Fixed point theory is one of the most interesting areas of research in mathematics. In this direction, we study some unique common fixed point results for a pair of self-mappings without continuity on fuzzy metric spaces under the generalized contraction conditions by using "the triangular property of fuzzy metric". Moreover, we present weak-contraction and generalized Ćirić-contraction theorems. The results are supported by suitable examples. Further, we establish a supportable application of the fuzzy differential equations to ensure the existence of a unique common solution to validate our main work.
Citation: Iqra Shamas, Saif Ur Rehman, Thabet Abdeljawad, Mariyam Sattar, Sami Ullah Khan, Nabil Mlaiki. Generalized contraction theorems approach to fuzzy differential equations in fuzzy metric spaces[J]. AIMS Mathematics, 2022, 7(6): 11243-11275. doi: 10.3934/math.2022628
[1] | Liqian Bai, Xueqing Chen, Ming Ding, Fan Xu . A generalized quantum cluster algebra of Kronecker type. Electronic Research Archive, 2024, 32(1): 670-685. doi: 10.3934/era.2024032 |
[2] | Yongjie Wang, Nan Gao . Some properties for almost cellular algebras. Electronic Research Archive, 2021, 29(1): 1681-1689. doi: 10.3934/era.2020086 |
[3] | Quanguo Chen, Yong Deng . Hopf algebra structures on generalized quaternion algebras. Electronic Research Archive, 2024, 32(5): 3334-3362. doi: 10.3934/era.2024154 |
[4] | Peigen Cao, Fang Li, Siyang Liu, Jie Pan . A conjecture on cluster automorphisms of cluster algebras. Electronic Research Archive, 2019, 27(0): 1-6. doi: 10.3934/era.2019006 |
[5] | Xiu-Jian Wang, Jia-Bao Liu . Quasi-tilted property of generalized lower triangular matrix algebras. Electronic Research Archive, 2025, 33(5): 3065-3073. doi: 10.3934/era.2025134 |
[6] | Agustín Moreno Cañadas, Robinson-Julian Serna, Isaías David Marín Gaviria . Zavadskij modules over cluster-tilted algebras of type A. Electronic Research Archive, 2022, 30(9): 3435-3451. doi: 10.3934/era.2022175 |
[7] | Fang Li, Jie Pan . On inner Poisson structures of a quantum cluster algebra without coefficients. Electronic Research Archive, 2021, 29(5): 2959-2972. doi: 10.3934/era.2021021 |
[8] | Ming Ding, Zhiqi Chen, Jifu Li . The properties on F-manifold color algebras and pre-F-manifold color algebras. Electronic Research Archive, 2025, 33(1): 87-101. doi: 10.3934/era.2025005 |
[9] | Kailash C. Misra, Sutida Patlertsin, Suchada Pongprasert, Thitarie Rungratgasame . On derivations of Leibniz algebras. Electronic Research Archive, 2024, 32(7): 4715-4722. doi: 10.3934/era.2024214 |
[10] | Agustín Moreno Cañadas, Pedro Fernando Fernández Espinosa, José Gregorio Rodríguez-Nieto, Odette M Mendez, Ricardo Hugo Arteaga-Bastidas . Extended Brauer analysis of some Dynkin and Euclidean diagrams. Electronic Research Archive, 2024, 32(10): 5752-5782. doi: 10.3934/era.2024266 |
Fixed point theory is one of the most interesting areas of research in mathematics. In this direction, we study some unique common fixed point results for a pair of self-mappings without continuity on fuzzy metric spaces under the generalized contraction conditions by using "the triangular property of fuzzy metric". Moreover, we present weak-contraction and generalized Ćirić-contraction theorems. The results are supported by suitable examples. Further, we establish a supportable application of the fuzzy differential equations to ensure the existence of a unique common solution to validate our main work.
Cluster algebras were introduced by Fomin and Zelevinsky in [1]. The core idea to define cluster algebra of rank n is that one should have a cluster seed and an operator on cluster seeds, called mutation. Roughly, a cluster seed Σt0 is a collection of variables x1;t0,⋯,xn;t0 (cluster variables) and binomials F1;t0,⋯,Fn;t0 (exchange polynomials). One can apply mutation to a cluster seed to produce a new seed, i.e., new variables and new binomials. Note that the exchange polynomial in cluster algebra is always a binomial. One of the main results in cluster algebras is that they have the Laurent phenomenon [1].
In the theory of cluster algebras, the following are interesting conjectures on seeds of cluster algebras: in a cluster algebra of rank n, (1) each seed is uniquely defined by its cluster; (2) any two seeds with n−1 common cluster variables are connected with each other by one step of mutation. One can refer to [2,3] for detailed proof.
Significant notations of the upper cluster algebra, upper bound and lower bound associated with the cluster seed were introduced by Berenstein, Fomin and Zelevinsky to study the structure of cluster algebras in [4]. There are some theorems of upper bounds and lower bounds: (a) under the condition of coprimeness, the upper bound is invariant under seed mutations; (b) the standard monomials in x1,x′1,…,xn,x′n are linearly independent over ZP if and only if the cluster seed is acyclic; (c) under the conditions of acyclicity and coprimeness, then the upper bound coincides with the lower bound.
Muller showed that locally acyclic cluster algebras coincide with their upper cluster algebras in [5]. Gekhtman, Shapiro and Vainshte in [3] proved (a) for generalized cluster algebras, then Bai, Chen, Ding and Xu demonstrated (c) and the sufficiency of (b) in [6]. Besides, Bai discovered that acyclic generalized cluster algebras coincide with their generalized upper cluster algebras.
Laurent phenomenon (LP) algebras were introduced by Lam and Pylyavskyy in [7], which generalize cluster algebras from the perspective of exchange relations. The exchange polynomials in LP algebras were allowed to have arbitrarily many monomials, rather than being just binomials. It turns out that the Laurent phenomenon also appears in LP algebras [7].
One should note that our method also works for cluster algebras and generalized cluster algebras. We do not talk much about generalized cluster algebras in this paper, and one can refer to [2,6,8,9,10] for details.
In this paper, we first affirm the conjectures on seeds of cluster algebras with respect to LP algebras.
Theorem 1.1. In a LP algebra of rank n,
1) (Theorem 3.1) each LP seed is uniquely defined by its cluster.
2) (Theorem 3.7) any two LP seeds with n−1 common cluster variables are connected with each other by one step of mutation.
Second, we affirm theorems of upper bounds and lower bounds with respect to LP algebras under some conditions, by using the similar methods developed in [4].
Condition 1.2. Let Mk be the lexicographically first monomial in the irreducible polynomial Fk and fk(xi) be the polynomial on xi in R[x2,…,ˆxi,…,ˆxk,…,xn](xi) without constant terms in Fk for any i≠k. Assume that for a LP seed (x,F) of rank n, ∀k∈[1,n], Fk satisfies the following conditions:
(i) ˆFk=Fk.
(ii) Mk is of the form xvk={xvk+1,kk+1⋯xvn,knk∈[1,n−1]1k=n, where vk∈Zn−k≥0 for k∈[1,n−1].
(iii) when x1∈Fk for k≠1, Fk=Mk+fk(x1).
(iv) when x1∉Fk for k≠1 or 2, if there exist an index i in [2,k−1] such that xk∈Mi, then Fk=Mk+fk(xi).
Theorem 1.3. (a) (Theorem 4.9) Under (i) of Condition 1.2, the upper bound is invariant under LP mutations.
(b) (Theorem 4.16) Under (i) and (ii) of Condition 1.2, the standard monomials in x1,x′1,…,xn,x′n form an R-basis for L(Σ).
(c) (Theorem 4.9) Under Condition 1.2, the upper bound coincides with the lower bound.
This paper is organized as follows: In Section 2, some basic definitions are given. In Section 3, we prove Theorem 1.1, and we give the corresponding results and applications in cluster algebras. In Section 4, we affirm Theorem 1.3.
Let a,b be positive integers satisfying a≤b, write [a,b] for {a,a+1,…,b}.
Let R be a unique factorization domain over Z, and the ambient field F be the rational function field in n independent variables over the field of fractions Frac(R). Recall that an element f of R is irreducible if it is non-zero, not a unit, and not be expressed as the product f=gh of two elements g,h∈R which are non-units.
Definition 2.1. A Laurent phenomenon (LP) seed of rank n in F is a pair (x,F), in which
(i) x={x1⋯,xn} is a transcendence basis for F over Frac(R), where x is called the cluster of (x,F) and x1⋯,xn are called cluster variables.
(ii) F={F1,⋯,Fn} is a collection of irreducible polynomials in R[x1,⋯,xn] such that for each i,j∈[1,n], xj∤Fi (Fi is not divisible by xj) and Fi does not depend on xi, where F1,⋯,Fn are called the exchange polynomials of (x,F).
The following notations, definitions and propositions can refer to [7,11].
Let F,N be two rational functions in x1,⋯,xn. Denote by F|xj←N the expression obtained by substituting xj in F by N. And if F involves the variable xi, then we write xi∈F. Otherwise, we write xi∉F.
Definition 2.2. Let (x,F) be a LP seed in F. For each Fj∈F, define a Laurent polynomial ˆFj=Fjxa11⋯xaj−1j−1xaj+1j+1⋯xann, where ak∈Z≥0 is maximal such that Fakk divides Fj|xk←Fk/x′k as an element in R[x1,⋯,xk−1,(x′k)−1,xk+1,⋯,xn]. The Laurent polynomials in ˆF:={ˆF1,⋯,ˆFn} are called the exchange Laurent polynomials.
From the definition of exchange Laurent polynomials, we know that Fj/ˆFj is a monomial in R[x1,⋯,ˆxj,⋯,xn], where ˆxj means xj vanishes in the {x1,⋯,xn}. And ˆFj|xk←Fk/x′k is not divisible by Fk.
Proposition 2.3. (Lemma 2.4 of [7])Let (x,F) be a LP seed in F, then F={F1,⋯,Fn} and ˆF={ˆF1,⋯,ˆFn} determine each other uniquely.
Proposition 2.4. (Lemma 2.7 of [7])If xk∈Fi, then xi∉Fk/ˆFk. In particular, xk∈Fi implies that ˆFk|xi←0 is well defined and ˆFk|xi←0∈R[x±11,⋯,ˆxi,⋯,ˆxk,⋯,x±1n].
Definition 2.5. Let (x,F) be a LP seed in F and k∈[1,n]. Define a new pair
({x′1,⋯,x′n},{F′1,⋯,F′n}):=μk(x,F), |
where x′k=ˆFk/xk and x′i=xi for i≠k. And the exchange polynomials change as follows:
(1) F′k:=Fk;
(2) If xk∉Fi, then F′i:=riFi, where ri is a unit in R;
(3) If xk∈Fi, then F′i is obtained from the following three steps:
(i) Define Gi:=Fi|xk←Nk, where Nk=ˆFk|xi←0x′k. Then we have
Gi∈R[x±11,⋯,ˆxi,⋯,x′k−1,⋯,x±1n]=R[x′1±1,⋯,^x′i,⋯,x′k−1,⋯,x′n±1]. |
(ii) Define Hi to be Gi with all common factors (in R[x1,⋯,ˆxi,⋯,ˆxk,⋯,xn]) removed. Note that Hi is unique up to a unit in R and Hi∈R[x′1±1,⋯,^x′i,⋯,x′k−1,⋯,x′n±1].
