast Asian
Bulletin of
Mathematics
c SEAMS. 2013
⃝
Generalized Power Series over Zip and Weak Zip Rings
R.M. Salem
Department of Mathematics, Faculty of Science, Al-Azhar University, Nasr City 11884,
Cairo, Egypt
Email: rsalem 02@hotmail.com
Received 15 July 2011
Accepted 21 March 2012
Communicated by N.V. Sanh
AMS Mathematics Subject Classification(2000): 06F05, 16W60, 16D25, 16P60
Abstract. In this paper we show that, if 𝑅 is a generalized Armendariz (semicommutative right Noetherian ) ring and 𝑆 is a strictly (strictly totally) ordered monoid, then
the generalized power series ring Λ = [[𝑅𝑆,≤ ]] is a right zip (weak zip) ring if and only
if 𝑅 is.
Keywords: Strictly ordered monoid; Artinian and narrow subset; Generalized power
series rings; Zip rings, Weak zip rings; Armendari rings; Semicommutative rings.
1. Introduction
The concept of zip rings, was initiated by Zelmanowitz [16] although it wasn’t
so called zip at that time. However, Faith in [4] called a ring 𝑅 right zip if the
right annihilator π‘Ÿπ‘… (𝑋) of a subset 𝑋 ⊆ 𝑅 is zero, then π‘Ÿπ‘… (𝑋0 ) = 0 for a finite
subset 𝑋0 of 𝑋, equivalently for a left ideal 𝐿 of 𝑅, if the π‘Ÿπ‘… (𝐿) = 0, then there
exists a finitely generated left ideal 𝐿1 ⊆ 𝐿 such that π‘Ÿπ‘… (𝐿1 ) = 0. 𝑅 is called a
zip ring if it is both right and left zip. In [16] Zelmanowitz showed that any ring
satisfying the descending chain condition on right annihilators is a right zip ring
but the converse doesn’t hold. Extensions of zip rings were studied by several
authors, Beach and Blair [1] showed that if 𝑅 is a commutative zip ring, then
𝑅[π‘₯] is a zip ring.
∑
∑
∑π‘ž
∑𝑙
𝑗
′
𝑖
′
𝑗
Now, let 𝑓 = 𝑛𝑖=0 π‘Žπ‘– π‘₯𝑖 , 𝑔 = π‘š
𝑗=0 𝑏𝑗 π‘₯ (𝑓 =
𝑖=𝑠 π‘Žπ‘– π‘₯ , 𝑔 =
𝑗=𝑑 𝑏𝑗 π‘₯ ) be
polynomials over a ring 𝑅. Rege and Chhawchharia in [10] introduced the notion
of an Armendariz ring, a ring 𝑅 is called Armendariz if whenever 𝑓, 𝑔 ∈ 𝑅[π‘₯]
such that 𝑓 𝑔 = 0 then π‘Žπ‘– 𝑏𝑗 = 0 for each 0 ≤ 𝑖 ≤ 𝑛 and 0 ≤ 𝑗 ≤ π‘š. In [8,
260
R.M. Salem
Theorem 1] Hong et al showed that if 𝑅 is an Armendariz ring, then 𝑅 is a right
zip ring if and only if 𝑅[π‘₯] is a right zip ring.
In [3, Theorem 2.8], W. Cortes studied skew polynomial extension over zip
rings, in general he showed that:
(i) If 𝜎 is an automorphism of 𝑅 and 𝑓, 𝑔 ∈ 𝑅[π‘₯, 𝜎] (𝑓 ′ , 𝑔 ′ ∈ 𝑅[π‘₯, π‘₯−1 , 𝜎]) such
that 𝑓 𝑔 = 0 (𝑓 ′ 𝑔 ′ = 0), implies π‘Žπ‘– 𝜎 𝑖 (𝑏𝑗 ) = 0 for each 0 ≤ 𝑖 ≤ 𝑛 and
0 ≤ 𝑗 ≤ 𝑛 (𝑠 ≤ 𝑖 ≤ π‘ž and 𝑑 ≤ 𝑗 ≤ 𝑙). Then the following are equivalent:
(a) 𝑅 is a right zip ring
(b) 𝑅[π‘₯, 𝜎] is a right zip ring
(c) 𝑅[π‘₯, π‘₯−1 , 𝜎] is a right zip ring
(ii) If 𝜎 is an automorphism of 𝑅 and 𝑓, 𝑔 ∈ 𝑅[[π‘₯, 𝜎]] (𝑓 ′ , 𝑔 ′ ∈ 𝑅[[π‘₯, π‘₯−1 , 𝜎]])
such that 𝑓 𝑔 = 0, implies π‘Žπ‘– 𝜎 𝑖 (𝑏𝑗 ) = 0 for each 𝑖 ≥ 0 and 𝑗 ≥ 0 (𝑖 ≥ 𝑠 and
𝑗 ≥ 𝑑). Then the following are equivalent:
(a) 𝑅 is right zip
(b) 𝑅[[π‘₯, 𝜎]] is right zip
(c) 𝑅[[π‘₯, π‘₯−1 , 𝜎]] is right zip
Motivated by the above L. Ouyang in [9] introduced the notion of right weak
zip rings, i.e., rings provided that if the right weak annihilator of a subset 𝑋
of 𝑅, 𝑁 π‘Ÿπ‘… (𝑋) ⊆ 𝑛𝑖𝑙(𝑅), then there exists a finite subset 𝑋0 ⊆ 𝑋 such that
𝑁 π‘Ÿπ‘… (𝑋0 ) ⊆ 𝑛𝑖𝑙(𝑅), where 𝑛𝑖𝑙(𝑅) is the set of all nilpotent elements of 𝑅 and
𝑁 π‘Ÿπ‘… (𝑋) = {π‘Ž ∈ π‘…βˆ£π‘₯π‘Ž ∈ 𝑛𝑖𝑙(𝑅) for each π‘₯ ∈ 𝑋}. The author in [9] studied the
relationship between right (left) weak zip property of 𝑅 and the Ore extension
𝑅[π‘₯, 𝜎, 𝛿]. He showed the following results:
(i) A ring 𝑅 is right (left) weak zip if and only if the upper triangular matrix
𝑇𝑛 (𝑅) is right (left) weak zip, where 𝑛 ≥ 2.
(ii) If 𝑅 is a (𝜎, 𝛿)-compatible and reversible ring, then 𝑅 is a right (left) weak
zip ring if and only if 𝑅[π‘₯, 𝜎, 𝛿] is right (left) weak zip. A ring 𝑅 is called 𝜎compatible if for π‘₯, 𝑦 ∈ 𝑅, π‘₯𝑦 = 0 ⇔ π‘₯πœŽπ‘¦ = 0 and 𝑅 is called 𝛿-compatible
if π‘₯𝑦 = 0 ⇒ π‘₯𝛿𝑦 = 0. If 𝑅 is both 𝜎-compatible and 𝛿-compatible, then 𝑅
is called (𝜎, 𝛿)-compatible.
We recall the following notions for a ring 𝑅.
(i) 𝑅 is reduced if for π‘₯ ∈ 𝑅 such that π‘₯2 = 0, then π‘₯ = 0
(ii) 𝑅 is reversible if for π‘₯, 𝑦 ∈ 𝑅 such that π‘₯𝑦 = 0, then 𝑦π‘₯ = 0
(iii) 𝑅 is semicommutative if for π‘₯, 𝑦 ∈ 𝑅 such that π‘₯𝑦 = 0, then π‘₯𝑅𝑦 = 0
The above definitions implies that, the class of reduced rings is a subclass of
the class of reversible rings which is a subclass of the class of semicommutative
rings.
