The Zariski cancellation problem for Poisson algebras
Jason Gaddis, Xingting Wang

TL;DR
This paper investigates the Zariski cancellation problem for Poisson algebras, establishing positive results for specific classes and introducing new invariants to analyze broader cases.
Contribution
It provides affirmative solutions for certain classes of Poisson algebras and introduces Poisson analogues of invariants to address the cancellation problem more generally.
Findings
Confirmed Zariski cancellation for connected graded Poisson algebras without degree one Poisson central elements.
Proved cancellation for Poisson integral domains of Krull dimension two with nontrivial brackets.
Developed Poisson versions of the Makar-Limanov invariant and discriminant for broader analysis.
Abstract
We study the Zariski cancellation problem for Poisson algebras asking whether implies when and are Poisson algebras. We resolve this affirmatively in the cases when and are both connected graded Poisson algebras finitely generated in degree one without degree one Poisson central elements and when is a Poisson integral domain of Krull dimension two with nontrivial Poisson bracket. We further introduce Poisson analogues of the Makar-Limanov invariant and the discriminant to deal with the Zariski cancellation problem for other families of Poisson algebras.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
The Zariski cancellation problem for Poisson algebras
Jason Gaddis
Miami University, Department of Mathematics, 301 S. Patterson Ave., Oxford, Ohio 45056
and
Xingting Wang
Department of Mathematics, Howard University, Washington DC, 20059
Abstract.
We study the Zariski cancellation problem for Poisson algebras asking whether implies when and are Poisson algebras. We resolve this affirmatively in the cases when and are both connected graded Poisson algebras finitely generated in degree one without degree one Poisson central elements and when is a Poisson integral domain of Krull dimension two with nontrivial Poisson bracket. We further introduce Poisson analogues of the Makar-Limanov invariant and the discriminant to deal with the Zariski cancellation problem for other families of Poisson algebras.
Key words and phrases:
Zariski cancellation problem, Poisson algebra, Locally nilpotent derivation, Discriminant
2010 Mathematics Subject Classification:
17B36, 16W25, 14R10
Contents
- 1 Introduction
- 2 Preliminaries
- 3 Cancellation properties for Poisson algebras
- 4 Graded versus ungraded
- 5 Artinian center and detectability
- 6 The Makar-Limanov invariant and retractability
- 7 Poisson discriminant and effectiveness
- 8 Some remarks and questions
1. Introduction
It is both surprising and interesting to know that many fundamental questions about the geometry and symmetries of the affine space , which is a basic object in algebraic geometry, still remain open in the 21st century. Among those open questions is the cancellation property, which was asked by Zariski in the following sense.
Question 1.1** ((Zariski Cancellation Problem)).**
Does an isomorphism imply an isomorphism , for any affine variety ?
The interested reader is directed to the survey by Gupta for further background on this problem [18]. In recent years, several works have extended the Zariski cancellation problem from commutative algebraic geometry to noncommutative projective algebraic geometry, where the cancellation property is studied for certain types of Artin-Schelter regular algebras, which are noncommutative graded analogues of commutative polynomial rings; see [4, 5, 25, 26].
In this paper, we extend the original Zariski cancellation problem in a different direction. Instead of generalizing the affine variety into a noncommutative algebraic variety, which is represented by a noncommutative algebra as its coordinate ring, we assume to have extra structure, namely we assume that there exists a bivector satisfying a vanishing Schouten-Nijenhuis bracket . In algebraic terms, is an affine Poisson variety whose coordinate ring turns out to be a commutative Poisson algebra. Therefore, we are interested in the following question.
Question 1.2** ((Zariski Cancellation Problem for Poisson Algebras)).**
When is a Poisson algebra cancellative? That is, when does an isomorphism of Poisson algebras for another Poisson algebra imply an isomorphism as Poisson algebras?
The notion of the Poisson bracket, first introduced by Siméon Denis Poisson, arises naturally in Hamiltonian mechanics and differential geometry. Poisson algebras have become deeply entangled with non-commutative geometry, integrable systems, and topological field theories. They are essential in the study of the noncommutative discriminant [6, 33] and representation theory of noncommutative algebras [43, 42]. In addition, there has been renewed interest in enveloping algebras of Poisson algebras [27, 28].
Recently, Adjamagbo and van den Essen [2] proved that the famous Jacobian conjecture for polynomial algebras has an equivalent statement for Poisson algebras. As pointed out by van den Essen [40], the Zariski cancellation problem (especially in dimension two) is closely related to the Jacobian conjecture. Therefore, one of our motivations is to study the Zariski cancellation problem for Poisson algebras with a potential link to the above-mentioned Poisson version of the Jacobian conjecture.
While we are searching for general methods and techniques in this direction, we find many theories developed by Bell and Zhang in the (associative) noncommutative setting [4, 5] can be adapted to the Poisson setting. Their work relies both on the idea of the noncommutative discriminant [7, 8] and the Makar-Limanov invariant [30]. Other work on noncommutative Zariski cancellation has focused on path algebras and their quotients [12, 25] as well as a Morita invariance version [26].
In Section 2, we provide background on Poisson algebras and related concepts that are critical to our study. Section 3 continues this and introduces the Zariski cancellation problem for Poisson algebras, where we discuss various versions of the Poisson cancellation property and the relations between them. In Section 4 we restrict our attention to graded Poisson algebras and establish a graded Poisson version of the Zariski cancellation problem (Theorem 4.5). This is a consequence of the following theorem, which is an analogue of a result of Bell and Zhang in the graded (associative) algebra setting [4].
Theorem 1.3** ((Theorem 4.2)).**
Let and be two connected graded Poisson algebras finitely generated in degree one. If as ungraded Poisson algebras, then as graded Poisson algebras.
This result is then applied to the family of skew quadratic Poisson algebras to obtain all possible isomorphisms between them based on the coefficient matrices given by their Poisson brackets (Theorem 4.6).
Much of our remaining work makes use of the Poisson center and this turns out to be a suitable replacement for the (algebra) center used in [5]. We study the Poisson center and its implications for Poisson cancellation in Section 5.
Theorem 1.4**.**
Let be a Poisson algebra.
- (1)
(Corollary 5.4) If is noetherian with artinian Poisson center, then is Poisson cancellative. 2. (2)
(Theorem 5.5) If has trivial Poisson center, then is Poisson cancellative.