(iii) Let M be a Laurent monomial in x′1,⋯,^x′i,⋯,x′n with coefficient a unit in R such that F′i:=MHi∈R[x′1,⋯,x′n] and is not divisible by any variable in {x′1,⋯,x′n}. Thus
F′i∈R[x′1,⋯,^x′i,⋯,x′k,⋯,x′n]. |
Then we say that the new pair μk(x,F) is obtained from the LP seed (x,F) by the LP mutation in direction k.
Example 2.6. Let R=Z and F=Q(a,b,c). Consider the LP seed (x,F), where x={a,b,c} and F={b+1, a+c, b+1}. From the definition of exchange Laurent polynomials, we can get ˆFa=Fac,ˆFb=Fb,ˆFc=Fca.
Let (x′,F′)=μa(x,F), then we have a′=^Faa=b+1ac, b′=b, c′=c. From the definition of the LP mutation, the exchange polynomial Fa does not change. Since a∉Fc, we have F′c=b+1 (or up to a unit). Since Fb depends on a, to compute F′b, we need to procedure the above three steps. By (i), we get Na=1a′c and Gb=1a′c+c. By (ii), we get Hb=Gb up to a unit in R. By (iii), M=a′c and F′b=MHb=a′c2+1. Thus the new seed can be chosen to be
(x′,F′)={(a′,b+1),(b,a′c2+1),(c,b+1)}. |
It is not clear a priori that the LP mutation μk(x,F) of a LP seed (x,F) is still a LP seed because of the irreducibility requirement for the new exchange polynomials. But it can be seen from the following proposition that μk(x,F) is still a LP seed in F.
Proposition 2.7. (Proposition 2.15 of [7])Let (x,F) be a LP seed in F, then μk(x,F) is also a LP seed in F.
Proposition 2.8. (Proposition 2.16 of [7])If (x′,F′) is obtained from (x,F) by LP mutation at k, then (x,F) can be obtained from (x′,F′) by LP mutation at k. In this sense, LP mutation is an involution.
Remark 2.9. It is important to note that because of (ii), F′i is defined up to a unit in R. And this is the motivation to consider LP seeds up to an equivalence relation.
Definition 2.10. Let Σt1=(xt1,Ft1) and Σt2=(xt2,Ft2) be two LP seeds in F. Σt1 and Σt2 are equivalent if for each i∈[1,n], there exist ri,r′i which are units in R such that xi;t2=rixi;t1 and Fi;t2=r′iFi;t1.
Denote by [Σt] the equivalent class of Σt, that is, [Σt] is the set of LP seeds which are equivalent to Σt.
Proposition 2.11. (Lemma 3.1 of [7])Let Σt1=(xt1,Ft1) and Σt2=(xt2,Ft2) be two LP seeds in F, and Σtu=μk(Σt2), Σtv=μk(Σt1).If [Σt1]=[Σt2], then [Σtv]=[Σtu].
Let Σt=(xt,Ft) be a LP seed in F. By the above proposition, it is reasonable to define LP mutation of [Σt] at k given by μk([Σt]):=[μk(Σt)].
Definition 2.12. A Laurent phenomenon (LP) pattern S in F is an assignment of each LP seed (xt,Ft) to a vertex t of the n-regular tree Tn, such that for any edge tk_t′,(xt′,Ft′)=μk(xt,Ft).
We always denote by xt={x1;t,⋯,xn;t} and Ft={F1;t,⋯,Fn;t}.
Definition 2.13. Let S be a LP pattern, the Laurent phenomenon (LP) algebra A(S) (of rank n) associated with S is the R-subalgebra of F generated by all the cluster variables in the seeds of S.
If Σ=(x,F) is any seed in F, we shall write A(Σ) to mean the LP algebra A(S) associated with S containing the seed Σ.
Theorem 2.14. (Theorem 5.1 of [7], the Laurent phenomenon) Let A(S) be a LP algebra, and (xt0,Ft0) be a LP seed of A(S). Then any cluster variable xi;t of A(S) is in the Laurent polynomial ring R(t±10):=R[x±11;t0,⋯,x±1n;t0].
Definition 2.15. Let Σ=(x,F) be a LP seed of rank n and k∈[1,n]. A new seed Σ∗=(x∗,F∗) of rank n−1 is defined as follows:
1) let R∗=R[x±1k].
2)x∗=x−{xk}.
3)let F∗={F∗j=Fjxak | j∈[1,n]−k, a is the power of xk in ˆFj/Fj}.
The seed Σ∗ is in fact a LP seed, then Σ∗ is called the freezing of the LP seed Σ at xk. A(Σ∗)⊂F=Frac(R∗[x1,…,^xk,…,xn]) is defined to be the subalgebra generated by all the cluster variables from LP seeds mutation-equivalent to Σ∗. Then A(Σ∗) is called the freezing of the LP algebra A(Σ) at xk.
Example 2.16. Consider the LP seed Σ=(x,F)={(a,b+1),(b,a+c),(c,b+1)} over R=Z from Example 2.6. We produce the freezing of (x,F) at c as follows: first, remove (c,b+1); next, since the powers of c in ^Fa and ^Fb are -1 and 0 respectively, we have F∗a=Fac−1=b+1c and F∗b=Fb. Then the LP seed Σ∗ are {(a,b+1c),(b,a+c)} over Z[c±1].
Proposition 2.17. (Proposition 3.7 of [7])The algebra A(Σ∗) is a LP algebra.
Corollary 2.18. The freezing of the LP seed at xi is compatible with the mutation in direction j for j≠i.
Recall that an integer matrix Bn×n=(bij) is called skew-symmetrizable if there is a positive integer diagonal matrix D such that DB is skew-symmetric, where D is said to be the skew-symmetrizer of B. Bn×n=(bij) is sign-skew-symmetric if bijbji<0 or bij=bji=0 for any i,j∈[1,n]. A sign-skew-symmetric B is totally sign-skew-symmetric if any matrix B′ obtained from B by a sequence of mutations is sign-skew-symmetric. It is known that skew-symmetrizable integer matrices are always totally sign-skew-symmetric.
The diagram for a sign-skew-symmetric matrix Bn×n is the directed graph Γ(B) with the vertices 1,2,⋯,n and the directed edges from i to j if bij>0. Bn×n is called acyclic if Γ(B) has no oriented cycles. As shown in [12], an acyclic sign-skew-symmetric integer matrix B is always totally sign-skew-symmetric.
Let P be the coefficient group, its group ring ZP is a domain [1]. We take an ambient field F to be the field of rational functions in n independent variables with coefficients in ZP.
Definition 2.19. A cluster seed in F is a triplet Σ=(x,y,B) such that
(i) x={x1,⋯,xn} is a transcendence basis for F over Frac(ZP). x is called the cluster of (x,y,B) and {x1⋯,xn} are called cluster variables.
(ii) y={y1,⋯,yn} is a subset of P, where {y1,⋯,yn} are called coefficients.
(iii) B=(bij) is a n×n totally sign-skew-symmetric matrix, called an exchange matrix.
Let (x,y,B) be a cluster seed in F, one can associate binomials {F1,⋯,Fn} defined by
Fj=yj1⊕yj∏bij>0xbiji+yj1⊕yj∏bij<0x−biji. |
{F1,⋯,Fn} are called the exchange polynomials of (x,y,B).
Note that the coefficients and the exchange matrices in a cluster algebra are used for providing the exchange polynomials and explaining how to produce new exchange polynomials when doing a mutation (see Definition 2.20) on a cluster seed.
Definition 2.20. Let Σ=(x,y,B) be a cluster seed in F. Define the mutation of Σ in the direction k∈[1,n] as a new triple Σ′=(x′,y′,B′):=μk(Σ) in F, where
x′i={Fk/xki=kxii≠k.,y′i={y−1ki=kyiymax(bki,0)k(1⨁yk)−bkii≠k., |
and b′ij={−biji=k or j=kbij+sgn(bik)max(bikbkj,0)otherwise. |
It can be seen that μk(Σ) is also a cluster seed and the mutation of a cluster seed is an involution, that is, μk(μk(Σ))=Σ.
Definition 2.21. A cluster pattern S is an assignment of a seed Σt=(xt,yt,Bt) to every vertex t of the n-regular tree Tn, such that for any edge tk_t′,Σ′t′=(xt′,yt′,Bt′)=μk(Σt).
We always denote by xt=(x1;t,⋯,xn;t),yt=(y1;t,⋯,yn;t),Bt=(btij).
Definition 2.22. Let S be a cluster pattern, the cluster algebra A(S) (of rank n) associated with the given cluster pattern S is the ZP-subalgebra of the field F generated by all cluster variables of S.
Theorem 2.23. (Theorem 3.1 of [1], the Laurent phenomenon)Let A(S) be a cluster algebra, and Σt0=(xt0,yt0,Bt0) be a cluster seed of A(S). Then any cluster variable xi;t of A(S) is in the Laurent polynomial ring ZP(t±10):=ZP[x±11;t0,⋯,x±1n;t0].
Example 2.24. Let B=(03−30), then the exchange polynomials of the cluster seed (x,B) are the following two polynomials
F1=x32+1=(x2+1)(x22−x2+1), |
F2=x31+1=(x1+1)(x21−x1+1). |
It is easy to see that the exchange polynomials F1,F2 of (x,B) are both reducible in the above example. Thus the cluster x and the exchange polynomial F of (x,B) can not define a LP seed. From [7], we know that sometimes a cluster algebra defines a LP algebra indeed.
Theorem 2.25. (Theorem 4.5 of [7]) Every cluster algebra with principal coefficients is a Laurent phenomenon algebra.
Theorem 3.1. Let A(S) be a LP algebra of rank n, and (xt1,Ft1),(xt2,Ft2) be two LP seeds of A(S).
1) If there exists a permutation σ of [1,n] and a unit ri∈R such that xi;t2=rixσ(i);t1 for i∈[1,n], then Fi;t2=r′iFσ(i);t1 as polynomials for a certain unit r′i in R.
2) each LP seed is uniquely defined by its cluster.
Proof. For any fixed k∈[1,n], let (xu,Fu)=μk(xt2,Ft2) and (xv,Fv)=μσ(k)(xt1,Ft1), we consider the Laurent expansion of xk;u with respect to xv and the Laurent expansion of xσ(k);v with respect to xu.
From the definition of the LP mutation, we know
xi;u={xi;t2if i≠kˆFk;t2xk;t2if i=k and xi;v={xi;t1if i≠σ(k)ˆFσ(k);t1xσ(k);t1if i=σ(k). | (3.1) |
Since xi;t2=rixσ(i);t1 for i∈[1,n], we have xi;u=rixσ(i);v for i≠k. By (3.1), we get
xk;u=ˆFk;t2(x1;t2,⋯,ˆxk;t2,⋯,xn;t2)/xk;t2=ˆFk;t2(r1xσ(1);t1,⋯,ˆxσ(k);t1,⋯,rnxσ(n);t1)/(rkxσ(k);t1)=ˆFk;t2(r1xσ(1);v,⋯,ˆxσ(k);v,⋯,rnxσ(n);v)/(rkxσ(k);t1);xσ(k);v=ˆFσ(k);t1(x1;t1,⋯,ˆxσ(k);t1,⋯,xn;t1)/xσ(k);t1=ˆFσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v)/xσ(k);t1. |
Thus xk;uxσ(k);v=ˆFk;t2(r1xσ(1);v,⋯,ˆxσ(k);v,⋯,rnxσ(n);v)rkˆFσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v) and we get that
xk;u=xσ(k);vˆFk;t2(r1xσ(1);v,⋯,ˆxσ(k);v,⋯,rnxσ(n);v)rkˆFσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v). | (3.2) |
From the definition of the exchange Laurent polynomial, we know the above equation has the form of
xk;u=xσ(k);vFk;t2(r1xσ(1);v,⋯,ˆxσ(k);v,⋯,rnxσ(n);v)rkFσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v)M, | (3.3) |
where the Laurent monomial M is of the form xm11;v⋯xmσ(k)−1σ(k)−1;vxmσ(k)+1σ(k)+1;v⋯xmnn;v and mj is integer for j∈[1,n]−σ(k). Thus Eq (3.3) is the Laurent expansion of xk;u with respect to xv.