In [11, 12], P. Ribenboim studied extensively the rings of generalized power
series.
The motivation of this paper is to continue the studying of the transfer of
some algebraic properties between the base ring 𝑅 and the generalized power
series ring [[𝑅𝑆,≤ ]] see [6, 13] and to extend W. Cortes & L. Ouyang results to
the generalized power series over zip and weak zip rings.
Generalized Power Series
261
Throughout this paper 𝑅 is an associative ring with identity and (𝑆, ≤) is
a strictly ordered monoid (i.e. a monoid 𝑆 with a compatible order ≤ such
that if π‘Ž < 𝑏 in 𝑆, then π‘Ž + 𝑠 < 𝑏 + 𝑠 for all 𝑠 ∈ 𝑆). By a generalized power
series ring Λ = [[𝑅𝑆,≤ ]] we mean the set of all functions 𝑓 : 𝑆 → 𝑅 such that
𝑠𝑒𝑝𝑝(𝑓 ) = {𝑠 ∈ π‘†βˆ£π‘“ (𝑠) βˆ•= 0} is artinian and narrow, i.e. every strictly decreasing
sequence of elements of 𝑠𝑒𝑝𝑝(𝑓 ) is finite and every subset of pairwise orderincomparable elements of 𝑠𝑒𝑝𝑝(𝑓 ) is finite, with pointwise addition and product
operation called convolution defined by
∑
(𝑓 𝑔)(𝑠) =
𝑓 (𝑒)𝑔(𝑣)
for each 𝑓, 𝑔 ∈ Λ
(𝑒,𝑣)∈𝑋𝑠 (𝑓,𝑔)
where
𝑋𝑠 (𝑓, 𝑔) = {(𝑒, 𝑣) ∈ 𝑆 × π‘†βˆ£π‘’ + 𝑣 = 𝑠 and 𝑓 (𝑒) and 𝑔(𝑣) are nonzero}
is a finite set see [11]. By [11, 2.1] Λ = [[𝑅𝑆,≤ ]] becomes a ring called a generalized
power series ring with coefficients in 𝑅 and exponents in 𝑆, with the identity
map 𝑒 : 𝑆 → 𝑅 defined by 𝑒(0) = 1 and 𝑒(𝑠) = 0 for each nonzero 𝑠 ∈ 𝑆.
The mapping 𝑒𝑠 ∈ Λ defined by 𝑒𝑠 (𝑠) = 1 and 𝑒𝑠 (𝑑) = 0 for each 𝑑 βˆ•= 𝑠 is an
embedding of 𝑆 into Λ. Also, 𝑅 is canonically embedded as a subring of Λ via
π‘Ÿ 7→ π‘π‘Ÿ such that π‘π‘Ÿ (0) = π‘Ÿ and π‘π‘Ÿ (𝑠) = 0 for each nonzero 𝑠 ∈ 𝑆. So, 𝑅 ≃ 𝑐𝑅
and we can identify π‘Ÿ ∈ 𝑅 with π‘π‘Ÿ ∈ Λ.
Let πœ‹(𝑓 ) denote the set of all minimal elements of 𝑠𝑒𝑝𝑝(𝑓 ). If (𝑆, ≤) is strictly
totally ordered, then πœ‹(𝑓 ) consists of only one element which is still denoted by
πœ‹(𝑓 ).
So, the structure of generalized power series rings generalizes the structure
of some classical rings such as polynomial rings, formal power series rings, group
rings and semigroup rings.
If 𝑅 is a ring and 𝑆 is a strictly ordered monoid, then the ring 𝑅 is called
a generalized Armendariz ring if for each 𝑓, 𝑔 ∈ Λ = [[𝑅𝑠,≤ ]] such that 𝑓 𝑔 = 0
implies that 𝑓 (𝑒𝑖 )𝑔(𝑣𝑗 ) = 0 for each 𝑒𝑖 ∈ 𝑠𝑒𝑝𝑝(𝑓 ) and 𝑣𝑗 ∈ 𝑠𝑒𝑝𝑝(𝑔). In [14]
Zhongkui called such ring 𝑠-Armendariz ring.
Let 𝑇 = 𝐢(𝑓 ) be the content of 𝑓 , i.e. 𝐢(𝑓 ) = {𝑓 (𝑠)βˆ£π‘  ∈ 𝑠𝑒𝑝𝑝(𝑓 )} ⊆ 𝑅.
Since, 𝑅 ≃ 𝑐𝑅 we can identify, the content of 𝑓 with
𝑐𝐢(𝑓 ) = {𝑐𝑓 (𝑒𝑖 ) βˆ£π‘’π‘– ∈ 𝑠𝑒𝑝𝑝(𝑓 )} ⊆ Λ
2. Generalized Power Series Over Right Zip Ring
Hirano [7] and W. Cortes [3] observed relations between the right (left) annihilators in the ring 𝑅 and the right (left) annihilators in 𝑅[π‘₯] and 𝑅[π‘₯, 𝜎]
respectively.
In this note we investigate the relations between the right (left) annihilators in
the ring 𝑅 and the right (left) annihilators in the ring Λ = [[𝑅𝑆,≤ ]] of generalized
power series.
262
R.M. Salem
Lemma 2.1. Let 𝑅 be a ring, 𝑆 a strictly ordered monoid, Λ = [[𝑅𝑆,≤ ]] the
generalized power series ring and π‘ˆ ⊆ 𝑅. Then π‘ŸΛ (π‘ˆ ) = π‘Ÿπ‘… (π‘ˆ )Λ.
Proof. Let 𝑔 ∈ π‘ŸΛ (π‘ˆ ). Then for each 𝑒 ∈ π‘ˆ , 𝑐𝑒 𝑔 = 0. Thus, 0 = (𝑐𝑒 𝑔)(𝑣) =
𝑒𝑔(𝑣) = 0, for each 𝑣 ∈ 𝑠𝑒𝑝𝑝(𝑔) and we obtain that 𝑔(𝑣) ∈ π‘Ÿπ‘… (𝑒), for each
𝑣 ∈ 𝑠𝑒𝑝𝑝(𝑔). Consequently, 𝑔 ∈ π‘Ÿπ‘… (𝑒)Λ and we have that π‘ŸΛ (π‘ˆ ) ⊆ π‘Ÿπ‘… (π‘ˆ )Λ.
On the other hand, let 𝑔 ∈ π‘Ÿπ‘… (π‘ˆ )Λ. Then for each 𝑣 ∈ 𝑠𝑒𝑝𝑝(𝑔) we have that
𝑔(𝑣) ∈ π‘Ÿπ‘… (π‘ˆ ). Thus for each 𝑒 ∈ π‘ˆ we have that 0 = 𝑒𝑔(𝑣) = (𝑐𝑒 𝑔)(𝑣) which
implies that 𝑔 ∈ π‘ŸΛ (𝑒). Hence, 𝑔 ∈ π‘ŸΛ (π‘ˆ ). So, π‘Ÿπ‘… (π‘ˆ )Λ ⊆ π‘ŸΛ (π‘ˆ ).
With the above lemma we have a map πœ™ = π‘Ÿπ‘… (2𝑅 ) → π‘ŸΛ (2Λ ) defined by
πœ™(𝐼) = πΌΛ for every 𝐼 ∈ π‘Ÿπ‘… (2𝑅 ) = {π‘Ÿπ‘… (π‘ˆ )βˆ£π‘ˆ ⊆ 𝑅}. Obviously, πœ™ is injective.