Theorem 1.4 (2) can be applied to show that Poisson integral domains of Krull dimension two which have nontrivial Poisson brackets are Poisson cancellative (Corollary 5.6). Here the non-triviality of the Poisson bracket plays an essential role since by [9, 41] there are commutative domains of Krull dimension two that are not cancellative.
It is well-known that locally nilpotent derivations are important in the study of cancellation. This is the crux of the work of Makar-Limanov [30]. Bell and Zhang exploit these ideas to great effect in their work on cancellation [5]. We further these ideas by establishing a connection between locally nilpotent Poisson derivations and Poisson cancellation in Section 6.
Theorem 1.5** ((Theorem 6.12)).**
Assume is a field of characteristic zero. Let be an affine Poisson domain over with finite Krull dimension. If has no nontrivial locally nilpotent Poisson derivations, then is Poisson cancellative.
In positive characteristic, we can prove a similar result by replacing Poisson derivations by higher Poisson derivations introduced by Launois and Lecoutre [23].
In Section 7 we introduce the Poisson discriminant as well as the notion of effectiveness for these discriminants. We show that effectiveness controls the locally nilpotent Poisson derivations, which in the noncommutative setting was first observed by Bell and Zhang [5] where discriminants are defined for noncommutative algebras that are module-finite over their centers.
Theorem 1.6** ((Theorem 7.16)).**
Let be an affine Poisson domain with affine Poisson center. If the Poisson discriminant exists and is effective either in or its Poisson center, then is Poisson cancellative.
It is important to mention that for Poisson algebras in characteristic zero their Poisson centers are usually not large enough for us to emulate the definition of discriminants for noncommutative algebras by simply replacing the algebraic center by Poisson center. Hence, we follow the idea in [26, §2] to introduce the notion of Poisson discriminant from a representation-theoretic point of view. This is leveraged to study a variety of cancellation results related to Poisson algebras for which we can identify a discriminant relative to some property of Poisson algebras. We give two specific examples: one is the affine space with a nontrivial unimodular Poisson bracket (Example 7.17) and the other is derived from the Poisson order on the center of a three-dimensional PI Sklyanin algebra (Example 7.18).
Finally, in Section 8 we discuss relations between various concepts we introduce in this paper related to the Zariski cancellation problem for Poisson algebras. We further post several open questions that are continuations of the topics that are covered in this paper.
Acknowledgements. Part of this research work was done during the second author’s visit to the Department of Mathematics at Miami University in March 2019. He is grateful for the first author’s invitation and wishes to thank Miami University for its hospitality. The authors would also like to thank Jason Bell and James Zhang for helpful conversations and suggestions.
2. Preliminaries
Throughout the paper, we work over a base field . Many of our results still work when is only a commutative domain. We will leave the reader to figure out these cases.
2.1. Poisson algebras
A Poisson algebra (over ) is a commutative -algebra equipped with a bilinear product , called a Poisson bracket, such that is a Lie algebra under and the map is a -derivation of for all . When is a Poisson algebra, the Poisson center of is denoted by
[TABLE]
We say has trivial Poisson center if . A homomorphism of Poisson algebras is an algebra homomorphism such that for all , . An isomorphism of Poisson algebras is a bijective Poisson homomorphism. An ideal of a Poisson algebra is a Poisson ideal if . If is a Poisson ideal of , then is a Poisson algebra with bracket for all .
The Gelfand-Kirillov (GK) dimension of an affine Poisson algebra (or more generally, a -affine associative algebra ) is defined to be
[TABLE]
where varies over all finite-dimensional -vector subspaces of . If is affine commutative, then [22, Theorem 4.5], where denotes the Krull dimension of .
2.2. Locally nilpotent (Poisson) derivations
Denote the space of (-)derivations (resp. locally nilpotent (-)derivations) of an algebra by (resp. ). A higher derivation (or Hasse-Schmidt derivation) on is a sequence of -linear endomorphisms such that
[TABLE]
The collection of higher derivations of is denoted by . A higher derivation is called iterated if for all . A higher derivation is called locally nilpotent if
- (1)
for all there exists such that for all , 2. (2)
the map defined by and , for all , is an algebra automorphism of .
The collection of locally nilpotent higher derivations of is denoted by .
Now let be a Poisson algebra. A derivation of is called a Poisson derivation if
[TABLE]
We denote the space of Poisson derivations of by and the space of locally nilpotent Poisson derivations of by . Thus, we have the inclusions and . A higher Poisson derivation on is a higher derivation of such that
[TABLE]
If, in addition, is locally nilpotent and the map defined above is a Poisson algebra automorphism, then is said to be a locally nilpotent higher Poisson derivation. We denote the collection of locally nilpotent higher Poisson derivations of by .
2.3. Extensions of Poisson algebras
If and are Poisson algebras, then there is a natural Poisson structure on defined by extending linearly the bracket
[TABLE]
Thus, the polynomial algebra has a natural bracket, extending the bracket on , defined by setting for all . We will often identify and with their images under the obvious embeddings and , respectively.
The following lemma will be used frequently in this paper without mentioning it explicitly.
Lemma 2.1**.**
Let be a Poisson algebra and let be an affine Poisson algebra with trivial Poisson bracket. Then .
Proof.
It is clear from (1) that . Let with all ’s being linearly independent over . Assume some . Then there exists such that . Thus,
[TABLE]
It follows that the component is nonzero and so . ∎
Also there is a natural Poisson structure on defined by extending linearly the bracket
[TABLE]
Suppose is an idempotent of a commutative algebra . Then can be decomposed into a direct sum of two algebras and . If is a Poisson algebra, then and are Poisson algebras with Poisson bracket inherited from .
Let be a Poisson algebra and let be a Poisson derivation of . A linear map is called a Poisson -derivation if it satisfies
- (1)
; 2. (2)
,
for all . Let be a Poisson -derivation of . The Poisson-Ore extension is the polynomial ring with Poisson bracket
[TABLE]
For simplicity, we write and for any .
2.4. Filtered and graded Poisson algebras
An ascending Poisson -filtration on a Poisson algebra is a collection of subspaces satisfying
- (1)
, 2. (2)
, 3. (3)
, and 4. (4)
, for all .
If is a Poisson algebra with a Poisson -filtration , then the associated graded Poisson algebra is defined as
[TABLE]
with . We drop the subscript if the filtration is implied. Similarly, we can define a descending Poisson -filtration on a Poisson algebra .
Example 2.2**.**
(1) Let be the first Poisson Weyl algebra. That is, with Poisson bracket . Set the ascending filtration on by defining and to have degree 1 with being the span of all monomials with degree at most . Then is the polynomial algebra with trivial Poisson bracket.