Similarly, the following equation is the Laurent expansion of xσ(k);v with respect to xu.
xσ(k);v=xk;uFσ(k);t1(r−11xσ−1(1);u,⋯,ˆxk;u,⋯,r−1nxσ−1(n);u)r−1kFk;t2(x1;u,⋯,ˆxk;u,⋯,xn;u)M−1, | (3.4) |
where M−1 is also a Laurent monomial in R[x±11;u,⋯,ˆxk;u,⋯,x±1n;u] since xi;u=rixσ(i);v for i≠k.
We know that both Fσ(k);t1 ( r−11xσ−1(1);u ,⋯, ˆxk;u ,⋯, r−1nxσ−1(n);u ) Fk;t2 ( x1;u ,⋯, ˆxk;u ,⋯, xn;u ) =Fσ(k);t1 ( x1;v ,⋯, ˆxσ(k);v ,⋯, xn;v ) Fk;t2 ( r1xσ(1);v ,⋯, ˆxσ(k);v ,⋯, rnxσ(n);v ) and Fk;t2 ( r1xσ(1);v ,⋯, ˆxσ(k);v ,⋯, rnxσ(n);v ) Fσ(k);t1 ( x1;v ,⋯, ˆxσ(k);v ,⋯, xn;v ) are Laurent polynomials in
R[x±11;v,⋯,ˆxσ(k);v,⋯,x±1n;v]=R[x±11;u,⋯,ˆxk;u,⋯,x±1n;u] |
by the Laurent phenomenon.
Thus both Fk;t2 ( r1xσ(1);v ,⋯, ˆxσ(k);v ,⋯, rnxσ(n);v ) Fσ(k);t1 ( x1;v ,⋯, ˆxσ(k);v ,⋯, xn;v ) and Fσ(k);t1 ( x1;v ,⋯, ˆxσ(k);v ,⋯, xn;v ) Fk;t2 ( r1xσ(1);v ,⋯, ˆxσ(k);v ,⋯, rnxσ(n);v ) are units in R[x±11;v ,⋯, ˆxσ(k);v ,⋯, x±1n;v].
Because both Fk;t2 and Fσ(k);t1 are irreducible and xj;t2∤Fk;t2, xj;t1∤Fσ(k);t1 for each j∈[1,n], so that both Fk;t2 ( r1xσ(1);v ,⋯, ˆxσ(k);v ,⋯, rnxσ(n);v ) Fσ(k);t1 ( x1;v ,⋯, ˆxσ(k);v ,⋯, xn;v ) and Fσ(k);t1 ( x1;v ,⋯, ˆxσ(k);v ,⋯, xn;v ) Fk;t2 ( r1xσ(1);v ,⋯, ˆxσ(k);v ,⋯, rnxσ(n);v ) are units in R.
Hence
Fk;t2(r1xσ(1);v,⋯,ˆxσ(k);v,⋯,rnxσ(n);v)=r′kFσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v), |
for some unit r′k in R, i.e., Fk;t2(x1;u,⋯,ˆxk;u,⋯,xn;u)=r′kFσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v). Thus Fk;t2=r′kFσ(k);t1 as polynomials, for each k∈[1,n].
Remark 3.2. From the proof of the above theorem, we can see that the choice of the unit ri such that xi;t2=rixσ(i);t1 does not matter when proving Fi;t2/Fσ(i);t1 is a unit in R. Similarly, in the proof which is based on the Laurent phenomenon and need to use the ratio of two exchange polynomials (for example, the proof of Lemma 3.6), we can assume that ri=1.
Next, we will give the proof of the conjecture for cluster algebras that each seed is uniquely defined by its cluster, and main points of proof that are different from the previous one.
Theorem 3.3. Let A(S) be a cluster algebra, and Σtl=(xtl,ytl,Btl),l=1,2 be two cluster seeds of A(S).If there exists a permutation σ of [1,n] such that xi;t2=xσ(i);t1 for i∈[1,n], then
(i) Either yk;t2=yσ(k);t1, bt2ik=bt1σ(i)σ(k) or yk;t2=y−1σ(k);t1, bt2ik=−bt1σ(i)σ(k) for i,k∈[1,n].
(ii) In both cases, Fi;t2=Fσ(i);t1 as polynomials for i∈[1,n].
Proof. By the same method with the proof of Theorem 3.1, the version of the equation (3.2) for the cluster algebra is just
xk;u=xσ(k);vFk;t2(xσ(1);v,⋯,ˆxσ(k);v,⋯,xσ(n);v)Fσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v), | (3.5) |
and note that xi;u=xσ(i);v for any i≠k, we also have
xσ(k);v=xk;uFσ(k);t1(xσ−1(1);u,⋯,ˆxk;u,⋯,xσ−1(n);u)Fk;t2(x1;u,⋯,ˆxk;u,⋯,xn;u). | (3.6) |
We know that Eq (3.5) is the Laurent expansion of xk;u with respect to xv and Eq (3.6) is the Laurent expansion of xσ(k);v with respect to xu. Then by the Laurent phenomenon, both Fk;t2 ( xσ(1);v ,⋯, ˆxσ(k);v ,⋯, xσ(n);v ) Fσ(k);t1 ( x1;v ,⋯, ˆxσ(k);v ,⋯, xn;v ) and Fσ(k);t1 ( xσ−1(1);u ,⋯, ˆxk;u ,⋯, xσ−1(n);u ) Fk;t2 ( x1;u ,⋯, ˆxk;u ,⋯, xn;u ) are Laurent polynomials, and this implies that Fk;t2 ( xσ(1);v ,⋯, ˆxσ(k);v ,⋯, xσ(n);v ) Fσ(k);t1 ( x1;v ,⋯, ˆxσ(k);v ,⋯, xn;v ) is a Laurent monomial in ZP[x±11;v ,⋯, ˆxσ(k);v ,⋯, x±1n;v]. We know that
Fk;t2(xσ(1);v,⋯,ˆxσ(k);v,⋯,xσ(n);v)=yk;t21⊕yk;t2∏bt2ik>0xbt2ikσ(i);v+11⊕yk;t2∏bt2ik<0x−bt2ikσ(i);v, |
Fσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v)=yσ(k);t11⊕yσ(k);t1∏bt1iσ(k)>0xbt1iσ(k)i;v+11⊕yσ(k);t1∏bt1iσ(k)<0x−bt1iσ(k)i;v. |
Because Fk;t2(xσ(1);v,⋯,ˆxσ(k);v,⋯,xσ(n);v)Fσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v) is a Laurent monomial, we must have either yk;t2=yσ(k);t1, bt2ik=bt1σ(i)σ(k) or yk;t2=y−1σ(k);t1, bt2ik=−bt1σ(i)σ(k). In both cases, we have
Fk;t2(xσ(1);v,⋯,ˆxσ(k);v,⋯,xσ(n);v)=Fσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v), |
i.e., Fk;t2(x1;u,⋯,ˆxk;u,⋯,xn;u)=Fσ(k);t1(x1;v,⋯,ˆxσ(k);v,⋯,xn;v). Thus Fk;t2=Fσ(k);t1 as polynomials.
Lemma 3.4. Let A(S) be a skew-symmetrizable cluster algebra with skew-symmetrizer D, and(xt1,yt1,Bt1),(xt2,yt2,Bt2) be two cluster seeds of A(S).If there exists a permutation σ of [1,n] such that xi;t2=xσ(i);t1 for i∈[1,n], then bt2ik=dkdσ(k)bt1σ(i)σ(k).
Proof. Let Pσ be the permutation matrix define by the permutation σ. By the cluster formula (see Theorem 3.5 of [2]), we have Pσ(Bt1D−1)P⊤σ=Bt2D−1. Then Bt2=(PσBt1P⊤σ)(PσD−1P⊤σ)D. The result follows.
By the proof of the first statement and the definition of equivalence for two cluster seeds, we conclude the second statement.
From Theorem 3.3 and Lemma 3.4, we can affirm a conjecture for skew-symmetrizable cluster algebra proposed by Fomin and Zelevinsky in [13], which says every seed of a cluster algebra is uniquely determined by its cluster.
Corollary 3.5. Let A(S) be a skew-symmetrizable cluster algebra with skew-symmetrizer D, and (xt1,yt1,Bt1),(xt2,yt2,Bt2) be two cluster seeds of A(S).If there exists a permutation σ of [1,n] such that xi;t2=xσ(i);t1 for i∈[1,n], then yk;t2=yσ(k);t1, bt2ik=bt1σ(i)σ(k), dk=dσ(k) for any i and k.
The result (1) of Theorem 3.1 shows that when xi;t2=rixσ(i);t1 where i∈[1,n] and ri∈R, then Fi;t2=r′iFσ(i);t1 where r′i∈R. In fact, the proof of result (1) also works for any generalized cluster algebra and any (totally sign-skew-symmetric) cluster algebra with coefficients. The reason is that the proof mainly relies on the Laurent phenomenon and is independent of the form of exchange polynomials. In the meantime, the unit ri such that xi;t2=rixσ(i);t1 can be chosen to be 1 (see Remark 3.2). Although LP algebras, cluster algebras and generalized cluster algebras have different forms of exchange polynomials, and they are not included in each other, they all have the Laurent phenomenon. Thus for cluster algebras or generalized cluster algebras, it is the same method in proving that two clusters up to a permutation imply their corresponding exchange polynomials to be the same up to a permutation.
For cluster algebras or generalized cluster algebras, the equivalence for seeds is defined as two clusters and their corresponding exchange matrices up to a permutation. We know that exchange polynomials are defined by exchange matrices. And exchange polynomials up to a permutation can not imply exchange matrices to be equivalent up to a permutation. So in order to prove the conjecture (1) that each seed is uniquely defined by its cluster, we need to prove that the corresponding exchange matrices are equivalent up to a permutation. For cluster algebras, based on the result that exchange polynomials are the same up to a permutation, we give the proof in Corollary 3.5. For generalized cluster algebras, we will not discuss them more in this article, but will further discuss them in the next work.
Let A(S) be a LP algebra, if there is a seed (xt0,Ft0) of A(S) such that the exchange polynomials in Ft0 are all nontrivial, we say that A(S) is a LP algebra having no trivial exchange relations.
Note that if there is a trivial exchange polynomial in a LP seed (xt0,Ft0), from the definition of LP mutation, this trivial exchange polynomial remain invariant under any sequence of LP mutations. So if A(S) is a LP algebra having no trivial exchange relations, then each exchange polynomial of A(S) is a nontrivial polynomial.