In the following lemma we show that πœ™ is a bijective map if and only if 𝑅 is a
generalized Armendariz ring.
Lemma 2.2. Let 𝑅 be a ring, 𝑆 a strictly ordered monoid and Λ = [[𝑅𝑠,≤ ]] the
ring of generalized power series. Then the following are equivalent:
(1) 𝑅 is a generalized Armendariz ring.
(2) The function πœ™ : π‘Ÿπ‘… (2𝑅 ) → π‘ŸΛ (2Λ ) is bijective, where πœ™(𝐼) = 𝐼Λ
Proof. Let π‘Œ ⊆ Λ and 𝑇 = ∪𝑓 ∈π‘Œ 𝐢(𝑓 ). From Lemma 2.1 it is sufficient to show
that π‘ŸΛ (𝑓 ) = π‘Ÿπ‘… (𝐢(𝑓 ))Λ for all 𝑓 ∈ π‘Œ . In fact, let 𝑔 ∈ π‘ŸΛ (𝑓 ). Then 𝑓 𝑔 = 0
and by assumption 𝑓 (𝑒𝑖 )𝑔(𝑣𝑗 ) = 0 for each 𝑒𝑖 ∈ 𝑠𝑒𝑝𝑝(𝑓 ) and each 𝑣𝑗 ∈ 𝑠𝑒𝑝𝑝(𝑔).
Then for a fixed 𝑒𝑖 and each 𝑣𝑗 ∈ 𝑠𝑒𝑝𝑝(𝑔), 0 = 𝑓 (𝑒𝑖 )𝑔(𝑣𝑗 ) = (𝑐𝑓 (𝑒𝑖 ) 𝑔)(𝑣𝑖 ) and it
follows that 𝑔 ∈ π‘Ÿπ‘… ∪𝑒𝑖 ∈𝑠𝑒𝑝𝑝(𝑓 ) 𝑐𝑓 (𝑒𝑖 ) Λ = π‘Ÿπ‘… 𝐢(𝑓 )Λ. So π‘ŸΛ (𝑓 ) ⊆ π‘Ÿπ‘… 𝐢(𝑓 )Λ.
Conversely,
( let )𝑔 ∈ π‘Ÿπ‘… 𝐢(𝑓 )Λ, then 𝑐𝑓 (𝑒𝑖 ) 𝑔 = 0 for each 𝑒𝑖 ∈ 𝑠𝑒𝑝𝑝(𝑓 ).
Hence, 0 = 𝑐𝑓 (𝑒𝑖∑
) 𝑔 (𝑣𝑗 ) = 𝑓 (𝑒𝑖 )𝑔(𝑣𝑗 ) for each 𝑒𝑖 ∈ 𝑠𝑒𝑝𝑝(𝑓 ) and 𝑣𝑗 ∈ 𝑠𝑒𝑝𝑝(𝑔).
Thus, (𝑓 𝑔)(𝑠) = (𝑒𝑖 ,𝑣𝑗 )∈𝑋𝑠 (𝑓,𝑔) 𝑓 (𝑒𝑖 )𝑔(𝑣𝑗 ) = 0 and it follows that 𝑔 ∈ π‘ŸΛ (𝑓 ).
Hence π‘Ÿπ‘… 𝐢(𝑓 )Λ ⊆ π‘ŸΛ (𝑓 ) and it follows that π‘Ÿπ‘… 𝐢(𝑓 )Λ = π‘ŸΛ (𝑓 ). So π‘ŸΛ (π‘Œ ) =
∩𝑓 ∈π‘Œ π‘ŸΛ (𝑓 ) = ∩𝑓 ∈π‘Œ π‘Ÿπ‘… 𝐢(𝑓 )Λ = π‘Ÿπ‘… (𝑇 )Λ.
Now, let 𝑓, 𝑔 ∈ Λ be such that 𝑓 𝑔 = 0. Then 𝑔 ∈ π‘ŸΛ (𝑓 ) and by assumption
π‘ŸΛ (𝑓 ) = 𝑇 Λ for some right ideal 𝑇 of 𝑅. Consequently, 0 = 𝑓 𝑐𝑔(𝑣𝑗 ) and for
any 𝑒𝑖 ∈ 𝑠𝑒𝑝𝑝(𝑓 ), 0 = (𝑓 𝑐𝑔(𝑣𝑗 ) )(𝑒𝑖 ) = 𝑓 (𝑒𝑖 )𝑔(𝑣𝑗 ) for each 𝑒𝑖 ∈ 𝑠𝑒𝑝𝑝(𝑓 ) and
𝑣𝑗 ∈ 𝑠𝑒𝑝𝑝(𝑔). Hence, 𝑅 is a generalized Armendariz ring.
Now, we are able to prove the main theorem of this section.
Theorem 2.3. Let 𝑅 be a generalized Armendarinz ring, 𝑆 a strictly ordered
monoid and Λ = [[𝑅𝑠,≤ ]] the ring of generalized power series. Then Λ is a right
zip ring of and only if 𝑅 is.
Proof. Suppose that Λ is a right zip ring and 𝑋 ⊆ 𝑅 such that π‘Ÿπ‘… (𝑋) = 0. Let
π‘Œ = {𝑐π‘₯ ∣π‘₯ ∈ 𝑋}. Then it can be easily shown that π‘ŸΛ (π‘Œ ) = 0. Since, Λ is a
right zip ring, then there exists a finite subset π‘Œ0 ⊆ π‘Œ such that π‘ŸΛ (π‘Œ0 ) = 0. Let
Generalized Power Series
263
π‘Œ0 = 𝑐𝑋0 = {𝑐π‘₯𝑖 βˆ£π‘– = 1, ⋅ ⋅ ⋅ , 𝑛} and 0 βˆ•= π‘Ž ∈ π‘Ÿπ‘… (𝑋0 ). Then 0 = π‘₯𝑖 π‘Ž = (𝑐π‘₯𝑖 π‘π‘Ž )(0)
for each 𝑖 ∈ {1, ⋅ ⋅ ⋅ , 𝑛}. Hence 0 = 𝑐𝑋0 π‘π‘Ž and it follows that π‘π‘Ž ∈ π‘ŸΛ π‘π‘‹0 =
π‘ŸΛ π‘Œ0 = 0. So, π‘Ž = 0 and 𝑅 is a right zip ring.