(2) Any Poisson ideal of a Poisson algebra defines a descending -adic Poisson -filtration on by setting for all (). We denote the corresponding associated graded Poisson algebra by .
Let be a Poisson algebra. When for some Poisson -filtration , then is said to be -graded and in this case we generally write for . Additionally, if we say is connected graded and we say is generated in degree one if is generated by as an -algebra.
While every Poisson filtration on a Poisson algebra naturally restricts to a filtration on the underlying associative algebra, not every algebra filtration on a Poisson algebra extends to a Poisson filtration.
3. Cancellation properties for Poisson algebras
The main goal of our paper is to study the Zariski cancellation problem for Poisson algebras (abbreviated as PZCP). It can be thought of as a natural extension of the Zariski cancellation problem for (noncommutative) algebras (abbreviated as ZCP) [5, 25, 26]. We regard all isomorphisms as isomorphisms of Poisson algebras unless otherwise noted.
Definition 3.1**.**
A Poisson algebra is universally Poisson cancellative if implies for every Poisson algebra and every finitely generated commutative domain over with trivial Poisson bracket such that the natural map is an isomorphism for some Poisson ideal . In the special case that (resp. for all ), we say is Poisson cancellative (resp. strongly Poisson cancellative).
Our first concern is to construct examples of non-cancellative Poisson algebras with nontrivial Poisson bracket. The following lemma provides us an easy way of producing non-cancellative Poisson algebras from non-cancellative commutative algebras.
Lemma 3.2**.**
Let be a commutative algebra that is not (strongly/universally) cancellative. Let be a Poisson algebra with trivial Poisson center. Then is not (strongly/universally) Poisson cancellative.
Proof.
We only prove the result for the universally Poisson cancellative case. The cancellative and strongly cancellative cases can be proved similarly. Since is not universally cancellative, there exist another commutative algebra and an affine commutative domain together with an ideal such that is an isomorphism such that . As a consequence, . If were universally Poisson cancellative, it would imply that and hence
[TABLE]
a contradiction. So is not universally Poisson cancellative. ∎
Example 3.3**.**
The following examples of commutative algebras are known to be non-cancellative.
- (1)
Hochster showed that the coordinate ring of the real sphere is not cancellative [20]. 2. (2)
Let and let be the coordinate ring of the complex surface . Danielewski proved that is not cancellative. More precisely, but for all [41]. 3. (3)
Gupta showed that the polynomial ring is not cancellative whenever and [16, 17].
Now by Lemma 3.2, we can take tensor products of the non-cancellative commutative algebras listed above with Poisson algebras with trivial Poisson center (see Corollary 5.6 and Example 5.7) to produce non-cancellative Poisson algebras.
The next lemma is useful in cancellation problems. The proofs are standard.
Lemma 3.4**.**
Let and be two Poisson algebras such that for some integer . Then the following hold.
- (1)
has trivial Poisson bracket if and only if does. 2. (2)
is noetherian if and only if is. 3. (3)
. 4. (4)
is artinian if and only if is. 5. (5)
is a field if and only if is. 6. (6)
is a finite direct sum of fields if and only if is. 7. (7)
is local artinian if and only if is.
The notion of detectability was introduced in [25, Definition 3.1] to deal with the ZCP for noncommutative algebras. The notion of retractability is due to Abhyankar, Eakin, and Heinzer [1, p. 311], called invariance at the time, and was also meant to handle the ZCP. It was later called retractability by Lezama, Wang, and Zhang in [25, Definition 2.1]. We adapt both of these for Poisson algebras here.
Definition 3.5**.**
Let be a Poisson algebra with Poisson center .
- (1)
We say that is strongly Poisson detectable (resp. strongly -detectable) if, for any Poisson algebra and any integer , a Poisson algebra isomorphism implies that (resp. ), 2. (2)
We say that is strongly Poisson retractable (resp. strongly -retractable) if, for any Poisson algebra and integer , any Poisson algebra isomorphism implies that (resp. ).
If either holds only when , then we say is simply Poisson detectable (resp. -detectable) or Poisson retractable (resp. -retractable).
The condition is equivalent to for all .
The following observation is clear.
Lemma 3.6**.**
Let be a Poisson algebra.
- (1)
If is (strongly) Poisson retractable, then is (strongly) Poisson detectable, (strongly) -retractable, (strongly) -detectable, and (strongly) Poisson cancellative. 2. (2)
If is (strongly) -retractable, then is (strongly) -detectable. 3. (3)
If is (strongly) -detectable, then is (strongly) Poisson detectable. 4. (4)
If is (strongly) retractable as an algebra, then is (strongly) -retractable. 5. (5)
If is (strongly) detectable as an algebra, then is (strongly) -detectable.
Proof.
Here we only show (3). All of the remaining cases are verified easily. Suppose that is a Poisson isomorphism for some Poisson algebra and some . Then induces an isomorphism between their Poisson centers:
[TABLE]
Thus because is (strongly) -detectable. Hence is (strongly) Poisson detectable. ∎
Lemma 3.7**.**
The following properties of Poisson algebras are preserved under finite direct sums:
- (1)
being (strongly) Poisson cancellative, 2. (2)
being (strongly) Poisson retractable, 3. (3)
being (strongly) -retractable, 4. (4)
being (strongly) Poisson dectectable, and 5. (5)
being (strongly) -detectable.
Proof.
We only prove (1) and (2)-(5) can be proved similarly. Let and be Poisson algebras that are (strongly) Poisson cancellative and let be an isomorphism for some Poisson algebra and some . Let be the two orthogonal idempotents corresponding to the decomposition . It is easy to check that and are two orthogonal idempotents of , which are indeed orthogonal idempotents of . Write the corresponding decomposition of as . So, induces two isomorphisms and . Since and are (strongly) Poisson cancellative, we have , , and . Thus, is (strongly) Poisson cancellative. ∎
4. Graded versus ungraded
In this section, we deal with the PZCP related to connected graded Poisson algebras following the ideas in [4]. To do this, we study the isomorphism problem for graded Poisson algebras. There has been significant progress in studying this problem in a variety of noncommutative situations [3, 4, 10, 11, 15, 24]. By [4, Theorem 1], an (ungraded) isomorphism between two connected graded algebras finitely generated in degree one implies the existence of a graded isomorphism. We prove a corresponding result in the Poisson setting.
The following lemma is extracted from [4, Lemma 1.1]. We keep its proof for the sake of completeness.