Lemma 3.6. Let A(S) be a LP algebra having no trivial exchange relations, and Σt=(xt,Ft), Σt0=(xt0,Ft0) be two LP seeds of A(S) with xi;t=rixi;t0, where ri is a unit in R for any i≠k. If xk;t=Mxk;t0 for some Laurent monomial M in R[x±11;t0,⋯,ˆxk;t0,⋯,x±1n;t0], then M is a unit in R, and [Σt]=[Σt0].
Proof. Without loss of generality, we assume that ri=1 for i≠k. It does not make difference to the proof.
Assume that M=r∏i≠kxaii;t0=r∏i≠kxaii;t, where r is a unit in R. If there exists some j≠k such that aj<0, then we consider the LP seed (xw,Fw)=μj(Σt0). From the definition of LP mutation, we know that xi;w=xi;t0 for i≠j and xj;wxj;t0=ˆFj;t0(x1;t0,⋯,ˆxj;t0,⋯,xn;t0). Then we have
xk;t=(r∏i≠kxaii;t0)xk;t0=(r∏i≠j,kxaii;w)x−ajj;wˆF−ajj;t0(x1;w,⋯,ˆxj;w,⋯,xn;w)xk;w, |
which can be written as the following equation, from the definition of the exchange Laurent polynomial.
xk;t=(r∏i≠j,kxaii;w)x−ajj;wLF−ajj;t0(x1;w,⋯,ˆxj;w,⋯,xn;w)xk;w, | (3.7) |
where L is a Laurent monomial in R[x±11;w,⋯,ˆxj;w,⋯,x±1n;w]. Thus Eq (3.7) is the expansion of xk;t with respect to xw. Because A(S) has no trivial exchange relations, Fj;t0 is a nontrivial polynomial. And we know that Fj;t0 is irreducible and xs∤Fj;t0 for each s∈[1,n], thus Eq (3.7) will contradict the Laurent phenomenon. So each aj is nonnegative.
Similarly, by considering that xk;t0=M−1xk;t=(r∏i≠kx−aii;t)xk;t, we can get each −aj is nonnegative. Thus each aj is 0, thus M=r is a unit in R. Then by Theorem 3.1, we have [Σt]=[Σt0].
Theorem 3.7. Let A(S) be a LP algebra of rank n having no trivial exchange relations, and Σt1=(xt1,Ft1), Σt2=(xt2,Ft2) be two LP seeds of A(S). If xi;t1=rixi;t2 holds for any i≠k, where ri is a unit in R, then [Σt1]=[Σt2] or [Σt1]=μk[Σt2], that is, any two LP seeds with n>1 common cluster variables are connected with each other by one step of mutation.
Proof. Without loss of generality, we assume that ri=1 for i≠k. It does not make difference to the proof.
By the Laurent phenomenon, we assume that xk;t2=f(x1;t1,⋯,xn;t1) and xk;t1=g(x1;t2,⋯,xn;t2), where f∈R[x±11;t1,⋯,x±1n;t1] and g∈R[x±11;t2,⋯,x±1n;t2]. Since xi;t1=xi;t2 for any i≠k, we know that xk;t1 entries f with exponent 1 or −1; Thus xk;t2 has the form of xk;t2=L1x±1k;t1+L0, where L1≠0 and L0 are Laurent polynomials in
R[x±11;t1,⋯,ˆxk;t1,⋯,x±1n;t1]=R[x±11;t2,⋯,ˆxk;t2,⋯,x±1n;t2]. |
Let (xu,Fu)=μk(Σt2) and (xv,Fv)=μk(Σt1). From the definition of the LP mutation, we know
xi;u={xi;t2if i≠kˆFk;t2/xk;t2if i=k and xi;v={xi;t1if i≠kˆFk;t1/xk;t1if i=k. |
Thus xk;u=ˆFk;t2(x1;t2,⋯,ˆxk;t2,⋯,xn;t2)/xk;t2=ˆFk;t2(x1;t1,⋯,ˆxk;t1,⋯,xn;t1)L1x±1k;t1+L0. From the definition of the exchange Laurent polynomial, we know the above equation has the form of
xk;u=Fk;t2(x1;t1,⋯,ˆxk;t1,⋯,xn;t1)L1x±1k;t1+L0M, | (3.8) |
where M is a Laurent monomial in R[x±11;t1,⋯,ˆxk;t1,⋯,x±1n;t1]. The above equation is just the expansion of xk;u with respect to xt1. By the Laurent phenomenon, and the fact xk;t1∉Fk;t2(x1;t1,⋯,ˆxk;t1,⋯,xn;t1), we obtain that L0=0 and Fk;t2(x1;t1,⋯,ˆxk;t1,⋯,xn;t1)L1 is a Laurent polynomial in R[x±11;t1,⋯,ˆxk;t1,⋯,x±1n;t1]. Thus we have that xk;t2=L1x±1k;t1 and xk;u has the form of xk;u=˜Mx∓1k;t1, where ˜M is a Laurent polynomial in R[x±11;t1,⋯,ˆxk;t1,⋯,x±1n;t1].
We claim that Fk;t2(x1;t1,⋯,ˆxk;t1,⋯,xn;t1)L1 is actually a Laurent monomial, i.e., ˜M is a Laurent monomial in R[x±11;t1,⋯,ˆxk;t1,⋯,x±1n;t1].
Case (i): xk;t2=L1xk;t1. Then xk;t1=L−11xk;t2, which is the expansion of xk;t1 with respect to xt2. By the Laurent phenomenon, we can get that L1 is a Luarent monomial in
R[x±11;t1,⋯,ˆxk;t1,⋯,x±1n;t1]. |
Then by Lemma 3.6, L1 is a unit in R and [Σt1]=[Σt2].
Case (ii): xk;t2=L1x−1k;t1, in this case, xk;u=˜Mxk;t1. By the same argument in case (i), we can get that ˜M is a Luarent monomial in R[x±11;t1,⋯,ˆxk;t1,⋯,x±1n;t1]. Then by Lemma 3.6, ˜M is a unit in R and [Σt1]=[(xu,Fu)]=[μk(Σt2)]=μk([Σt2]).
Remark 3.8. The same method also works for cluster algebras and one can get the similar result.
The following definitions are natural generalizations of the corresponding notions of cluster algebras in [4].
For i∈[1,n], we define the adjacent cluster xi by xi=(x−{xi})∪{x′i} where the cluster variables xi and x′i are related by the exchange Laurent polynomial ˆFi. Let R[x±1] be the ring of Laurent polynomials in x1,…,xn with coefficients in R.
Definition 4.1. The upper bound U(Σ) and lower bound L(Σ) associated with a LP seed Σ=(x,F) is defined by
U(Σ)=R[x±1]∩R[x±11]∩⋯∩R[x±1n], L(Σ)=R[x1,x′1,…,xn,x′n] |
Thus, L(Σ) is the R-subalgebra of F generated by the union of n+1 clusters x±1,x±11,…,x±1n. Note that L(Σ)⊆A(Σ)⊆U(Σ).
Remark 4.2. The method of the proof of results for LP algebras in this section is a little similar to those for cluster algebras in [4]. The concepts of LP algebras and cluster algebras are essentially different, since LP algebras and cluster algebras are not included with each other. In general, the calculation for LP algebras is more complicated. Now, we give the following three points to explain the specific differences between LP algebras and cluster algebras.
(1) For cluster algebras, the exchange polynomials are binomials. While for LP algebras, those are multinomials, so that the calculation using the exchange polynomials becomes complicated.
(2) For cluster algebras, coprimeness is necessary for the proof of properties for the upper bound and lower bound, and it is easy to check that the coprimeness keeps under mutations for a certain seed. While for LP algebras, the concept of coprimeness is not yet defined. In order to obtain a LP seed with coprimeness, we assume that a LP seed satisfies the condition that ˆFk=Fk for any k. But it is not obvious whether the condition keeps under mutations (see Example 4.12 which shows that the condition does not keep under mutations), so that in the proof that involves mutations and requires that condition, we need to show that the condition keeps under mutations.
(3) In cluster algebras, the acyclic seed has good property that the lexicographically first monomial of its any exchange polynomial Fj is a monomial in {xi|i>j}. While in LP algebras, the concept of the acyclic seed is not yet defined. In order to obtain a LP seed with such good property, we assume that the LP seed satisfies certain conditions, see Condition 1.2.
For any LP seed Σ=(x,F), the following lemma and corollary hold parallel to the corresponding results in [4].
Lemma 4.3.
U(Σ)=n⋂j=1R[x±11,…,x±1j−1,xj,x′j,x±1j+1,…,x±1n]. | (4.1) |
Proof. It is sufficient to show that
R[x±1]∩R[x±11]=R[x1,x′1,x±12,…,x±1n]. |
The inclusion ⊇ is clear, we only need to prove the converse inclusion.
For any y∈R[x±1]∩R[x±11], y is of the form y=N∑m=−Mcmxm1, where M,N∈Z≥0 and cm∈R[x±12,…,x±1n]. If M≥0, it is easy to see that
y∈R[x1,x±12,…,x±1n]⊆R[x1,x′1,x±12,…,x±1n]. |
If MN≠0, from the definition of LP seeds, x1∉ˆF1, then
y|x1←ˆF1x′1=∑Nm=−Mcm(ˆF1x′1)m=∑Mm=1c−mˆF−m1x′m1+∑Nm=0cmˆFm1x′−m1. |
Since y∈R[x±11], y can be written as N′∑p=M′cpxp1 where cp∈R[x±12,…,x±1n], then we have c−mˆF−m1∈R[x±12,…,x±1n]. Thus,
y=M∑m=1c−mˆF−m1x′m1+N∑m=0cmxm1∈R[x1,x′1,x±12,…,x±1n]. |
If N=0, by similar discussion, we have y∈R[x′1,x±12,…,x±1n]⊆R[x1,x′1,x±12,…,x±1n].
Corollary 4.4. For j∈[1,n], y∈R[x±11,…,x±1j−1,xj,x′j,x±1j+1,…,x±1n] if and only if y is of the form y=N∑m=−Mcmxmj where M,N∈Z≥0, cm∈R[x±11,…,^xj±1,…,x±1n] and c−m is divisible by ˆFmj in R[x±11,…,^xj±1,…,x±1n] for m∈[1,M].
Lemma 4.5. Suppose that ˆFj=Fj for j∈[1,2], then R[x1,x±12]∩R[x±11,x2,x′2]=R[x1,x2,x′2].
Proof. The inclusion ⊇ is clear, we only need to prove the converse inclusion. For y∈R[x1,x±12]∩R[x±11,x2,x′2], y is of the form y=∑m∈Zxm1(cm+c′m(x2)+c″m(x′2)), where cm∈R, c′m(x2) and c″m(x′2) are polynomials over R without constant terms.
Let M be the smaller integer such that cM+c′M(x2)+c″M(x′2)≠0. If M≥0, then it is easy to see that y∈R[x1,x2,x′2].
Otherwise, the Laurent expression of y is ∑m∈Zxm1(cm+c′m(x2)+c″m(F2x2)) by the assumption. Let r2 be the sum of monomials in F2 without x1. Then there are nonzero terms with smallest power of x1 in the Laurent expression of y, which are xM1(cM+c′M(x2)+c″M(r2x2))≠0, which contradicts the condition that y∈R[x1,x±12].
Lemma 4.6. Suppose that ˆFj=Fj for j∈[1,2], then
R[x1,x′1,x±12]=R[x1,x′1,x2,x′2]+R[x1,x±12]. |
Proof. The inclusion ⊇ is clear, we only need to prove the converse inclusion. It is enough to show that ∀M,N>0, x′N1x−M2∈R[x1,x′1,x2,x′2]+R[x1,x±12].