Conversely, suppose that 𝑅 is a right zip ring and π‘Œ ⊆ Λ be such that
π‘ŸΛ (π‘Œ ) = 0 and 𝑇 = 𝐢(π‘Œ ) be the content of each 𝑓 ∈ π‘Œ , i.e., 𝑇 = ∪𝑓 ∈π‘Œ {𝑓 (𝑒)βˆ£π‘’ ∈
𝑠𝑒𝑝𝑝(𝑓 )}. Then by Lemma 2.2 π‘ŸΛ (π‘Œ ) = π‘Ÿπ‘… (𝑇 )Λ = 0 and it follows that π‘Ÿπ‘… 𝑇 =
0. Since, 𝑅 is a right zip ring there exists a finite subset 𝑇0 ⊆ 𝑇 such that
π‘Ÿπ‘… (𝑇0 ) = 0. Let π‘Žπ‘– ∈ 𝑇0 , where 𝑇0 can be written as 𝑇0 = {π‘Žπ‘– βˆ£π‘– ∈ {1, ⋅ ⋅ ⋅ , 𝑛}},
π‘“π‘Žπ‘– ∈ π‘Œ be such that π‘Žπ‘– ∈ {π‘“π‘Žπ‘– (𝑒)βˆ£π‘’ ∈ 𝑠𝑒𝑝𝑝(π‘“π‘Žπ‘– )}, π‘Œ0 be a minimal subset of π‘Œ
which contains {π‘“π‘Žπ‘– ∈ π‘Œ βˆ£π‘Žπ‘– ∈ 𝑇0 } and let 𝑇1 = ∪π‘“π‘Žπ‘– {π‘“π‘Žπ‘– (𝑒)βˆ£π‘’ ∈ 𝑠𝑒𝑝𝑝(π‘“π‘Žπ‘– )}. So,
𝑇0 ⊆ 𝑇1 and it follows that π‘Ÿπ‘… 𝑇1 ⊆ π‘Ÿπ‘… 𝑇0 = 0. So, by lemma 2.2 we have that
π‘ŸΛ (π‘Œ0 ) = π‘Ÿπ‘… (𝑇1 )Λ = 0 and it follows that Λ is a right zip ring.
3. Generalized Power Series Over Right Weak Zip Ring
Proposition 3.1. Let 𝑅 be a semicommutative ring, (𝑆, +, ≤) a strictly totally
ordered monoid and Λ = [[𝑅𝑠,≤ ]] the generalized power series ring. If 𝑓 ∈ Λ is
nilpotent, then 𝑓 (𝑒) ∈ 𝑅 is nilpotent for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ).
Proof. Let 𝑓 ∈ Λ be a nilpotent element. Then there exists π‘˜ ∈ β„• such that
𝑓 π‘˜ = 0. Since, 𝑆 is a strictly totally
∑ ordered monoid, then πœ‹(𝑓 ) = 𝑒0 . By the fact
that, 0 = 𝑓 π‘˜ (π‘˜π‘’0 ) = (𝑓 (𝑒0 ))π‘˜ + (𝑑1 ,⋅⋅⋅ ,π‘‘π‘˜ )∈π‘‹π‘˜π‘’ (𝑓,⋅⋅⋅ ,𝑓 )−{(𝑒0 ,⋅⋅⋅ ,𝑒0 )} 𝑓 (𝑑1 ) ⋅ ⋅ ⋅ 𝑓 (π‘‘π‘˜ )
0
and πœ‹(𝑓 ) = 𝑒0 , there exists 𝑖 ∈ {1, ⋅ ⋅ ⋅ , π‘˜} such that 𝑑𝑖 ≥ 𝑒0 . Thus, 𝑒0 +⋅ ⋅ ⋅+𝑒0 ≤
𝑑1 + ⋅ ⋅ ⋅ + π‘‘π‘˜ = 𝑒0 + ⋅ ⋅ ⋅ + 𝑒0 = π‘˜π‘’0 a contradiction. So, 𝑑𝑖 = 𝑒0 for each
𝑖 ∈ {1, ⋅ ⋅ ⋅ , π‘˜} and it follows that 0 = 𝑓 π‘˜ (π‘˜π‘’0 ) = (𝑓 (𝑒0 ))π‘˜ and 𝑓 (𝑒0 ) is nilpotent.
Consider now, 𝑓 = (𝑓 − 𝑐𝑓 (𝑒0 ) 𝑒𝑒0 ) + 𝑐𝑓 (𝑒0 ) 𝑒𝑒0 = (𝑓 − 𝑓0′ ) + 𝑓0′ = 𝑓0 + 𝑓0′ where
𝑠𝑒𝑝𝑝(𝑓0 ) = 𝑠𝑒𝑝𝑝(𝑓 ) − {𝑒0 } and 𝑠𝑒𝑝𝑝(𝑓0′ ) = {𝑒0 }.
Hence,
0 =𝑓 π‘˜ = (𝑓0 + 𝑓0′ )π‘˜ = (𝑓0 + 𝑓0′ )(𝑓0 + 𝑓0′ ) ⋅ ⋅ ⋅ (𝑓0 + 𝑓0′ )
=𝑓0π‘˜ + 𝑓0 𝑓0′π‘˜−1 + ⋅ ⋅ ⋅ + 𝑓0′π‘˜−2 𝑓0 𝑓0′ + 𝑓0′π‘˜−1 𝑓0 + 𝑓02 𝑓0′π‘˜−2 + 𝑓0 𝑓0′π‘˜−2 𝑓0 + ⋅ ⋅ ⋅
+ 𝑓0′π‘˜−2 𝑓02 + ⋅ ⋅ ⋅ + 𝑓0′π‘˜ = 𝑓0π‘˜ + Δ + 𝑓0′π‘˜ ,
where 𝑠𝑒𝑝𝑝𝑓0′π‘˜ ⊆ 𝑠𝑒𝑝𝑝𝑓0′ + ⋅ ⋅ ⋅ + 𝑠𝑒𝑝𝑝𝑓0′ π‘˜-times
.
⊆ π‘˜ 𝑠𝑒𝑝𝑝𝑓0′ = {π‘˜π‘’0 }.
Thus, 0 = 𝑓0′π‘˜ (π‘˜π‘’0 ) = (𝑓 (𝑒0 ))π‘˜ and 𝑓0′ is nilpotent.
Now, it is clear that Δ is the sum of monomials each monomial is the product
of β„“ copies of 𝑓0 and (π‘˜ − β„“) copies of 𝑓0′ , where 𝑠𝑒𝑝𝑝 each monomial ⊆ the sum
of β„“ copies of 𝑠𝑒𝑝𝑝𝑓0 and (π‘˜ − β„“) copies of 𝑠𝑒𝑝𝑝𝑓0′ .
Since, 𝑓 ′ (𝑒0 ) = 𝑓 (𝑒0 ) is nilpotent and 𝑅 is a semicommutative ring, then by
[13, lemma 3.1] we obtain that each monomial is a nilpotent element of Λ. Thus
𝑓0π‘˜ is also nilpotent.
If 𝑓 = 𝑓0′ , then 𝑓 ∈ Λ is a nilpotent element of Λ and 𝑓0′ (𝑒0 ) =
(𝑐𝑓 (𝑒0 ) 𝑒𝑒0 )(𝑒0 ) = 𝑓 (𝑒0 ) is a nilpotent element of 𝑅 and there is nothing to
264
R.M. Salem
prove.
Now, suppose that 0 βˆ•= 𝑓0 = 𝑓 − 𝑓0′ and let πœ‹(𝑓0 ) = πœ‹(𝑓 − 𝑓0′ ) = 𝑒𝑖 .
Since, 0 βˆ•= (𝑓 − 𝑓0′ )(𝑒𝑖 ) = (𝑓 − 𝑐𝑓 (𝑒0 ) 𝑒𝑒0 )(𝑒𝑖 ) = 𝑓 (𝑒𝑖 ) and 𝑓0 (𝑒0 ) = (𝑓 −
′
𝑓0 )(𝑒0 ) = (𝑓 − 𝑐𝑓 (𝑒0 ) 𝑒𝑒0 )(𝑒0 ) = 𝑓 (𝑒0 ) − 𝑓 (𝑒0 )𝑒𝑒0 (𝑒0 ) = 𝑓 (𝑒0 ) − 𝑓 (𝑒0 ) = 0 then
𝑒0 < 𝑒𝑖 .