Lemma 4.1**.**
Let and be two connected graded algebras that are generated in degree one. Suppose . If is an isomorphism as ungraded algebras, then .
Proof.
For any connected graded algebra with any codimension 1 ideal , the tangent dimension of is said to be . It is easy to show that the tangent dimension of the augmentation ideal is the upper bound for that of any codimension 1 ideal of . Now since is also generated by , any codimension 1 ideal of has tangent dimension at most . The isomorphism yields a bijection between their codimension 1 ideals sharing the same tangent dimensions. This implies that . ∎
The following result uses the proof of the isomorphism lemma [4, Theorem 1] for noncommutative algebras.
Theorem 4.2**.**
Let and be two connected graded Poisson algebras finitely generated in degree one. If as ungraded Poisson algebras, then as graded Poisson algebras.
Proof.
Let be such an ungraded isomorphism. From Lemma 4.1, . Say and . Note that is a codimension 1 Poisson ideal in . By changing bases, without loss of generality, we can write
[TABLE]
where and . Write and . If , then . After taking the associated graded Poisson algebras, we have as graded Poisson algebras.
Now we assume . We first show the following two claims together by induction on the degree .
Claim I: If is a homogeneous relation of degree in , then in .
Claim II: .
Claims I and II are trivial for . Suppose they hold for . Now let be a homogeneous relation of degree in . Write
[TABLE]
where is a homogeneous term in of degree . Let be the largest integer such that in . If , then . By equation (3), the lowest degree term of
[TABLE]
is . So, in . If , write with and
[TABLE]
whose lowest degree term is . Since , we get in but in by the choice of . By the induction hypotheses, Claims I and II in degree imply that the map for all induces an isomorphism between and as vector spaces. So, in implies that in as well. This contradicts our choice of . Hence this case cannot happen and Claim I is true in degree . Claim II in degree follows from Claim I since the map sends all relations of degree to [math]. Then . By symmetry, the inverse isomorphism implies that , so . In conclusion, if a Poisson algebra isomorphism satisfies Equation (3), then the modified map via for is a well-defined algebra isomorphism.
It remains to show that for all homogeneous elements . We will proceed by induction on the total degree . If , it is trivial. If , without loss of generality, we can assume . Because the Poisson bracket is homogeneous, we can write
[TABLE]
for some and . Applying the Poisson isomorphism in (3) to the above equation, we get
[TABLE]
Since the left hand side is of degree , the degree 0 part of the right hand side yields that . A similar argument suggests that we can assume and . Then the degree 1 part of the right hand side gives that . So we have all and . Then by comparing the degree 2 parts in both sides, we get
[TABLE]
Finally, suppose that . We can write with . Note that is an algebra map and by induction hypothesis we have
[TABLE]
This completes our proof. ∎
The following corollaries are immediate consequences of the isomorphism lemma for connected graded Poisson algebras.
Corollary 4.3**.**
Suppose that a Poisson algebra has two graded Poisson algebra decompositions such that
[TABLE]
where
- (1)
, 2. (2)
is generated by (respectively by ) as an algebra, and 3. (3)
either or is finite dimensional over .
Then there is a Poisson automorphism such that for all integers .
We will use (resp. ) to denote the group of all (graded) Poisson automorphisms of a (graded) Poisson algebra .
Corollary 4.4**.**
Retain the hypotheses of Corollary 4.3. If , then for all integers .
Now we can state the main result regarding the PZCP in the connected graded case with a restriction on the Poisson center being generated in degree at least . It can be viewed as a Poisson version of [4, Theorem 9].
Theorem 4.5**.**
Let and be two connected graded Poisson algebras finitely generated in degree one. Suppose either or is generated in degree at least . If as ungraded Poisson algebras, then as connected graded Poisson algebras.
Proof.
If we set for all , both and are connected graded Poisson algebras that are finitely generated in degree one. Since , by Theorem 4.2, there is a graded Poisson algebra isomorphism . Without loss of generality, we can take which implies . The ’s are in the Poisson center of . Therefore for all . By a dimension argument, . Modulo and we obtain an induced connected graded Poisson algebra isomorphism . ∎
As an application, we can solve the isomorphism problem between skew quadratic Poisson algebras. It can be viewed as the semiclassical limit version of the isomorphism problem for skew polynomial algebras, which was studied in [4].
Theorem 4.6**.**
Suppose is some skew-symmetric matrix with for all . Let be the quotient of a skew quadratic Poisson algebra where and is a graded Poisson ideal in . Let be another graded Poisson algebra where is a graded Poisson ideal in . If is isomorphic to as ungraded Poisson algebras, then and there is a permutation such that for all . Furthermore, after an elementary change of generators in , .
Proof.
First of all, we show under the restrictions on that for any , is a Poisson ideal if and only if for some scalar . One direction is clear. Conversely, let where the coefficients for at least two variables are not zero. We need to show that is not a Poisson ideal. Without loss of generality, we can take . Suppose it is not true. Then for all we can write
[TABLE]
for some scalars . Taking the above identity in the quotient Poisson algebra we get . Since the relation ideal of is contained in and in , then in and for all . Back in , and
[TABLE]
The same argument as before shows that for all . Let . Then for all . In particular, , which contradicts the assumptions on .
Next, by Theorem 4.2, is isomorphic to as graded Poisson algebras. In particular, . Let be such a graded Poisson isomorphism. Since the are Poisson ideals in generated by one degree one element, by the above discussion, for some and some permutation . Up to an elementary change of basis of , sends to for all . The result follows. ∎
5. Artinian center and detectability
In this section, we prove that if a Poisson algebra has artinian Poisson center, then is strongly Poisson cancellative. As an application, we will show Poisson domains of Krull dimension two with nontrivial Poisson bracket are universally Poisson cancellative.
Lemma 5.1**.**
Let be a Poisson algebra over and let be any field extension.
- (1)
If is (strongly) Poisson detectable, then so is . 2. (2)
If is (strongly) -detectable, then so is .
Proof.
(1) Let be an isomorphism for some Poisson algebra over and some . Write and . Then induces an isomorphism . Since is (strongly) Poisson detectable, we have
[TABLE]
Because is flat, we get and is (strongly) Poisson detectable.
(2) can be proved similarly by noting that . ∎
A Poisson ideal that is also a prime ideal is called Poisson prime. Recall that the nilradical of a commutative algebra is the intersection of all its prime ideals which consists of all nilpotent elements. In particular, for any noetherian Poisson algebra , [13, Lemma 1.1(d)] shows that any prime ideal of contains a Poisson prime. This implies that the nilradical of is the intersection of all Poisson prime ideals, which turns out to be a Poisson ideal. We denote the reduced Poisson algebra of by .