By the assumption, we have x2x′2=ˆF2=F2=g(x1)+r2, where g(x1)=m∑i=1gixi1, gi∈R and r2≠0∈R since F2 is not divisible by x1. If g(x1)=0, then x−12=r−12x′2, which implies that x′N1x−M2∈R[x1,x′1,x2,x′2].
Otherwise, let p(x1)=−g(x1)r2∈R[x1], then x2x′2=g(x1)+r2 can be written as
x−12=r−12x′2+p(x1)x−12. |
Repeatedly substituting x−12 in the RHS of the above equation by r−12x′2+p(x1)x−12, we obtain x−12=P(x1,x′2)+pN(x1)x−12, where P(x1,x′2)=r−12x′2N−1∑i=0pi(x1)∈R[x1,x′2].
Then we have
x′N1x−M2=x′N1PM(x1,x′2)+x′N1pMN(x1)x−M2+x′N1∑M−1i=1(Mi)(P(x1,x′2))M−i(p(x1)Nx−12)i, | (4.2) |
where the first term of (4.2) that is, x′N1PM(x1,x′2)∈R[x1,x′1,x′2].
For p(x1)=−1r2g(x1)=−1r2m∑i=1gixi1, the smallest power of x1 in pN(x1) is N and the greatest is Nm. Thus we can rewrite pN(x1) in the form xN1(N(m−1)∑i=0pixi1) where pi∈R, implying that for any integer K>0, we have pNK(x1)∈xN1R[x1]. Since x1x′1=ˆF1=F1∈R[x2], we have x′N1pNK(x1)∈R[x1,x2].
Then the middle term of (4.2) is obvious in R[x1,x±12], and the last term of (4.2) is equal to x′N1M−1∑i=1(Mi)(P(x1,F2x2))M−i(p(x1)Nx−12)i∈R[x1,x±12].
Thus we finish the proof.
Proposition 4.7. Suppose that n≥2 and ˆFj=Fj for j∈[1,n], then
U(Σ)=n⋂j=2R[x1,x′1,x±12,…,x±1j−1,xj,x′j,x±1j+1,…,x±1n]. | (4.3) |
Proof. Comparing (4.1) with (4.3), it is sufficient to show that
R[x1,x′1,x2,x′2,x±13,…,x±1n]=R[x1,x′1,x±12,…,x±1n]∩R[x±11,x2,x′2,x±13,…,x±1n]. |
Freeze the cluster variables x3,…,xn and view R[x±13,…,x±1n] as the new ground ring R, then the above equality reduces to
R[x1,x′1,x2,x′2]=R[x1,x′1,x±12]∩R[x±11,x2,x′2]. | (4.4) |
Suppose F1=f(x2)+r1, F2=g(x1)+r2, where r1≠0,r2≠0∈R and f(x2),g(x1) are polynomials over R without constant terms. It is easy to see that Lemma 4.5 and Lemma 4.6 hold for four cases which are: (C1) x2∉F1 and x1∉F2, that is, f(x2)=0 and g(x1)=0; (C2) x2∈F1 and x1∈F2; (C3) x2∉F1 but x1∈F2; (C4) x2∈F1 but x1∉F2. Combining Lemma 4.5 and Lemma 4.6 with the fact that R[x1,x′1,x2,x′2]⊆R[x±11,x2,x′2], we obtain:
R[x1,x′1,x±12]∩R[x±11,x2,x′2]=(R[x1,x′1,x2,x′2]+R[x1,x±12])∩R[x±11,x2,x′2]=R[x1,x′1,x2,x′2]+(R[x1,x±12]∩R[x±11,x2,x′2])=R[x1,x′1,x2,x′2] |
Thus we have (4.4).
Lemma 4.8. For a LP seed (x,F), let x′2 and x″2 be the cluster variables exchanged with x2 in the LP seeds μ2(x,F) and μ2μ1(x,F) respectively, then
R[x1,x′1,x2,x′2,x±13,…,x±1n]=R[x1,x′1,x2,x″2,x±13,…,x±1n]. | (4.5) |
Proof. We can freeze the cluster variables x3,…,xn and view R[x±13,…,x±1n] as the new ground ring R. Then we will prove the following equality can be reduced from (4.5):
R[x1,x′1,x2,x′2]=R[x1,x′1,x2,x″2]. |
We first show that x″2∈R[x1,x′1,x2,x′2].
In (C1), it is easy to see that x″2=rx′2 for certain r∈R, which implies x″2∈R[x1,x′1,x2,x′2].
In (C2), let (x′,F′)=μ1(x,F), then x″2 is obtained by x2x″2=ˆF′2. Recall that x1x′1=ˆF1=F1=f(x2)+r1 and x2x′2=ˆF2=F2=g(x1)+r2, where g(x1)=m∑i=1gixi1, gi∈R and r2≠0∈R. Because F2 depends on x1, from the definition of LP mutations, we have:
1) G2=F2|x1←N2=g(r1x′1)+r2.
2) H2=G2/c, where c is the product of all common factors of giri1 for i∈[1,m] and r2.
3) F′2=MH2=x′m1H2=h(x′1)+r3,
where r3=gmrm1c, h(x′1)=m∑i=1hix′i1, hj={gm−irm−i1/cj∈[1,m−1],r2/cj=m.
By Proposition 2.4, there exist x2 in F′1=F1 in (x′,F′), so that there is no x′1 in F′2/ˆF′2, thus we have ˆF′2=F′2. It follows that
x2x″2=1cr2x′m1+∑m−1j=1hjx′j1+r3=1c(x2x′2−g(x1))x′m1+∑m−1j=1hjx′j1+r3=x2(1cx′2x′m1)−(1cg(x1)x′m1−(∑m−1j=1hjx′j1)−r3), |
where 1cg(x1)x′m1=(m∑i=1gicxi1)x′m1=m∑i=1gic(x1x′1)ix′m−i1=m∑i=1gic(f(x2)+r1)ix′m−i1. Recall that f(x2) is a polynomial in x2 without constant terms, then gmc(f(x2)+r1)m can be written as x2Pm+gmrm1c=x2Pm+r3, where Pm is a polynomial in x2.
For i∈[1,m−1], we have gic(f(x2)+r1)ix′m−i1=x2Pi+giri1cx′m−i1=x2Pi+hm−ix′m−i1, where Pi is a polynomial in x2.
Then 1cg(x1)x′m1−(m−1∑j=1hjx′j1)−r3=x2(m∑i=1Pi), which implies that
x2x″2=x2(1cx′2x′m1−m∑i=1Pi). |
Thus x″2∈R[x1,x′1,x2,x′2].
For (C3) and (C4), it is enough to show for (C3) by symmetry. At this time, F′2 is the same as that in (C2). Since f(x2)=0, (F′1=F1)|x2←F′2/x″2=r1 is not divisible by F′2, so that ˆF′2=F′2. As a consequence, we have x2x″2=x2(1cx′2x′m1). Thus x″2∈R[x1,x′1,x2,x′2].
On the other hand, we can prove similarly that x′2∈R[x1,x′1,x2,x″2]. Then, (4.5) follows truely.
Theorem 4.9. Assume that a LP seed Σ=(x,F) satisfied ˆFj=Fj for j∈[1,n] and Σ′=(x′,F′) is the LP seed obtained from the LP seed Σ by mutation in direction k. Then the corresponding upper bounds coincide, that is, U(Σ)=U(Σ′).
Proof. Without loss of generality, we assume that k=1. Combining Proposition 4.7 and Lemma 4.8, we finish the proof.
Proposition 4.10. If the exchange polynomials of a LP seed satisfy ˆFk=Fk for any k∈[1,n], then Fi≠Fk for any i≠k. Furthermore, any two of the exchange polynomials {Fk|k∈[1,n]}of a LP seed are coprime.
Proof. We will prove by contradiction. If Fi=Fk, then
ˆFi|xk←Fk/x′k=Fi|xk←Fk/x′k=Fk|xk←Fk/x′k=Fk |
for xk∉Fk, implying that Fk divides ˆFi|xk←Fk/x′k, which contradicts the definition of exchange Laurent polynomials.
Besides, since the irreducibility of exchange polynomials for LP seeds, we conclude that the exchange polynomials of a LP seed are coprime under the condition that ˆFk=Fk for any k∈[1,n].
Remark 4.11. When a cluster seed is a LP seed, the coprimeness of the cluster seed is equivalent to the condition that ˆFk=Fk for any k∈[1,n].
Example 4.12. Consider the LP seed (x,F)={(a,b+c),(b,a+c),(c,a+(a+1)b)} over R=Z, which satisfies the condition that ˆFk=Fk for any k∈{a,b,c}. Then the LP seed obtained by mutation at b is
(x′,F′)={(a,1+d),(d,a+c),(c,a+d+1)} |
where d=a+Cb. It is easy to see that ^F′c=F′ca, meaning that the condition that ˆFk=Fk for any k for a LP seed does not keep under mutations.
Definition 4.13. Let Σ=(x,F) be a LP seed, the upper LP algebra ¯A(Σ) defined by Σ is the intersection of the subalgebras U(Σ′) for all LP seeds Σ′ mutation-equivalent to Σ.
Theorem 4.9 has the following direct implication.
Corollary 4.14. Assume that all LP seeds mutation equivalent to a LP seed Σ=(x,F) satisfy the condition that ˆFj=Fj for j∈[1,n], then the upper bound U(Σ) is independent of the choice of LP seeds mutation-equivalent to Σ, and so is equal to the upper LP algebra ¯A(Σ).
Definition 4.15. Let (x,F) be a LP seed. A standard monomial in {xi,x′i|i∈[1,n]} is a monomial that contains no product of the form xix′i.
Let xa=xa11…xann be a Laurent monomial where a∈Zn. For a Laurent polynomial in x1,…,xn, we order the each two terms xa and xa′ lexicographically as follows:
a≺a′ if the first nonzero difference a′j−aj is positive. | (4.6) |
We set the term with the smallest lexicographical order as the first term in a Laurent polynomial.
Theorem 4.16. Assume that a LP seeds Σ=(x,F) satisfies
1) ˆFj=Fj for j∈[1,n].
2) in any Fj, the lexicographically first monomial is of the form
xvj={xvj+1,jj+1⋯xvn,jnj∈[1,n−1]1j=n |
where vj∈Zn−j≥0 for j∈[1,n−1].
Then the standard monomials in x1,x′1,…,xn,x′n form an R-basis for L(Σ).
Proof. The proof is using the same technique as in [4]. We denote the standard monomials in x1,x′1,…,xn,x′n by x(a)=x(a1)1⋯x(an)n, where a=(a1,…,an)∈Zn and
x(ai)i={xaii,ai≥0x′−aii,ai<0. |
Note that x(a) is a Laurent polynomial in x1,…,xn and for any i, we have x(−1)i=x′i=x−1iˆFi=x−1iFi. By the assumption for Fi, it follows that the lexicographically first monomial in x(−1)i is x−1ixvi, then the power of xi in x(ai)i is ai and there is no x1,…,xi−1 in x(ai)i. Then the lexicographically first monomial in x(a) is the product of xaii(ai>0) and xaii(xvi)−ai(ai<0).
We assume that a≺a′ such that ai=a′i for i∈[1,k−1] and ak<a′k. Let P, M and Q be the lexicographically first monomial of k−1∏j=1x(aj)j, x(ak)k and n∏j=k+1x(aj)j respectively. Then the lexicographically first monomial of x(a) is PMQ, similarly that of x(a′) is P′M′Q′.