By the fact that, 𝑓0π‘˜ is nilpotent, then there exists a positive integer 𝑛 such
that (𝑓0π‘˜ )𝑛 = 0.
′
Using the same procedure above it can be easily shown that 0 = 𝑓0π‘˜ (π‘˜ ′ 𝑒𝑖 ) =
′
(𝑓 (𝑒𝑖 ))π‘˜ where π‘˜ ′ = π‘˜π‘›. Continuing on this process 𝑓 = π‘“πœ‡ + π‘“πœ‡′ , where π‘“πœ‡′ :
𝑆 → 𝑅 is defined by π‘“πœ‡′ (π‘’π‘š ) = (𝑐𝑓 (π‘’π‘š ) π‘’π‘’π‘š )(π‘’π‘š ) = 𝑓 (π‘’π‘š ) which is nilpotent for
each π‘’π‘š ∈ 𝑠𝑒𝑝𝑝(𝑓 ), π‘š ≤ πœ‡ and π‘“πœ‡ is a nilpotent element of Λ. Let πœ‹(π‘“πœ‡ ) =
πœ‹(𝑓 − π‘“πœ‡′ ) = πœ‹(𝑓 − 𝑐𝑓 (π‘’πœ‡ ) π‘’π‘’πœ‡ ) = π‘’πœƒ , π‘’πœ‡ < π‘’πœƒ .
Using [12, 5.3] we can define a relation on Λ called section relation for π‘“πœ‡′ and
′
π‘“πœˆ ∈ Λ as follows:
i)
ii)
iii)
iv)
π‘“πœ‡′ βͺ― π‘“πœˆ′ if πœ‡ < 𝜈
πœ‹(𝑓 − π‘“πœ‡′ ) < πœ‹(𝑓 − π‘“πœˆ′ ) where πœ‡ < 𝜈
𝑒 < πœ‹(𝑓 − π‘“πœ‡′ ) for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(π‘“πœ‡′ )
π‘“πœ‡ = 𝑓 − π‘“πœ‡′ ∈ Λ is nilpotent and π‘“πœ‡′ (π‘’π‘š ) ∈ 𝑅 is nilpotent for each π‘’π‘š ∈
𝑠𝑒𝑝𝑝(𝑓 ), π‘š ≤ πœ‡.
Let ∗ denotes the section relation βͺ― with the above properties. Let 𝛼 be an
ordinal such that card 𝛼 > π‘π‘Žπ‘Ÿπ‘‘ ∣ 𝑠𝑒𝑝𝑝𝑓 ∣ and Γ be the set of all ordinals πœ† < 𝛼.
We show that for each πœ† ∈ Γ there exists π‘“πœ†′ ∈ Λ such that ∗ is satisfied. In fact
let πœ† ∈ Γ and assume that we have already found the element π‘“πœ‡′ ∈ Λ for every
πœ‡ < πœ† satisfying ∗ for ordinals πœ‡ < 𝜈 < πœ†.
Now, we will determine an element π‘“πœ†′ where ∗ is satisfied for πœ‡ < 𝜈 ≤ πœ†.
Suppose that there exists an ordinal πœ‚ such that πœ† = πœ‚ + 1. If 𝑓 − π‘“πœ‚′ = 0, then
𝑓 = π‘“πœ‚′ is nilpotent. Thus π‘“πœ‚′ (π‘’π‘š ) = (𝑐𝑓 (π‘’π‘š ) π‘’π‘’π‘š )(π‘’π‘š ) = 𝑓 (π‘’π‘š ) ∈ 𝑅 is nilpotent
for each π‘’π‘š ∈ 𝑠𝑒𝑝𝑝(𝑓 ), π‘š ≤ πœ‚ and there is nothing to prove.
Hence, suppose that π‘“πœ‚ = 𝑓 − π‘“πœ‚′ βˆ•= 0. Let πœ‹(π‘“πœ‚ ) = πœ‹(𝑓 − π‘“πœ‚′ ) = π‘’πœ† and
π‘“πœ†′ : 𝑆 → 𝑅 be defined by π‘“πœ†′ = π‘“πœ‚′ + 𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† . So π‘“πœ†′ ∈ Λ and we show that
π‘“πœ‚′ βͺ― π‘“πœ†′ which implies that π‘“πœ‡′ βͺ― π‘“πœ†′ for every πœ‡ < πœ†.
In fact 0 βˆ•= (𝑓 − π‘“πœ‚′ )(π‘’πœ† ) = 𝑓 (π‘’πœ† ) and it follows that 𝑠𝑒𝑝𝑝(π‘“πœ†′ − π‘“πœ‚′ ) =
𝑠𝑒𝑝𝑝(𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† ) = {π‘’πœ† }. If 𝑒 ∈ 𝑠𝑒𝑝𝑝(π‘“πœ‚′ ), then by ∗ 𝑒 < πœ‹(𝑓 − π‘“πœ‚′ ) = π‘’πœ† ∈
𝑠𝑒𝑝𝑝(π‘“πœ†′ − π‘“πœ‚′ ). Thus π‘“πœ‚′ βͺ― π‘“πœ†′ , if π‘“πœ‚′ = π‘“πœ†′ , and we have that 𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† = 0 which
is a contradiction. If π‘“πœ‡′ = π‘“πœ†′ , then π‘“πœ‡′ βͺ― π‘“πœ‚′ βͺ― π‘“πœ†′ = π‘“πœ‡′ and π‘“πœ†′ = π‘“πœ‚′ which is
again a contradiction. Hence, π‘“πœ‡′ βˆ•= π‘“πœ†′ for each πœ‡ < πœ‚ ≤ πœ†, and it can be easily
shown that π‘“πœ† = 𝑓 −π‘“πœ†′ is nilpotent and π‘“πœ†′ (π‘’π‘š ) = (𝑐𝑓 (π‘’π‘š ) π‘’π‘’π‘š )(π‘’π‘š ) = 𝑓 (π‘’π‘š ) ∈ 𝑅
is nilpotent for each π‘š ≤ πœ† .
If π‘“πœ† = 𝑓 − π‘“πœ†′ = 0, there is nothing to prove, otherwise there exists π‘’πœ‰ ∈
𝑠𝑒𝑝𝑝(𝑓 ) such that πœ‹(𝑓 − π‘“πœ†′ ) = π‘’πœ‰ where π‘“πœ†′ = π‘“πœ‚′ + 𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† . Since, (𝑓 − π‘“πœ†′ ) =
𝑓 − π‘“πœ‚′ − 𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† and (𝑓 − π‘“πœ†′ )(π‘’πœ† ) = (𝑓 − π‘“πœ‚′ − 𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† )(π‘’πœ† ) = 𝑓 (π‘’πœ† ) −
π‘“πœ‚′ (π‘’πœ† ) − 𝑓 (π‘’πœ† ) = 0 then π‘’πœ† < π‘’πœ‰ . By the fact that 0 βˆ•= (𝑓 − π‘“πœ†′ )(π‘’πœ‰ ) =
(𝑓 − π‘“πœ‚′ − 𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† )(π‘’πœ‰ ) = (𝑓 − π‘“πœ‚′ )(π‘’πœ‰ ) we have that π‘’πœ‰ ∈ 𝑠𝑒𝑝𝑝(𝑓 − π‘“πœ‚′ ). Hence,
Generalized Power Series
265
π‘’πœ† < π‘’πœ‰ and πœ‹(𝑓 − π‘“πœ‚′ ) < πœ‹(𝑓 − π‘“πœ† ) and this implies that πœ‹(𝑓 − π‘“πœ‡′ ) < πœ‹(𝑓 − π‘“πœ†′ )
for each πœ‡ < πœ†.