Lemma 5.2**.**
Let be a noetherian Poisson algebra. If is (strongly) Poisson detectable, so is .
Proof.
Suppose there is an isomorphism for some Poisson algebra and some . Since and similarly for , then induces an isomorphism . As is (strongly) Poisson detectable, then . In another way, we have
[TABLE]
Repeating this we get
[TABLE]
By Lemma 3.4(2), is noetherian and its nilradical is nilpotent. Therefore, we get and hence is (strongly) Poisson detectable. ∎
Theorem 5.3**.**
Let be a noetherian Poisson algebra.
- (1)
If or is (strongly) Poisson detectable, then is (strongly) Poisson cancellative. 2. (2)
If is (strongly) -detectable, then is (strongly) Poisson cancellative. 3. (3)
If is (strongly) -retractable, then is (strongly) Poisson cancellative.
Proof.
(1) By Lemma 5.2, is always (strongly) Poisson detectable. Let be an isomorphism for some Poisson algebra and some . Since sends to , there is a natural map
[TABLE]
By the fact that is (strongly) detectable, . Then is a surjective endomorphism of . By Lemma 3.4(2), is noetherian and so is . Since noetherian rings are Hopfian, is an automorphism by [19, Proposition(i)]. Therefore
[TABLE]
So, is (strongly) Poisson cancellative.
(2) is from (1) and Lemma 3.6(3).
(3) is from (3) and Lemma 3.6(2). ∎
Corollary 5.4**.**
Let be a noetherian Poisson algebra. If or has artinian Poisson center, then is strongly Poisson cancellative. As a consequence, any artinian Poisson algebra is strongly Poisson cancellative.
Proof.
Suppose has artinian Poisson center with nilradical . In the artinian case, is just the Jacobson radical of . By a possible base field extension, we can assume that is finite direct sum of . By Lemma 5.1 and Lemma 3.7(4), is (strongly) detectable. Hence is (strongly) detectable by Lemma 5.2. Then by Lemma 3.6(5) is (strongly) -detectable. Thus is (strongly) Poisson cancellative by Theorem 5.3(3). Similarly, if has artinian Poisson center, we can prove that is strongly Poisson detectable. So the result follows by Theorem 5.3(1).
The same argument applies to any artinian Poisson algebra. ∎
When the Poisson center of a Poisson algebra is trivial, we can push our result a little bit further in terms of being universally Poisson cancellative. This result and its proof are Poisson analogues of [5, Proposition 1.3].
Theorem 5.5**.**
Let be a Poisson algebra with trivial Poisson center. Then is universally Poisson cancellative.
Proof.
Let be an affine commutative domain equipped with a trivial Poisson bracket. Assume for some (Poisson) ideal and suppose is a Poisson algebra isomorphism for some Poisson algebra . Because , then and . Since restricts to an isomorphism of the Poisson centers, we have . Thus, is a domain with . It follows that is a field and since is a Poisson ideal such that , then and so . Then the induced isomorphism implies that . So . ∎
Corollary 5.6**.**
Let and let be an affine Poisson domain of Krull dimension two. If has nontrivial Poisson bracket, then is strongly Poisson cancellative. Moreover, if is algebraically closed, then is universally Poisson cancellative.
Proof.
We first show that the Poisson bracket of any affine Poisson integral domain of Krull dimension at most one has to be trivial. Take to be the quotient field of , which is a Poisson field after extending the Poisson bracket of to . Therefore, has transcendence degree at most one by [31, Chapter 5, §14]. We treat the case when has transcendence degree one. The transcendence degree zero case can be proved similarly. Suppose for some . Consider the minimal polynomial of over as for some and . Applying the derivation to the above minimal polynomial and noticing that , we get
[TABLE]
Since , , contradicting the minimality of . So, and . It remains to show that . If for some and we still apply the derivation to the minimal polynomial of over . Since , we get a similar contradiction of the minimal degree as before. So, has trivial Poisson bracket and hence has trivial Poisson bracket.
Now let . Denote by the localization of at and by the quotient field of . Since is an integral domain, then . As a consequence, [39, Corollary 2]. As is affine over , one sees that is affine over . Hence and by [22, Theorem 4.5]. By previous discussion, one sees that since it has nontrivial Poisson bracket. Hence, . So is an algebraic field extension of and is artinian. Hence is strongly Poisson cancellative by Corollary 5.4(1).
In particular, if is algebraically closed the above argument shows that . We now can apply Theorem 5.5 to conclude the proof. ∎
Example 5.7**.**
As applications of Corollary 5.4 and Theorem 5.5, we have the following examples.
- (1)
Let . The th Poisson symplectic algebra is with the Poisson bracket
[TABLE]
A straightforward check shows so is universally Poisson cancellative. 2. (2)
Let and . Let be the quadratic Poisson polynomial algebra with the Poisson bracket given by for all . If , then and is universally Poisson cancellative. 3. (3)
Let and be a finite-dimensional restricted Lie algebra over . Write and . Then is a finite-dimensional Poisson algebra with the Kirillov-Kostant Poisson bracket such that for all . So is strongly Poisson cancellative.
6. The Makar-Limanov invariant and retractability
In order to study the retractable property of Poisson algebras, we introduce an analogue of the Makar-Limanov invariant related to (higher) locally nilpotent Poisson derivations.
Definition 6.1**.**
Let be a Poisson algebra with Poisson center , let , and let be either blank or .
- (1)
For a higher Poisson derivation , the kernel of is defined to be
[TABLE] 2. (2)
The set of all (higher) locally nilpotent Poisson derivations of with respect to is denoted
[TABLE] 3. (3)
The Poisson Makar-Limanov invariant of is defined to be
[TABLE] 4. (4)
We say is -rigid if , or equivalently, . Moreover, we say is strongly -rigid if , for all . 5. (5)
We say is -rigid if . Moreover, we say is strongly -rigid if for all .
Remark 6.2**.**
- (1)
Throughout, will be either blank or . When is blank, we will further assume that . Note that there is an embedding from to given by when . 2. (2)
Note that for any . Hence when we will simply omit it and write , , (strongly) -rigid, or (strongly) -rigid. 3. (3)
When is a commutative algebra with trivial Poisson bracket, then and we simply write , , or (strongly) -rigid instead of , , (strongly) or, -rigid, respectively.