Since ai=a′i for i∈[1,k−1], we have P=P′=n∏j=1xpjj.
Since ak<a′k and the power of xk in x(ak)k is ak and there is no xi (i∈[1,k−1]) in x(ak)k, we obtain M=xakkn∏j=k+1xmjj and M′=xa′kkn∏j=k+1xm′jj.
And Q=(n∏j=k+1,aj>0xajj)(n∏j=k+1,aj<0xajj(xvj)−aj)=n∏j=k+1xqjj, similarly Q′=n∏j=k+1xq′jj.
It follows that
PMQ=(k−1∏j=1xpjj)(xpk+akk)(n∏j=k+1xpj+mj+qjj), P′M′Q′=(k−1∏j=1xpjj)(xpk+a′kk)(n∏j=k+1xpj+m′j+q′jj). |
Thus PMQ≺P′M′Q′, implying that
if a≺a′, the lexicographically first monomial of x(a)≺that of x(a′). | (4.7) |
The linearly independence of standard monomials over R follows at once from (4.7). Since the product xix′i for any i equals to ^Fi=Fi, which is the linear combination of standard monomials in x1,x′1,…,xn,x′n. Thus they form a basis for L(Σ).
In the following statements, we always assume that Σ=(x,F) is a LP seed of rank n satisfying Condition 1.2.
Notation 4.17. We denote by \mathit{\boldsymbol{\varphi}}:R[x_2, x_2', \dots, x_n, x_n']\rightarrow R[x_2^{\pm1}, \dots, x_n^{\pm1}] the algebra homomorphism defined as the composition \varphi_2\circ\varphi_1 , where
\begin{align} \varphi_1 & : R[x_2, x_2', \dots, x_n, x_n']\rightarrow R[x_1, x_2^{\pm1}, \dots, x_n^{\pm1}]{\text{ by }}x_i\mapsto x_i{\text{ and }}x_i'\mapsto F_i/x_i.\\ \varphi_2 & : R[x_1, x_2^{\pm1}, \dots, x_n^{\pm1}]\rightarrow R[x_2^{\pm1}, \dots, x_n^{\pm1}]{\text{ by }}x_1\mapsto 0{\text{ and }}x_i^{\pm1}\mapsto x_i^{\pm1}. \end{align} |
We denote by R^{st}[x_2, x_2', \dots, x_n, x_n'] (resp. R^{st}[x_1, x_2, x_2', \dots, x_n, x_n'] ) the R -linear span (resp. R[x_1] -linear span) of the standard monomials in x_2, x_2', \dots, x_n, x_n' .
Lemma 4.18. R[x_2, x_2', \dots, x_n, x_n'] = ker(\varphi)\oplus R^{st}[x_2, x_2', \dots, x_n, x_n'] .
Proof. For any y\in R[x_2, x_2', \dots, x_n, x_n'] , replace x_ix_i'\in y with F_i , then y\in R^{st}[x_1, x_2, x_2', \dots, x_n, x_n'] . Thus we have R[x_2, x_2', \dots, x_n, x_n']\subseteq R^{st}[x_1, x_2, x_2', \dots, x_n, x_n'] . It follows that
R[x_2, x_2', \dots, x_n, x_n'] = ker(\varphi)+ R^{st}[x_2, x_2', \dots, x_n, x_n']. |
Similarly using the tool of the proof of Theorem 4.16, For \mathbf{x^{(a)}}\in R^{st}[x_2, x_2', \dots, x_n, x_n'] , the lexicographically first monomial of \varphi(x_j^{(a_j)}) is a Laurent monomial in x_j, x_{j+1}, \dots, x_n whose the power of x_j is a_j , implying that if \mathbf{a}\prec\mathbf{a'} , then the lexicographically first monomial of \varphi(\mathbf{x^{(a)}}) precedes the one of \varphi(\mathbf{x^{(a')}}) .
Then the restriction of \varphi to R^{st}[x_2, x_2', \dots, x_n, x_n'] is injective.
Notation 4.19. Given a Laurent polynomial y\in R[x_1^{\pm1}, \dots, x_n^{\pm1}] , we denote by LT(y) as the sum of all Laurent monomials with the smallest power of x_1 in the Laurent expansion of y with nonzero coefficient.
The following results parallel to Lemmas 6.4 and 6.5 in [4] can be obtained similarly.
Lemma 4.20. Suppose that y = \sum_{m = a}^{b}c_mx_1^m where c_m\in R^{st}[x_2, x_2', \dots, x_n, x_n'] and c_a\neq0 , then LT(y) = \varphi(c_a)x_1^a .
Lemma 4.21. R[x_1, x_2^{\pm1}, \dots, x_n^{\pm1}]\cap R[x_1^{\pm1}, x_2, x_2', \dots, x_n, x_n'] = R[x_1, x_2, x_2', \dots, x_n, x_n'] .
Lemma 4.22. Im (\varphi) = R[x_2, x_2^{(-)}, \dots, x_n, x_n^{(-)}] , where x_j^{(-)} = \begin{cases}x_j', & \mathit{\mbox{if}} x_1\notin F_j \\x_j^{-1}, & \mathit{\mbox{otherwise}}\end{cases}.
Proof. By Condition 1.2, we have \varphi(x_j') = \begin{cases}x_j', & \mbox{if } x_1\notin F_j \\x_j^{-1}M_j, & \mbox{otherwise}\end{cases}. Thus the inclusion \subseteq is clear.
Let J be the set of indexes j\in [2, n] satisfying x_1\notin F_j . We set W_j = x_j^{-1}M_j . To prove the converse inclusion, it is enough to show that x_j^{-1}\in Im( \varphi ) for j\in [2, n]-J .
For \mathbf{m} = (m_2, \dots, m_n), \mathbf{l} = (l_2, \dots, l_n)\in \mathbb{Z}^{n-1} , let \mathbf{x^{m}} be a Laurent monomial in R[x_2^{\pm1}, \dots, x_n^{\pm1}] . Moreover, we set \mathbf{W^{l}} = \prod\limits_{j = 2}^{n}W_j^{l_j} . Then we have any \mathbf{x^{m}} can be written as \mathbf{W^{l}} satisfying
m_j = -l_j+\sum\limits_{2\leq i < j}v_{ji}l_i. |
Define the multiplicative monoid \mathcal{W} = \{\mathbf{x^{m}} = \mathbf{W^{l}} | l_i\geq0\ for\ i\in [2, n]\ and\ m_j\geq0\ for\ j\in J\} . Then \mathbf{W^{l}}\in \mathcal{W} if and only if
\begin{equation} (a)\; l_k\geq0\; \text{for}\; k\in [2, n], \; \;\;(b)\; \sum\limits_{2\leq i < j}v_{ji}l_i\geq l_j\text{ for }j\in J. \end{equation} | (4.8) |
By the equivalence condition (4.8) of \mathbf{W^{l}}\in \mathcal{W} , we obtain x_j^{-1}\in \mathcal{W} for j\in [2, n]-J , implying that it suffices to show that \mathcal{W}\subseteq Im (\varphi) .
For W = \mathbf{W^{l}}\in \mathcal{W} , we prove that W\in Im (\varphi) by induction on the degree of W . When deg(W) = 0 , we have W = 1\in R \subseteq Im (\varphi) . Assume that deg(W) > 0 and for any W'\in \mathcal{W} such that deg(W') < deg(W) , then W'\in Im (\varphi) .
Let j = max\{i|l_i > 0\ in\ W\} , then we have W/W_j\in \mathcal{W} by the equivalence condition (4.8) of \mathbf{W^{l}}\in \mathcal{W} . As a consequence, W/W_j\in Im (\varphi) under the induction assumption. If j\in [2, n]-J , then W_j\in Im (\varphi) so that W = (W/W_j)W_j\in Im (\varphi) .
Otherwise, since l_j > 0 , there exist i\in[2, j-1] such that v_{ji}l_i > 0 , where v_{ji}\neq0 implies that x_j\in M_i . Fix such an index i . By (iv) of Condition 1.2, F_j = f_j(x_i)+M_j and f_j(x_i) = \sum\limits_{t = 1}^{s_j}r_tc_t , where s_j is the number of terms of f_j(x_i) , r_t\in R and c_t = \prod\limits_{p\in [2, n]-{j}}x_p^{\gamma_{pt}} satisfying \gamma_{pt}\in \mathbb{Z}_{\geq0} and \gamma_{it}\neq0 .
From the definition of LP mutations and Condition 1.2(i)(ii), we have x_j' = x_j^{-1}f_j(x_i)+W_j . By multiplying both sides of that equation by W/W_j , we have
(W/W_j)x_j' = x_j^{-1}\sum\limits_{t = 1}^{s_j}r_tc_t(W/W_j)+W. |
Since (W/W_j)x_j'\in Im (\varphi) , we only need to show that for t\in [1, s_j] , x_j^{-1}c_t(W/W_j)\in Im (\varphi) .
Define W' = W_{i}^{l_i'}\cdots W_{j}^{l_j'} , where l_i' = 1 and l_p' = min\{l_p, \sum\limits_{i\leq q < p}v_{pq}l_q'\} for p\in [i+1, j] . Because W/W' = W_{2}^{l_2}\cdots W_{i-1}^{l_{i-1}}W_{i}^{l_i-1}W_{i+1}^{l_{i+1}-l_{i+1}'}\cdots W_{j}^{l_j-l_j'} , the equivalence condition (4.8) of W/W'\in \mathcal{W} can be written as
(a) for k\in[i, j] , l_k-l_k'\geq0 ;
(b) for k\in J , l_k-l_k'\leq\sum\limits_{2\leq h < k}v_{kh}(l_h-l_h') \Leftrightarrow -l_k'+\sum\limits_{2\leq h < k}v_{kh}l_h'\leq -l_k+\sum\limits_{2\leq h < k}v_{kh}l_h .
The inequalities of (a) are immediate from the definition of W' and the choice of i . And for inequality (b), we discuss in several cases:
1) if k\in [2, i-1] , (b) is equivalent to 0\leq -l_k+\sum\limits_{2\leq h < k}v_{kh}l_h .
2) if k = i , we have \sum\limits_{2\leq h < i}v_{kh}l_h' = 0 , (b) is equivalent to -1\leq -l_k+\sum\limits_{2\leq h < k}v_{kh}l_h .
3) if k\in [i+1, n] , when l_k' = l_k , (b) is equivalent to \sum\limits_{2\leq h < k}v_{kh}l_h'\leq\sum\limits_{2\leq h < k}v_{kh}l_h , when l_k'\leq l_k , l_k' = \sum\limits_{2\leq h < k}v_{kh}l_h' , then LHS of (b) is zero.
Since W\in \mathcal{W} and inequalities of (a) hold, we have inequality (b) holds for W/W' . Thus W/W' belongs to \mathcal{W} with deg(W/W') < deg(W) , so that W/W'\in Im (\varphi) .
Then we have
\begin{align} x_j^{-1}c_t(W/W_j) & = W\cdot \prod\nolimits_{p\in [2, n]-{j}}x_p^{\gamma_{pt}}/\mathbf{x^{v_j}}\\ & = (W/W')\cdot (W'\cdot (x_2^{\gamma_{2t}}\cdots x_{j-1}^{\gamma_{j-1, t}})\cdot(x_{j+1}^{\gamma_{j+1, t}-v_{j+1, j}}\cdots x_{n}^{\gamma_{nt}-v_{nj}})) \\ & = (W/W')\cdot P \end{align} |
The claim x_j^{-1}c_t(W/W_j)\in Im (\varphi) is a consequence of the statement that P\in R[x_2, \cdots, x_n] . Indeed, R[x_2, \cdots, x_n]\subseteq Im (\varphi) .