Now, we show that 𝑒 < πœ‹(𝑓 − π‘“πœ†′ ) for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(π‘“πœ†′ ). In fact 𝑠𝑒𝑝𝑝(π‘“πœ†′ ) =
𝑠𝑒𝑝𝑝(π‘“πœ‚′ + 𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† ) ⊆ 𝑠𝑒𝑝𝑝(π‘“πœ‚′ ) ∪ 𝑠𝑒𝑝𝑝(𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† ). If 𝑒 ∈ 𝑠𝑒𝑝𝑝(π‘“πœ‚′ ), then 𝑒 <
πœ‹(𝑓 − π‘“πœ‚′ ) and if 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑐𝑓 (π‘’πœ† ) π‘’π‘’πœ† ), then 𝑒 = π‘’πœ† = πœ‹(𝑓 − π‘“πœ‚′ ) < πœ‹(𝑓 − π‘“πœ†′ ).
Now, let πœ† be a limit ordinal for the family {π‘“πœ†′ βˆ£πœ‡ < πœ†} of elements π‘“πœ‡′ ∈ Λ and
it was proved in [12, 5.3] that there exists an element 𝑏 =βͺ― − sup(π‘“πœ‡′ )πœ‡<πœ† ∈ Λ
such that
i) π‘“πœ‡′ βͺ― 𝑏 for every πœ‡ < πœ†
ii) If 𝑏′ ∈ Λ and π‘“πœ‡′ βͺ― 𝑏′ for every πœ‡ < πœ†, then 𝑏 βͺ― 𝑏′ .
Let π‘“πœ†′ = 𝑏 =βͺ― − sup(π‘“πœ‡′ )πœ‡<πœ† . Then by i) we know that π‘“πœ‡′ βͺ― π‘“πœ†′ for every
πœ‡ < πœ†, π‘“πœ† = 𝑓 − π‘“πœ†′ is a nilpotent element of Λ and that π‘“πœ†′ (π‘’π‘š ) ∈ 𝑅 is nilpotent
′
′
for each π‘’π‘š ≤ π‘’πœ† . If π‘“πœ‡′ = π‘“πœ†′ , then π‘“πœ‡′ βͺ― π‘“πœ‡+1
βͺ― π‘“πœ†′ = π‘“πœ‡′ and π‘“πœ‡′ = π‘“πœ‡+1
which
′
′
is a contradiction. Hence, π‘“πœ‡ βˆ•= π‘“πœ† for every πœ‡ < πœ†.
Since, 𝑓 − π‘“πœ†′ = (𝑓 − π‘“πœ‡′ ) − (π‘“πœ†′ − π‘“πœ‡′ ) for every πœ‡ < πœ†, then by [12] π‘’πœ‰ =
πœ‹(𝑓 − π‘“πœ†′ ) ≥ min{πœ‹(𝑓 − π‘“πœ‡′ ), πœ‹(π‘“πœ†′ − π‘“πœ‡′ )}. Note that,
′
− π‘“πœ‡′ )
πœ‹(π‘“πœ†′ − π‘“πœ‡′ ) = πœ‹(π‘“πœ‡+1
′
= πœ‹(𝑓 − π‘“πœ‡′ ) − (𝑓 − π‘“πœ‡+1
)
≥ min{π‘’πœƒ , π‘’πœƒ+1 }
Hence, π‘’πœ‰ ≥ π‘’πœƒ for all πœ‡ < πœ† and π‘’πœƒ ≤ π‘’πœƒ+1 ≤ π‘’πœ‰ , then π‘’πœƒ < π‘’πœ‰ .
We now show that 𝑒 < πœ‹(𝑓 −π‘“πœ†′ ) for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(π‘“πœ†′ ). In fact, if 𝑠𝑒𝑝𝑝(π‘“πœ†′ ) =
∪πœ‡<πœ† 𝑠𝑒𝑝𝑝(π‘“πœ‡′ ), then there exists an ordinal πœ‡ < πœ† such that 𝑒 ∈ 𝑠𝑒𝑝𝑝(π‘“πœ‡′ ), thus
𝑒 < π‘’πœ‡ < π‘’πœ† .
Hence, for πœ‡, 𝜈 ∈ Γ, πœ‡ < 𝜈, then π‘’πœ‡ < π‘’πœˆ and we have that ∣{π‘’πœ† } such that
πœ† ∈ Γ∣ = ∣Γ∣ > βˆ£π‘†βˆ£ which is a contradiction. So, 𝑓 = π‘“πœ†′ and the Proposition is
proved.
Proposition 3.2. Let 𝑅 be a semicommutative right Noetherian ring, (𝑆, +, ≤)
a strictly ordered monoid and Λ = [[𝑅𝑆,≤ ]] the ring of generalized power series.
If 𝑓 ∈ Λ is such that 𝑓 (𝑒) is nilpotent for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ), then 𝑓 ∈ Λ is
nilpotent.
Proof. Let 𝑓 ∈ Λ be such that 𝑓 (𝑒) is a nilpotent element for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 )
and 𝐼 the ideal generated by {𝑓 (𝑒)βˆ£π‘’ ∈ 𝑠𝑒𝑝𝑝(𝑓 )}. Since, 𝑅 is a semicommutative right Noetherian ring, then by [13, lemma 3.1] 𝐼 is a finitely generated
nilpotent ideal.
Thus, there exists an π‘š ∈ β„• such that 𝐼 π‘š = (0). Since
∑
π‘š
𝑓 (𝑒) =
(𝑒1 ,⋅⋅⋅ ,π‘’π‘š )∈𝑋𝑒 (𝑓,⋅⋅⋅ ,𝑓 ) 𝑓 (𝑒1 ) ⋅ ⋅ ⋅ 𝑓 (π‘’π‘š ) for each 𝑒 ∈ 𝑆. Note that
for each summand belongs to ∈ 𝐼 π‘š , then each summand is equal zero. So,
𝑠𝑒𝑝𝑝(𝑓 π‘š ) = πœ™. Thus 𝑓 π‘š = 0 and 𝑓 is nilpotent.
The following example is due to D. Fields [5, Example 1] which shows us that
the right Noetherian condition can’t be deleted.
266
R.M. Salem
Example 3.3. Let 𝑆 = β„€/(𝑝) where 𝑝 is prime, {𝑋𝑖 }𝑖∈𝑁 a countable collection of
indeterminates over∑
𝑆, 𝑅 = 𝑆[𝑋0 , 𝑋1 , ⋅ ⋅ ⋅ , 𝑋𝑑 , ⋅ ⋅ ⋅ ]/(𝑋0𝑝 , 𝑋1𝑝 , ⋅ ⋅ ⋅ , 𝑋𝑑𝑝 , ⋅ ⋅ ⋅ ), 𝑓𝑖 =
∞
𝑋 𝑖 and let 𝑓 (𝑋) = 𝑖=0 𝑓𝑖 𝑋 𝑖 ∈ 𝑅[[𝑋]]. Then for each π‘˜ ∈ β„• (𝑓 (π‘˜) )𝑝 = 0. But
for each natural number 𝑛, 𝑓0 𝑓1 ⋅ ⋅ ⋅ 𝑓𝑛−1 ∈ (𝐴𝑓 )𝑛 and 𝑓0 𝑓1 ⋅ ⋅ ⋅ 𝑓𝑛−1 βˆ•= 0. Hence,
𝐴𝑓 isn’t nilpotent.