Example 6.3**.**
Let and let . Clearly, . Now consider the Poisson bracket on given by for some . One can check that . Hence, . Furthermore, one can show that when no nontrivial locally nilpotent derivation of is a Poisson derivation. It follows that and so is -rigid.
Lemma 6.4**.**
Let be a higher Poisson derivation of a Poisson algebra .
- (1)
Suppose is locally nilpotent. For every there is a Poisson algebra automorphism of , , defined by . 2. (2)
If is iterative and if for any there exists such that for all , then defined by and is a Poisson algebra automorphism. As a a consequence, is locally nilpotent. 3. (3)
Let be a Poisson -algebra automorphism. If for all , then for some . 4. (4)
If for some , then .
Proof.
(1) By definition, is a Poisson algebra automorphism of and . Note that is a Poisson ideal of . Then it yields a Poisson algebra automorphism of . One sees that the induced automorphism is exactly .
(2) First we show that is an algebra automorphism with inverse given by . Since is -linear, it suffices to show that, for all ,
[TABLE]
where all interchanging of summations can be justified by the fact that the sums are actually finite. Moreover, we have
[TABLE]
It remains to show that is a Poisson algebra homomorphism, which follows from, for any ,
[TABLE]
(3) Write for all . In a similar way to the proof of (2), we have that is in .
(4) is clear from the definitions of and . ∎
Lemma 6.5**.**
Let be a Poisson algebra, and . Then for any .
Proof.
First suppose that is blank. That is a -derivation of such that is clear. We claim it is a Poisson derivation. By induction, it suffices to show that is a Poisson derivation of .
Let and be in . Then
[TABLE]
Now suppose . For each , define the series of -linear operators , where
[TABLE]
We show that belongs to for all . Again, it suffices to show this for . Set for , where for and if . Then is clearly a higher derivation of and
[TABLE]
Thus, is a higher Poisson derivation of . Moreover, one can easily check that is iterative and for all . Hence by Lemma 6.4(2) belongs to .
Finally, in all cases one can check directly . ∎
Lemma 6.6**.**
Let be an -graded affine commutative domain. If is an affine subalgebra of containing such that , then .
Proof.
This can be proved using a similar argument as in [5, Lemma 3.2] after replacing with . ∎
Lemma 6.7**.**
Let be a Poisson algebra.
- (1)
Suppose is an affine domain. If , then is Poisson retractable. 2. (2)
Suppose is an affine domain. If is strongly -rigid, then is strongly Poisson retractable. 3. (3)
Suppose is an affine domain. If , then is -retractable. 4. (4)
Suppose is an affine domain. If is strongly -rigid, then is strongly -retractable.
Proof.
(2) Let be an isomorphism for some Poisson algebra and some . By Lemma 3.4(3), . For any locally nilpotent (higher) Poisson derivation of , is again a locally nilpotent (higher) Poisson derivation of . Reversing this argument shows that induces an isomorphism
[TABLE]
The last inclusion is due to Lemma 6.5 since has finite Krull dimension. It follows that . Let with , , and . Then Lemma 6.6 implies that . Thus, and is strongly Poisson retractable. The proof of (1) is analogous.
(4) Let be an isomorphism for some Poisson algebra and some . Since preserves the Poisson center, it induces an isomorphism
[TABLE]
Note that . Because is strongly -rigid, we get . Hence . As in (2), we get and is strongly -retractable. The proof of (3) is analogous. ∎
Our next main result shows that -rigidity implies Poisson cancellation, which suggests that the Poisson Makar-Limanov invariant is a useful invariant in the PZCP.
Theorem 6.8**.**
Let be a Poisson algebra.
- (1)
Suppose is an affine domain. If , then is Poisson cancellative. 2. (2)
Suppose is an affine domain. If is strongly -rigid, then is strongly Poisson cancellative. 3. (3)
Suppose is noetherian and is an affine domain. If , then is Poisson cancellative. 4. (4)
Suppose is noetherian and is an affine domain. If is strongly -rigid, then is strongly Poisson cancellative.
Proof.
(1)-(2) are from Lemma 6.7(1)-(2) and Lemma 3.6(1).
(3)-(4) are from Lemma 6.7(3)-(4) and Theorem 5.3(3). ∎
Our last result in this section is a Poisson analogue of [5, Theorem 3.6]. We adapt the arguments in [5, Lemma 3.4 and Lemma 3.5] and make the necessary changes for Poisson algebras.
Lemma 6.9**.**
Let and let be an affine Poisson domain. Denote by the fractional quotient of . Suppose that is endowed with a nonzero locally nilpotent Poisson derivation . Then the following hold.
- (1)
is embedded in the Poisson-Ore extension and is embedded in , where and is a Poisson derivation of . 2. (2)
. 3. (3)
The Poisson derivation can be extended to a locally nilpotent Poisson derivation of by declaring that and .
Proof.
(1) By [35, Lemma 3.2], extends uniquely to a Poisson derivation of . Let denote the kernel of this extension. Because is a Poisson derivation we have
[TABLE]
Hence, is a Poisson subalgebra of .
By hypothesis is nonzero and locally nilpotent and so there exists such that . Moreover, we may choose such that . Thus, for all ,
[TABLE]
In particular, induces a derivation of . By the Jacobi identity,
[TABLE]
Hence, induces a Poisson derivation of .
Let . Denote by the subalgebra of generated by and . Since , then is a Poisson subalgebra of . To prove , it suffices to show that . Let and let be the smallest integer such that . When , and the claim holds. We proceed inductively. Assume the result holds for all . Take and write with and . But so . Now and so .
As , then . Thus, embeds in the Poisson subalgebra . Since for all , then is isomorphic to a homomorphic image of the Poisson-Ore extension . Let denote the kernel of this image and choose a monic element of minimal -degree, say . Applying gives
[TABLE]
This contradicts the minimality of unless . Thus, embeds in . Finally, the remaining of the statements are easy to check by our construction. ∎
Lemma 6.10**.**
Let be a Poisson algebra and let be a Poisson derivation of the Poisson algebra . Suppose for some fixed . For , write for . Then the map defined by defines a Poisson derivation of .
Proof.
It is clear that is -linear because is and it is easy to check that is a derivation of . We show that is a Poisson derivation. Let . We use the notation to denote the coefficient of in . Then
[TABLE]
∎
Lemma 6.11**.**
Let and let be a Poisson algebra with some .