The only variable with negative power (namely, -1 ) in W' is x_i , since
\begin{align} W' & = W_{i}^{1}W_{i+1}^{l_{i+1}'}\cdots W_{j}^{l_j'} \\ & = x_i^{-1}\cdot (x_{i+1}^{v_{i+1, i}-l_{i+1}'}\cdot x_{i+2}^{(\sum\limits_{i\leq h < i+2}v_{i+2, h} \ \ \ \ l_h')-l_{i+2}'}\cdots x_{j}^{(\sum\limits_{i\leq h < j} v_{j, h} \ \ l_h')-l_{j}'})\cdot(x_{j+1}^{\sum\limits_{i\leq h\leq j}v_{j+1, h} \ \ l_h'}\cdots x_{n}^{\sum\limits_{i\leq h\leq j}v_{nh} \ \ l_h'})\\ & = x_i^{-1}\cdot Q \cdot (x_{j+1}^{\delta_{j+1, t}}\cdots x_{n}^{\delta_{nt}}) \end{align} |
where \delta_{pt} = \sum\limits_{i\leq h\leq j}v_{ph}l_h' for p\in [j+1, n] . Then we have
P = Q\cdot (\prod\limits_{q\in[2, i-1]\cup[i+1, j-1]}x_q^{\gamma_{qt}})\cdot x_i^{\gamma_{it}-1}\cdot (\prod\limits_{p\in [j+1, n]}x_p^{\delta_{pt}+\gamma_{pt}-v_{pj}}). |
For i is the fixed index such that \gamma_{it}\in \mathbb{Z}_{ > 0} , \gamma_{it}-1 > 0 , then the power of x_i is nonnegative.
For p\in[j+1, n] , we have
\delta_{pt}+\gamma_{pt}-v_{pj} = v_{pi}l_i'+\cdots+v_{pj}l_j'+\gamma_{pt}-v_{pj}\geq v_{pj}(l_j'-1). |
From the definition of W' , we obtain l_j' = min\{l_j, \sum\limits_{i\leq q < j}v_{jq}l_q'\} , and it is easy to see that l_j\geq1 and \sum\limits_{i\leq q < j}v_{jq}l_q' = v_{ji}+\sum\limits_{i < q < j}v_{jq}l_q'\geq v_{ji}\geq 1 by the choice of i and j . Then l_j'\geq1 . Thus the power of x_p is nonnegative. Hence the power of any cluster variable is nonnegative. It follows that P\in R[x_2, \cdots, x_n] .
By the same technique as in [4], we give the following theorem.
Theorem 4.23. If a LP seed \Sigma = (\mathbf{x, F}) satisfying Condition 1.2, \mathcal{L}(\Sigma) = \mathcal{U}(\Sigma) .
Proof. We apply the induction on n , that is, the rank of the LP seed. When n = 1 , by Lemma 4.3, we have \mathcal{L}(\Sigma) = R[x_1, x_1'] = \mathcal{U}(\Sigma) . Assume that n\geq2 and the statement holds for all algebras of rank 2 to n-1 . Then we consider about rank n .
By Lemma 4.3, we have
\mathcal{U}(\Sigma) = \bigcap\limits_{j = 2}^{n} R[x_1^{\pm1}, \dots, x_{j-1}^{\pm1}, x_j, x_j', x_{j+1}^{\pm1}, \dots, x_n^{\pm1}]\bigcap R[x_1, x_1', x_{2}^{\pm1}, \dots, x_n^{\pm1}]. |
For the seed \Sigma' obtained from \Sigma by freezing at x_1 , by the induction assumption, we have \mathcal{L}(\Sigma') = \mathcal{U}(\Sigma') , that is, \bigcap\limits_{j = 2}^{n}R[x_1^{\pm1}, \dots, x_{j-1}^{\pm1}, x_j, x_j', x_{j+1}^{\pm1}, \dots, x_n^{\pm1}] = R[x_1^{\pm1}, x_2, x_2', \dots, x_n, x_n'] . Then it is enough to show that
\begin{equation} R[x_1, x_1', x_{2}^{\pm1}, \dots, x_n^{\pm1}]\cap R[x_1^{\pm1}, x_2, x_2', \dots, x_n, x_n'] = R[x_1, x_1', \dots, x_n, x_n']. \end{equation} | (4.9) |
The inclusion \supseteq is clear, we only need to prove the converse inclusion.
For \forall y\in LHS of (4.9), let a be the smallest power of x_1 in y|_{x_ix_i'\leftarrow F_i} . Then y can be written as \sum\limits_{m = a}^{b}c_mx_1^m where c_m\in R^{st}[x_2, x_2', \dots, x_n, x_n'] . By Lemma 4.20, we have
LT(y) = \varphi(c_a)x_1^a\in R[x_1^{\pm1}, x_{2}^{\pm1}, \dots, x_n^{\pm1}]. |
If a\geq0 , by Lemma 4.21, we have y\in R[x_1, \dots, x_n, x_n'] \subseteq the RHS of (4.9).
Otherwise, we apply the induction on |a| . Since y\in R[x_1, x_1', x_{2}^{\pm1}, \dots, x_n^{\pm1}] , by Lemma 4.4 we have \varphi(c_a) is divisible by F_1^{|a|} , that is \varphi(c_a) = F_1^{|a|}z_a for certain z_a\in R[x_2^{\pm1}, \dots, x_n^{\pm1}] .
When J = \varnothing , we have Im (\varphi) = R[x_2^{\pm1}, \dots, x_n^{\pm1}] according to Lemma 4.22. Then z_a\in Im (\varphi) .
When J\neq\varnothing , we consider the LP seed \Sigma^{\ast} obtained from \Sigma by freezing at \{x_j|j\in[2, n]-J\} and removing x_1 . In view of Lemma 4.22, we have \mathcal{L}(\Sigma^{\ast}) = Im (\varphi) . Besides, by the induction assumption, we have \mathcal{L}(\Sigma^{\ast}) = \mathcal{U}(\Sigma^{\ast}) . Using Lemma 4.3, we obtain Im (\varphi) = \bigcap\limits_{j\in J}R[x_2^{\pm1}, \dots, x_j, x_j', \dots, x_n^{\pm1}] .
For certain j\in J , z_a can be written as \sum\limits_{s\in \mathbb{Z}}c_sx_j^s , where c_s\in R[x_2^{\pm1}, \dots, \hat{x_j}, \dots, x_n^{\pm1}] . Since F_1^{|a|}z_a = \sum\limits_{s\in \mathbb{Z}}(c_sF_1^{|a|})x_j^s\in Im (\varphi)\subseteq R[x_2^{\pm1}, \dots, x_j, x_j', \dots, x_n^{\pm1}] , by Corollary 4.4 c_sF_1^{|a|} is divisible by F_j^{|s|} for s < 0 . By Proposition 4.10, F_1 and F_j are coprime, implying that c_s is divisible by F_j^{|s|} . Using Corollary 4.4 again, we have z_a\in R[x_2^{\pm1}, \dots, x_j, x_j', \dots, x_n^{\pm1}] .
By the arbitrariness of j\in J , we obtain z_a\in Im (\varphi) .
Then there exist c_a'\in R[x_2^{\pm1}, \dots, x_n^{\pm1}] such that z_a = \varphi(c_a') . It implies that
LT(y) = \varphi(c_a)x_1^a = F_1^{|a|}z_ax_1^a = \varphi(c_a')F_1^{|a|}x_1^a = \varphi(c_a')x_1'^{|a|}. |
Then we have y = \sum\limits_{m = a}^{b}c_mx_1^m = \sum\limits_{m = a}^{-1}c_m'x_1'^{|m|}+\sum\limits_{m = 0}^{b}c_mx_1^m\in R[x_1, x_1', \dots, x_n, x_n'] .
Corollary 4.24. If a LP seed \Sigma = (\mathbf{x, F}) satisfied Condition 1.2, then the standard monomials in x_1, x_1', \dots, x_n, x_n' form an R -basis of the LP algebra \mathcal{A}(\Sigma) .
Proof. It is immediately from Theorem 4.16 and Theorem 4.9.
Example 4.25. Consider the LP seed (\mathbf{x, F}) = \{(a, bcd+1), (b, a+cd), (c, bd+1), (d, 1+abc)\} , it is easy to see that Condition 1.2 (i) (ii) (iii) hold. Since M_c = 1 and b|(F_c-M_c) , (iv) of Condition 1.2 holds. Besides, \varphi:\ b'\mapsto \frac{cd}{b}, \ c'\mapsto \frac{bd+1}{c} = c', \ d'\mapsto \frac{1}{d} , Then it is clear that d^{-1}\in Im (\varphi) and b^{-1} = \varphi(b'c'd'-d)\in Im (\varphi) . Thus by Theorem 4.9, we have \mathcal{L}(\Sigma) = \mathcal{U}(\Sigma) .
Note that this LP seed is not a cluster seed or a generalized cluster seed for c\in F_a since a\notin F_c .
The cluster seed is acyclic if and only if there exist a permutation \sigma such that for i > j , b_{\sigma(i), \sigma(j)}\geq0 . Renumbering if necessary the indexes of the initial acyclic cluster, we assume that for i > j , b_{ij}\geq0 . Then by the exchange polynomials for cluster algebras, we conclude that the cluster seed is acyclic if and only if for any j , F_j = \frac{y_j}{1\oplus y_j}\prod\limits_{i > j}x_i^{b_{ij}}+\frac{y_j}{1\oplus y_j}\prod\limits_{i < j}x_i^{-b_{ij}} .
Proposition 4.26. Condition 1.2 is equivalent to acyclicity and coprimeness of exchange polynomials for a cluster seed which is a also LP seed.
Proof. When a cluster seed is a LP seed, recall that (i) in Condition 1.2 is equivalent to coprimeness of exchange polynomials by Remark 4.11.
When a cluster seed satisfies the conditions (i) and (ii), for any j\in [2, n-1] , since F_j is a binomial, we have F_j = \mathbf{x^{v_j}}+\mathbf{x^{b_j}} . If x_i\in \mathbf{x^{b_j}} for some i > j , then x_j\in \mathbf{x^{v_i}} for b_{ji}b_{ij} < 0 , which contradicts to the condition (ii). For j = n , F_n = 1+\mathbf{x^{b_n}} . From the definition of exchange polynomials for cluster algebras, we have \mathbf{x^{b_n}} is of the form x_1^{|b_{1n}|}\cdots x_{n-1}^{|b_{n-1, n}|} . For j = 1 , since for any j > 1 , we have b_{j1} > 0 by the above discussion, so that we obtain F_1 = 1+x_2^{|b_{12}|}\cdots x_{n}^{|b_{1n}|} . Then the cluster seed satisfied the conditions (i) and (ii) is acyclic. Besides, it is easy to see that when a cluster seed is acyclic, it satisfies the conditions (i) and (ii).Thus the conditions (i) and (ii) are equivalent to acyclicity of exchange polynomials.
Under the conditions of acyclicity and coprimeness, the cluster seed in fact satisfies (iii) and (iv) in Condition 1.2.