We combine Proposition 3.1 and Proposition 3.2 to get the following.
Theorem 3.4. Let 𝑅 be a semicommutative right Noetherian ring, (𝑆, +, ≤) a
strictly totally ordered monoid and Λ = [[𝑅𝑠,≤ ]] the ring of generalized power
series. Then 𝑓 ∈ Λ is a nilpotent element if and only if 𝑓 (𝑒) ∈ 𝑅 is a nilpotent
element for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ).
Proof. Suppose that 𝑓 is a nilpotent element of Λ. Then by Proposition 3.1 𝑓 (𝑒)
is nilpotent for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ).
Conversely if 𝑓 (𝑒) is a nilpotent element of 𝑅 for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ), then by
Proposition 3.2 𝑓 is nilpotent.
Theorem 3.5. Let 𝑅 be a semicommutative right Noetherian ring, (𝑆, +, ≤) a
strictly totally ordered monoid and Λ = [[𝑅𝑆,≤ ]] the generalized power series
ring. Then 𝑅 is a right weak zip ring if and only if Λ is a right weak zip ring.
Proof. Suppose that Λ is a right weak zip ring and 𝑋 ⊆ 𝑅 with 𝑁 π‘Ÿπ‘… (𝑋) ⊆
𝑛𝑖𝑙(𝑅). Let π‘Œ = {𝑐π‘₯ ∈ Λ∣π‘₯ ∈ 𝑋} and 0 βˆ•= 𝑓 ∈ 𝑁 π‘ŸΛ (π‘Œ ). Then 𝑐π‘₯ 𝑓 ∈ 𝑛𝑖𝑙(Λ)
for each 𝑐π‘₯ ∈ π‘Œ and by Proposition 3.1 π‘₯𝑓 (𝑒) ∈ 𝑛𝑖𝑙(𝑅) for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ).
Thus, 𝑓 (𝑒) ∈ 𝑛𝑖𝑙(𝑅) for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ) and by Proposition 3.2 we obtain that
𝑓 ∈ 𝑛𝑖𝑙(Λ). So, 𝑁 π‘ŸΛ (π‘Œ ) ⊆ 𝑛𝑖𝑙(Λ).
Since Λ is a right weak zip ring, then there exists a finite subset π‘Œ0 ⊆ π‘Œ
such that 𝑁 π‘ŸΛ (π‘Œ0 ) ⊆ 𝑛𝑖𝑙(Λ), where π‘Œ0 = 𝑐𝑋0 , and 𝑋0 = {π‘₯𝑖 βˆ£π‘– = 1, ⋅ ⋅ ⋅ , 𝑛}.
Let 𝑇 = ∪𝑓 ∈𝑁 π‘ŸΛ (π‘Œ0 ) {𝑓 (𝑒)βˆ£π‘’ ∈ 𝑠𝑒𝑝𝑝𝑓 }. Then by Proposition 3.1 𝑇 ⊆ 𝑛𝑖𝑙(𝑅).
Therefore, 𝑇 = 𝑁 π‘Ÿπ‘… (𝑋0 ) and 𝑅 is a right weak zip ring.
Conversely, assume that 𝑅 is a right weak zip ring and π‘Œ ⊆ Λ with 𝑁 π‘ŸΛ (π‘Œ ) ⊆
𝑛𝑖𝑙(Λ). Let 𝑇 = 𝐢(π‘Œ ) be the content of π‘Œ , i.e. 𝐢(π‘Œ ) = ∪𝑓 ∈π‘Œ {𝑓 (𝑒)βˆ£π‘’ ∈
𝑠𝑒𝑝𝑝(𝑓 )}. Then 𝑓 (𝑒)π‘Ž ∈ 𝑛𝑖𝑙(𝑅) for each 𝑓 (𝑒) ∈ 𝑇 and π‘Ž ∈ 𝑁 π‘Ÿπ‘… (𝑇 ). So, for
each 𝑓 ∈ π‘Œ and 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ), 𝑓 (𝑒)π‘Ž = (𝑓 π‘π‘Ž )(𝑒) ∈ 𝑛𝑖𝑙(𝑅) and by Proposition 3.2
𝑓 π‘π‘Ž ∈ 𝑛𝑖𝑙(Λ). Thus, π‘π‘Ž ∈ 𝑁 π‘ŸΛ (π‘Œ ) ⊆ 𝑛𝑖𝑙(Λ). Then by Proposition 3.1 π‘Ž ∈ 𝑛𝑖𝑙(𝑅)
and it follows that 𝑁 π‘Ÿπ‘… (𝑇 ) ⊆ 𝑛𝑖𝑙(𝑅).
Since, 𝑅 is a right weak zip ring, then there exists a finite subset 𝑇0 ⊆ 𝑇
such that 𝑁 π‘Ÿπ‘… (𝑇0 ) ⊆ 𝑛𝑖𝑙(𝑅). So, for each 𝑑 ∈ 𝑇0 , there exists 𝑓𝑑 ∈ π‘Œ such that
𝑑 ∈ {𝑓𝑑 (𝑒)βˆ£π‘“π‘‘ ∈ π‘Œ for some 𝑒 ∈ 𝑠𝑒𝑝𝑝𝑓𝑑 }.
Let π‘Œ0 be a minimal subset of π‘Œ such that 𝑓𝑑 ∈ π‘Œ0 for each 𝑑 ∈ 𝑇0 . Then π‘Œ0
is a finite subset of π‘Œ . Let 𝑇1 = ∪𝑓𝑑 ∈π‘Œ0 {𝑓𝑑 (𝑒)βˆ£π‘’ ∈ 𝑠𝑒𝑝𝑝(𝑓𝑑 )}. Thus, 𝑇0 ⊆ 𝑇1 and
we obtain that 𝑁 π‘Ÿπ‘… (𝑇1 ) ⊆ 𝑁 π‘Ÿπ‘… (𝑇0 ) ⊆ 𝑛𝑖𝑙(𝑅).
Let 𝑔 ∈ 𝑁 π‘ŸΛ π‘Œ0 . Then 𝑓 𝑔 ∈ 𝑛𝑖𝑙(Λ) for each 𝑓 ∈ π‘Œ0 and by Proposition 3.1
Generalized Power Series
267
(𝑓 𝑔)(𝑀) ∈ 𝑛𝑖𝑙(𝑅) for each 𝑀 ∈ 𝑠𝑒𝑝𝑝(𝑓 𝑔).
∑
Since, (𝑓 𝑔)(𝑀) =
(𝑒𝑖 ,𝑣𝑗 )∈𝑋𝑀 (𝑓,𝑔) 𝑓 (𝑒𝑖 )𝑔(𝑣𝑗 ) ∈ 𝑛𝑖𝑙(𝑅) and 𝑆 is a totally
ordered monoid, then there exists 𝑒0 , 𝑣0 ∈ 𝑆 such that πœ‹(𝑓 ) = 𝑒0 , and πœ‹(𝑔) = 𝑣0 .
Thus
∑ by the same procedure used in Proposition 3.1 we have that (𝑓 𝑔)(𝑒0 +𝑣0 ) =
(𝑒𝑖 ,𝑣𝑖 )∈𝑋𝑒 +𝑣 𝑓 (𝑒𝑖 )𝑔(𝑣𝑖 ) = 𝑓 (𝑒0 )𝑔(𝑣0 ) is nilpotent.