- (1)
Suppose is an affine domain. If is -rigid, then . 2. (2)
Suppose is an affine domain. If is -rigid, then .
Proof.
(1) By Lemma 6.5, it suffices to show . Suppose by way of contradiction that there exists a locally nilpotent Poisson derivation of such that and . Let be a generating set of as a -algebra. Then and so there exists minimial such that . When we have and so because is -rigid, a contradiction. Thus, . Write, by Lemma 6.10,
[TABLE]
for some and . We consider three cases depending on the image of under . It will follow from these cases that .
Case 1 .
In this case we have and for all . Thus for every we have
[TABLE]
We get is locally nilpotent as is. Hence by Lemma 6.10, which implies that for is -rigid. This contradicts the minimality of so .
Case 2 for some in .
Since , then we have
[TABLE]
Thus, for all so . Define by for and . By [8, Lemma 4.11], . Moreover, we can check that is a Poisson derivation.
Let and be in . Then
[TABLE]
Thus . We can view as an associated graded Poisson derivation of . Since is locally nilpotent, so is and . We apply Lemma 6.9 to the Poisson algebra with respective to . Denote . There is an embedding for some . Moreover, extends to a locally nilpotent Poisson derivation of by setting and . Under this embedding, we have . It follows that
[TABLE]
Since , , a contradiction to the above equality.
Case 3 for some in and some . It is easy to see
[TABLE]
for all . So, can not be locally nilpotent, which is a contradiction.
Combining all above cases, we see that .
(2) can be proved similarly noting that any Poisson derivation of preserves its Poisson center . ∎
Theorem 6.12**.**
Let and let be a Poisson algebra.
- (1)
Suppose is an affine domain. If is (strongly) -rigid, then is (strongly) Poisson cancellative. 2. (2)
Suppose is noetherian and is an affine domain. If is (strongly) -rigid, then is (strongly) Poisson cancellative.
Proof.
It follows from Theorem 6.8 and Lemma 6.11. ∎
7. Poisson discriminant and effectiveness
For noncommutative algebras that are module-finite over their center, the notion of the discriminant has been introduced to compute their automorphism groups [5, 7, 8]. In this section we study the discriminant for Poisson algebras and its relation with the PZCP and Poisson automorphism groups. It is a matter of fact that a Poisson algebra appearing in characteristic zero usually does not possess a large Poisson center. Therefore, we introduce the Poisson discriminant from a representation-theoretic point of view following Lu, Wu, and Zhang [26, §2].
Let be a Poisson algebra. We denote by the set of units of . By a property we mean a property that is invariant under isomorphism within a class of algebras. Herein we generally assume this is the class of Poisson algebras.
Definition 7.1**.**
Let be a Poisson algebra and let . Let be a property defined for Poisson algebras. We define the following terms for sets/ideals in .
- (1)
(-locus) L_{\mathcal{P}}(A):=\{\mathfrak{m}\in\operatorname{MaxSpec}(\mathcal{Z}_{P}):A/\mathfrak{m}A\text{ has property \mathcal{P}}\}. 2. (2)
(-discriminant set) . 3. (3)
(-discriminant ideal) .
In the case that is a principal ideal, generated by , then is called the -discriminant of , denoted by . Observe that, if is a domain, is unique up to an element of .
Example 7.2**.**
Let be the first Poisson Weyl algebra. It is well-known that is Poisson simple. Now let be the Poisson homogenization of such that with Poisson bracket and . Suppose is algebraically closed of characteristic zero. The Poisson center of is . For each , let denote the corresponding maximal ideal of . Then for all . On the other hand, with the trivial Poisson bracket, whence not Poisson simple. It follows that if is the property of being Poisson simple, then the -locus is , the -discriminant set is , and the -discriminant exists, that is .
Definition 7.3**.**
Let be a class of Poisson -algebras. We say that a property is -stable if for every Poisson algebra in and every , as ideals of . If is a singleton , we say is -stable. On the other hand, if is the collection of all Poisson -algebras with affine Poisson center over , we say is stable.
The following lemma is obvious from Definition 7.1.
Lemma 7.4**.**
Let be a property. If is an isomorphism of Poisson algebras, then preserves the -locus, -discriminant set, -discriminant ideal, and, if it exists, the -discriminant.
Most of our results will require that is a stable property. The next lemma, which generalizes [26, Lemma 6.1], allows us to work with any property when is algebraically closed.
Lemma 7.5**.**
Let be algebraically closed. Then any property is stable.
Proof.
Let be any Poisson algebra over with affine Poisson center . By Hilbert’s Nullstellensatz, there is a natural projection such that as Poisson algebras over for any ideal . It now follows in a similar way to [26, Lemma 6.1] that and so is a stable property. ∎
Our next result is a generalization of Lemma 6.7 using the Poisson discriminant.
Lemma 7.6**.**
Let be a Poisson algebra with affine Poisson center . Let be a stable property and assume that the -discriminant of exists.
- (1)
Suppose is an affine domain. If , then is Poisson retractable. 2. (2)
Suppose is an affine domain. If is strongly -rigid, then is strongly Poisson retractable. 3. (3)
Suppose is a domain. If , then is -retractable. 4. (4)
Suppose is a domain. If is strongly -rigid, then is strongly -retractable.
Proof.
We prove (4). The proofs of (1)-(3) are similar.
Let be an isomorphism for some Poisson algebra and some . Note that the property is stable and the -discriminant ideal is invariant under any Poisson isomorphism by Lemma 7.4. Write as the identity element in . Hence,
[TABLE]
Thus, where is the -discriminant of . Moreover, the above computation implies that , where is the identity element in . However,
[TABLE]
is a domain, so . It follows that . Since is -rigid, we have
[TABLE]
where the last inclusion follows from Lemma 6.5. By Lemma 6.6, and is strongly -retractable. ∎
Now we can generalize our previous Theorem 6.8 by applying the Poisson discriminant.
Theorem 7.7**.**
Let be a Poisson algebra with affine Poisson center . Let be a stable property and assume that the -discriminant of exists.
- (1)
Suppose is an affine domain. If , then is Poisson cancellative. 2. (2)
Suppose is an affine domain. If is strongly -rigid, then is strongly Poisson cancellative. 3. (3)
Suppose is noetherian and is a domain. If , then is Poisson cancellative. 4. (4)
Suppose is noetherian and is a domain. If is strongly -rigid, then is strongly Poisson cancellative.
Proof.
(1)-(2) are from Lemma 7.6(1)-(2) and Lemma 3.6(1).