This project is supported by the National Natural Science Foundation of China (No.12071422 and No.12131015).
The authors declare there is no conflicts of interest.
[1] |
L. A. Zadeh, Fuzzy sets, Information and Control, 8 (1965), 338–353. https://doi.org/10.1016/S0019-9958(65)90241-X doi: 10.1016/S0019-9958(65)90241-X
![]() |
[2] | I. Kramosil, J. Michalek, Fuzzy metric and statistical metric spaces, Kybernetika, 11 (1975), 336–344. |
[3] |
A. George, P. Veeramani, On some results in fuzzy metric spaces, Fuzzy Set. Syst., 64 (1994), 395–399. https://doi.org/10.1016/0165-0114(94)90162-7 doi: 10.1016/0165-0114(94)90162-7
![]() |
[4] |
M. Grabiec, Fixed points in fuzzy metric spaces, Fuzzy Set. Syst., 27 (1988), 385–389. https://doi.org/10.1016/0165-0114(88)90064-4 doi: 10.1016/0165-0114(88)90064-4
![]() |
[5] |
F. Kiany, A. Amini-Haradi, Fixed point and endpoint theorems for set-valued fuzzy contraction maps in fuzzy metric spaces, Fixed Point Theory Appl., 2011 (2011), 94. https://doi.org/10.1186/1687-1812-2011-94 doi: 10.1186/1687-1812-2011-94
![]() |
[6] |
J. P. Aubin, J. Siegel, Fixed points and stationary points of dissipative multivalued maps, Proc. Amer. Math. Soc., 78 (1980), 391–398. https://doi.org/10.1090/S0002-9939-1980-0553382-1 doi: 10.1090/S0002-9939-1980-0553382-1
![]() |
[7] |
M. Fakhar, Endpoints of set-valued asymptotic contractions in metric spaces, Appl. Math. Lett., 24 (2010), 428–431. https://doi.org/10.1016/j.aml.2010.10.028 doi: 10.1016/j.aml.2010.10.028
![]() |
[8] |
V. Gregori, A. Sapena, On fixed-point theorems in fuzzy metric spaces, Fuzzy Set. Syst., 125 (2002), 245–252. https://doi.org/10.1016/S0165-0114(00)00088-9 doi: 10.1016/S0165-0114(00)00088-9
![]() |
[9] |
A. A. Harandi, Endpoints of set-valued contractions in metric spaces, Nonlinear Anal. Theor., 72 (2010), 132–134. https://doi.org/10.1016/j.na.2009.06.074 doi: 10.1016/j.na.2009.06.074
![]() |
[10] |
N. Hussain, A. A. Harandi, Y. J. Cho, Approximate endpoints for set-valued contractions in metric spaces, Fixed Point Theory Appl., 2010 (2010), 614867. https://doi.org/10.1155/2010/614867 doi: 10.1155/2010/614867
![]() |
[11] |
N. Mizoguchi, W. Takahashi, Fixed point theorems for multivalued mappings on complete metric spaces, J. Math. Anal. Appl., 141 (1989), 177–188. https://doi.org/10.1016/0022-247X(89)90214-X doi: 10.1016/0022-247X(89)90214-X
![]() |
[12] |
S. U. Rehman, H. Aydi, G.-X. Chen, S. Jabeen, S. U. Khan, Some set-valued and multi-valued contraction results in fuzzy cone metric spaces, J. Inequal. Appl., 2021 (2021), 110. https://doi.org/10.1186/s13660-021-02646-3 doi: 10.1186/s13660-021-02646-3
![]() |
[13] |
K. Wlodarczyk, D. Klim, R. Plebaniak, Existence and uniqueness of endpoints of closed set-valued asymptotic contractions in metric spaces, J. Math. Anal. Appl., 328 (2007), 46–57. https://doi.org/10.1016/j.jmaa.2006.05.029 doi: 10.1016/j.jmaa.2006.05.029
![]() |
[14] |
C. D. Bari, C. Vetro, A fixed point theorem for a family of mappings in fuzzy metric space, Rend. Circ. Mat. Palermo, 52 (2003), 315–321. https://doi.org/10.1007/bf02872238 doi: 10.1007/bf02872238
![]() |
[15] |
I. Beg, M. Abbas, Invariant approximation for fuzzy nonexpansive mappings, Math. Bohem., 136 (2011), 51–59. https://doi.org/10.21136/MB.2011.141449 doi: 10.21136/MB.2011.141449
![]() |
[16] |
I. Beg, S. Sedghi, N. Shobe, Fixed point theorems in fuzzy metric spaces, Int. J. Anal., 2013 (2013), 934145. https://doi.org/10.1155/2013/934145 doi: 10.1155/2013/934145
![]() |
[17] | C. D. Bari, C. Vetro, Fixed points, attractors and weak fuzzy contractive mappings in a fuzzy metric space, J. Fuzzy Math., 1 (2005), 973–982. |
[18] | M. Imdad, J. Ali, Some common fixed point theorems in fuzzy metric spaces, Math. Commun., 11 (2006), 153–163. |
[19] |
A. F. R. L-de Hierro, A. F. Fulga, A. Karapinar, N. Shahzad, Proinov-type fixed-point results in non-archimedean fuzzy metric spaces, Mathematics, 9 (2021), 1594. https://doi.org/10.3390/math9141594 doi: 10.3390/math9141594
![]() |
[20] |
M. Jleli, E. Karapinar, B. Samet, On cyclic (\psi, \phi)-contractions in Kaleva-Seikkala's type fuzzy metric spaces, J. Intell. Fuzzy Syst., 27 (2014), 2045–2053. https://doi.org/10.3233/IFS-141170 doi: 10.3233/IFS-141170
![]() |
[21] |
X. Li, S. U. Rehman, S. U. Khan, H. Aydi, J. Ahmad, N. Hussain, Strong coupled fixed point results and applications to Urysohn integral equations, Dynam. Syst. Appl., 30 (2021), 197–218. https://doi.org/10.46719/dsa20213023 doi: 10.46719/dsa20213023
![]() |
[22] | B. D. Pant, S. Chauhan, Common fixed point theorems for two pairs of weakly compatible mappings in menger spaces and fuzzy metric spaces, Sci. Stud. Res., 21 (2011), 81–96. |
[23] |
J. Rodriguez-Lopez, S. Romaguera, The Haudorff fuzzy metric on compact sets, Fuzzy Set. Syst., 147 (2008), 273–283. https://doi.org/10.1016/j.fss.2003.09.007 doi: 10.1016/j.fss.2003.09.007
![]() |
[24] |
S. U. Rehman, I. Shamas, N. Jan, A. Gumaei, M. Al-Rakhami, Some coincidence and common fixed point results in fuzzy metric space with an application to differential equations, J. Funct. Space., 2021 (2021), 9411993. https://doi.org/10.1155/2021/9411993 doi: 10.1155/2021/9411993
![]() |
[25] |
S. Chauhan, M. A. Khan, S. Kumar, Unified fixed point theorems in fuzzy metric spaces via common limit range property, J. Inequal. Appl., 2013 (2013), 182. https://doi.org/10.1186/1029-242X-2013-182 doi: 10.1186/1029-242X-2013-182
![]() |
[26] |
A.-F. Roldán-López-de-Hierro, E. Karapinar, S. Manro, Some new fixed point theorems in fuzzy metric spaces, J. Intell. Fuzzy Syst., 27 (2014), 2257–2264. https://doi.org/10.3233/IFS-141189 doi: 10.3233/IFS-141189
![]() |
[27] |
Z. Sadeghi, S. M. Vaezpour, C. Park, R. Saadati, C. Vetro, Set-valued mappings in partially ordered fuzzy metric spaces, J. Inequal. Appl., 2014 (2014), 157. https://doi.org/10.1186/1029-242X-2014-157 doi: 10.1186/1029-242X-2014-157
![]() |
[28] |
I. Shamas, S. U. Rehman, H. Aydi, T. Mahmood, E. Ameer, Unique fixed-point results in fuzzy metric spaces with an application to Fredholm integral equations, J. Funct. Space., 2021 (2021), 4429173. https://doi.org/10.1155/2021/4429173 doi: 10.1155/2021/4429173
![]() |
[29] |
I. Shamas, S. U. Rehman, N. Jan, A. Gumaei, M. Al-Rakhami, A new approach to fuzzy differential equations using weakly-compatible self-mappings in fuzzy metric spaces, J. Funct. Space., 2021 (2021), 6123154. https://doi.org/10.1155/2021/6123154 doi: 10.1155/2021/6123154
![]() |
[30] | T. Som, Some results on common fixed point in fuzzy metric spaces, Soochow Journal of Mathematics, 33 (2007), 553–561. |
[31] |
S. U. Rehman, R. Chinram, C. Boonpok, Rational type fuzzy-contraction results in fuzzy metric spaces with an application, J. Math., 2021 (2021), 6644491. https://doi.org/10.1155/2021/6644491 doi: 10.1155/2021/6644491
![]() |
[32] |
B. Schweizer, A. Sklar, Statical metric spaces, Pac. J. Math., 10 (1960), 313–334. https://doi.org/10.2140/pjm.1960.10.313 doi: 10.2140/pjm.1960.10.313
![]() |
[33] |
A. R. Alharbi, M. B. Almatrafi, A. R. Seadawy, Construction of the numerical and analytical wave solutions of the Joseph-Egri dynamical equation for the long waves in nonlinear dispersive systems, Int. J. Mod. Phys. B, 34 (2020), 2050289. https://doi.org/10.1142/S0217979220502896 doi: 10.1142/S0217979220502896
![]() |
[34] |
J. Chesters, R. Mottonen, K. E. Watkins, Transcranial direct current stimulation over left inferior frontal cortex improves speech fluency in adults who stutter, Brain, 141 (2018), 1161–1171. https://doi.org/10.1093/brain/awy011 doi: 10.1093/brain/awy011
![]() |
[35] |
Y. S. Ozkan, A. R. Seadawy, E. Yasar, Multi-wave, breather and interaction solutions to (3+1) dimensional Vakhnenko-Parkes equation arising at propagation of high-frequency waves in a relaxing medium, J. Taibah Univ. Sci., 15 (2021), 666–678. https://doi.org/10.1080/16583655.2021.1999053 doi: 10.1080/16583655.2021.1999053
![]() |
[36] |
A. R. Seadawy, K. K. Ali, R. Nuruddeen, A variety of soliton solutions for the fractional Wazwaz-Benjamin-Bona-Mahony equations, Results Phys., 12 (2019), 2234–2241. https://doi.org/10.1016/j.rinp.2019.02.064 doi: 10.1016/j.rinp.2019.02.064
![]() |
[37] |
U. Younas, A. R. Seadawy, M. Younis, S. T. R. Rizvi, Dispersive of propagation wave structures to the Dullin-Gottwald-Holm dynamical equation in a shallow water waves, Chinese J. Phys., 68 (2020), 348–364. https://doi.org/10.1016/j.cjph.2020.09.021 doi: 10.1016/j.cjph.2020.09.021
![]() |
[38] | V. Lakshmikantham, R. Mohapatra, Theory of fuzzy differential equations and inclusion, London: CRC Press, 2003. https://doi.org/10.1201/9780203011386 |
1. | Junyuan Huang, Xueqing Chen, Fan Xu, Ming Ding, Some properties of generalized cluster algebras of geometric type, 2024, 660, 00218693, 270, 10.1016/j.jalgebra.2024.07.013 |