0
0
Now, suppose that 𝑓 (𝑒)𝑔(𝑣) is nilpotent, so 𝑔(𝑣)𝑓 (𝑒) is also nilpotent for
each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ) and 𝑣 ∈ 𝑠𝑒𝑝𝑝(𝑔) such that 𝑒 + 𝑣 < 𝑀. Now, we show that
𝑓 (𝑒)𝑔(𝑣) and 𝑔(𝑣)𝑓 (𝑒) are nilpotent for each 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ) and 𝑣 ∈ 𝑠𝑒𝑝𝑝(𝑔) with
𝑒 + 𝑣 = 𝑀.
Since, 𝑋𝑀 (𝑓, 𝑔) = {(𝑒, 𝑣)βˆ£π‘’ + 𝑣 = 𝑀, 𝑒 ∈ 𝑠𝑒𝑝𝑝(𝑓 ) π‘Žπ‘›π‘‘ 𝑣 ∈ 𝑠𝑒𝑝𝑝(𝑔)} is a finite
subset, then 𝑋𝑀 (𝑓, 𝑔) = {(𝑒𝑖 , 𝑣𝑖 )βˆ£π‘– = 1, ⋅ ⋅ ⋅ , 𝑛}.
Since 𝑆 is a totally ordered monoid, in fact a cancellative monoid, then let
𝑒1 < 𝑒2 < ⋅ ⋅ ⋅ < 𝑒𝑛 . If 𝑒1 = 𝑒2 and 𝑒1 + 𝑣1 = 𝑒2 + 𝑣2 , then 𝑣1 = 𝑣2 and as <
is strictly order, if 𝑒1 < 𝑒2 and 𝑒1 + 𝑣1 = 𝑒2 + 𝑣2 , it must be 𝑣1 > 𝑣2 . Thus,
𝑣𝑛 < 𝑛𝑛−1 < ⋅ ⋅ ⋅ < 𝑣1 , now
∑
(𝑓 𝑔)(𝑀) =
𝑓 (𝑒𝑖 )𝑔(𝑣𝑖 ) ∈ 𝑛𝑖𝑙(𝑅)
(𝑒1 ,𝑣𝑖 )∈𝑋𝑀 (𝑓,𝑔)
= 𝑓 (𝑒1 )𝑔(𝑣1 ) + 𝑓 (𝑒2 )𝑔(𝑣2 ) + ⋅ ⋅ ⋅ + 𝑓 (𝑒𝑛 )𝑔(𝑣𝑛 )
(**)
For 𝑖 ≥ 2 𝑒1 + 𝑣𝑖 < 𝑒𝑖 + 𝑣𝑖 = 𝑀, then by induction hypothesis, we have
𝑓 (𝑒1 )𝑔(𝑣𝑖 ) and 𝑔(𝑣𝑖 )𝑓 (𝑒1 ) are nilpotent elements. Now multiplying ** from the
right by 𝑓 (𝑒1 ), then
𝑓 (𝑒1 )𝑔(𝑣1 )𝑓 (𝑒1 ) = (𝑓 𝑔)(𝑀)𝑓 (𝑒1 ) − 𝑓 (𝑒2 )𝑔(𝑣2 )𝑓 (𝑒1 ) − ⋅ ⋅ ⋅ − 𝑓 (𝑒𝑛 )𝑔(𝑣𝑛 )𝑔(𝑒1 ).
Since, 𝑅 is semicommutative, then nil(𝑅) is a nilpotent ideal and by induction
𝑓 (𝑒1 )𝑔(𝑣1 )𝑓 (𝑒1 ), hence 𝑓 (𝑒1 )𝑔(𝑣1 ) and 𝑔(𝑣1 )𝑓 (𝑒1 ) are nilpotent.
Therefore, by multiplying (**) from the right by 𝑓 (𝑒2 ), ⋅ ⋅ ⋅ , 𝑓 (𝑒𝑛 ) respectively yields 𝑓 (𝑒𝑖 )𝑔(𝑣𝑖 ) and 𝑔(𝑣𝑖 )𝑓 (𝑒𝑖 ) are nilpotent for each 𝑒𝑖 ∈ 𝑠𝑒𝑝𝑝(𝑓 ) and
𝑣𝑖 ∈ 𝑠𝑒𝑝𝑝(𝑔). Consequently, 𝑔(𝑣) ∈ 𝑁 π‘Ÿπ‘… (𝑇1 ) ⊆ 𝑛𝑖𝑙(𝑅) for each 𝑣𝑖 ∈ 𝑠𝑒𝑝𝑝(𝑔).
Thus by Proposition 3.2 𝑔 ∈ 𝑛𝑖𝑙(Λ). So, π‘π‘ŸΛ (π‘Œ0 ) ⊆ 𝑛𝑖𝑙(Λ) and Λ is a right weak
zip ring.
Acknowledgement. I’d like to express my deepest gratitude to Prof. M.H. Fahmy
for reading this manuscript and his valuable discussion, also I’m so grateful to
the referee for his/her valuable suggestions which helped me to improve the
presentation considerably.
References
[1] J.A. Beachy, W.D. Blair, Rings whose faithful left ideals are cofaithful, Pacific J.
Math. 58 (1) (1975) 1–13.
[2] F. Ced`
π‘œ, Zip rings and Mal’cev domains, Comm. in Algebra 19 (7) (1991) 1983–
1991.
[3] W. Cortes, Skew polynomial extensions over zip rings, Int. J. Math. and Math.
Sci. 10 (2008) 1–8.
268
R.M. Salem
[4] C. Faith, Annihilator ideals, associated primes and Kasch-McCoy commutative
rings, Comm. in Algebra 19 (7) (1991) 1867–1892.
[5] D.E. Fields, Zero divisors and nilpotent elements in power series rings, Proceeding
of the American Mathematical Society 27 (3) (1971).
[6] A.M. Hassanien, R.M. Salem, M. Farahat, Prüfer domains of generalized power
series, J. of the Egyptian Mathematical Society 15 (1) (2007) 11–19.
[7] Y. Hirano, On annihilator ideals of a polynomial ring over a noncommutative
ring, J. Pure Appl. Algebra 168 (1) (2002) 45–52.
[8] C.Y. Hong, N.K. Kim, T.K. Kwak, Y. Lee, Extensions of zip rings, J. Pure Appl.
Algebra 195 (3) (2005) 231–242.
[9] L. Ouyang, Ore extensions of weak zip rings, Glasgow MaTheorem J. 51 (2009)
525–537.
[10] M.B. Rege, S. Chhawchharia, Armendariz rings, Proc. Japan Acad. Ser. A Math.
Sci. 73 (1997) 14–17.
[11] P. Ribenboim, Rings of generalized power series: Nilpotent elements, Abh. Math.
Sem. Univ. Hamburg 61 (1991) 15–33.
[12] P. Ribenboim, Noetherian rings of generalized power series, J. Pure Appl. Algebra
79 (1992) 293–312.
[13] R.M. Salem, Prüer rings of generalized power series, Southeast Asian Bull. Math.
33 (2009) 527–534.
[14] L. Zhongkui, Special properties of rings of generalized power series, Comm. in
Algebra 32 (8) (2004) 3215–3226.
[15] L. Zhongkui, Z. Zhao, On weak Armendariz rings, Comm. in Algebra 34 (2006)
2607–2616.
[16] J.M. Zelmanowitz, The finite intersection property on annihilator right ideals,
Proc. Am. Math. Soc. 57 (2) (1976) 213–216.
Download