(3)-(4) are from Lemma 7.6(3)-(4) and Theorem 5.3(3). ∎
Corollary 7.8**.**
Let be algebraically closed of characteristic zero and let be a Poisson algebra with affine Poisson center . Let be any property and assume that the -discriminant of exists.
- (1)
Suppose is an affine domain. If is (strongly) -rigid, then is (strongly) Poisson cancellative. 2. (2)
Suppose is noetherian and is a domain. If is (strongly) -rigid, then is (strongly) Poisson cancellative.
Proof.
Since is algebraically closed, the property is stable by Lemma 7.5. Then the result follows from Theorem 7.7 and Lemma 6.11(1). ∎
Remark 7.9**.**
Let be a Poisson algebra with affine Poisson center , and be a stable property such that the -discriminant of exists. Applying the arguments in Lemma 7.6, we can show that the following hold.
- (1)
If is (strongly) -rigid, then it must be (strongly) -rigid. 2. (2)
If either is (strongly) -rigid or is (strongly) -rigid, then is (strongly) -rigid.
According to [5, §5], effectiveness of the discriminant controls -rigidity and hence solves the ZCP. For Poisson algebras, we will see that effectiveness of the Poisson discriminant plays the same role in the PZCP.
Definition 7.10**.**
Let be an affine (Poisson) commutative domain and suppose that generates as an algebra. An element is called (Poisson) effective if the following conditions hold.
- (1)
There is a (Poisson) -filtration on such that is a domain. With this filtration we define the degree of elements in , denoted by . 2. (2)
For every -filtered (Poisson) commutative algebra with being an -graded (Poisson) domain and for every testing subset satisfying
- (a)
it is linearly independent in the quotient -module , and 2. (b)
for all and for some ,
there is a presentation of of the form in the free algebra , such that either is zero or .
Remark 7.11**.**
- (1)
Our definition of effectiveness borrows from the definition for noncommutative algebras [5, Definition 5.1]. The key difference there is that the testing algebra should be PI. In our case, that is not necessary because we assume that every algebra is commutative. 2. (2)
There is another concept, called “dominating”, see [5, Definition 4.5] or [7, Definition 2.1(2)], which is slightly different from the definition of effectiveness. In this paper, we do not state this concept explicitly. 3. (3)
In many applications of Poisson discriminant we do not need the filtration appearing in the definition of effectiveness to be a Poisson filtration. So we will just use the effectiveness of Poisson discriminant for commutative algebras.
Example 7.12**.**
There are some simple examples of effective elements in polynomial algebras.
- (1)
For the polynomial algebra it is observed in [25, Example 2.8] that any nonzero element is effective. In a similar way it is Poisson effective. 2. (2)
Let be a Poisson algebra. A monomial in is said to have degree component-wise less than if for all and for some . We write if is a linear combination of monomials with degree compoent-wise less than . If satisfies and , then it follows from [25, Lemma 2.2(1)] that is effective in .
The next result is parallel to the corresponding result for discriminants of noncommutative (associative) algebras. We omit the proof and refer the reader to [5, 7, 8] for details.
Theorem 7.13**.**
Let be an affine Poisson algebra with generating subspace , and the associated graded algebra is a connected graded (Poisson) domain. Let be a stable property and assume that the -discriminant of exists. If is (Poisson) effective in , then is an algebraic group that fits into an exact sequence
[TABLE]
for some finite group .
Example 7.14**.**
Let be the Poisson algebra whose Poisson bracket is of Jacobian form given by the potential such that
[TABLE]
Since is homogeneous with isolated singularity at the origin, by [36, Proposition 4.2]. Let be the property such that the symplectic foliation appearing in the maximal spectrum of the corresponding Poisson algebra has no 0 dimensional skeletons. It is clear that has property for any if and only if it has no maximal ideal that is also a Poisson ideal. One checks that the zero ideal is the only maximal ideal of that is also a Poisson ideal. Hence the -discriminant of is given by . Note is neither effective nor Poisson effective in since contains Nagata’s wild automorphism [32] defined by
[TABLE]
On the contrary, is both effective and Poisson effective in the Poisson center , which implies that is strongly Poisson cancellative (see Theorem 7.16(2) later).
Lemma 7.15**.**
Let be a Poisson algebra with affine Poisson center . Let be a stable property and assume that the -discriminant of exists.
- (1)
Suppose is an affine domain. If is effective (respectively, Poisson effective) in , then is strongly -rigid. 2. (2)
Suppose is a domain. If is effective in , then is strongly -rigid.
Proof.
We will only prove (1). The proof of (2) is similar. Suppose is not strongly -rigid. Let be a minimal generating space of . Then there is for some such that for some . By Lemma 6.4(1,4), we have satisfying and
[TABLE]
Since is (Poisson) effective in , it implies that has a (Poisson) -filtration which naturally induces a filtration on its extension by assigning arbitrary degrees on and . It is easy to see that once , the -filtered (Poisson) algebra has a set where and in particular . But by (Poisson) effectiveness of , we must have
[TABLE]
a contradiction. Hence is strongly -rigid. ∎
Theorem 7.16**.**
Let be a Poisson algebra with affine Poisson center . Let be a stable property and assume that the -discriminant of exists.
- (1)
Suppose is an affine domain. If is effective (respectively, Poisson effective) in , then is strongly Poisson cancellative. 2. (2)
Suppose is noetherian and is a domain. If is effective in , then is strongly Poisson cancellative.
Proof.
The result follows from Lemma 7.15 and Theorem 7.7(2,4). ∎
Example 7.17**.**
Let be a polynomial algebra over an algebraically closed field of characteristic zero. Suppose has a Jacobian bracket given by some homogeneous potential such that , , . Suppose has isolated singularities. That is, suppose is a finite-dimensional -algebra, or equivalently, is a regular sequence of length 3 in . By [36, Proposition 4.2] we have . Let be the property of not having finite-dimensional simple Poisson modules. Then for any , has property if and only if no maximal ideal of is a Poisson ideal by [21, Lemma 3.1]. Moreover, any maximal ideal for any is a Poison ideal if and only if , which is a finite set of points in . Hence the -discriminant of is given by
[TABLE]
is a nonzero element in , which is effective in by Example 7.12(1). So is always strongly Poisson cancellative according to Theorem 7.16(2).
Example 7.18**.**
Let be a three-dimensional Sklyanin algebra over an algebraically closed field of characteristic zero that is module-finite over its center . Let be the associated geometric data of . It is well-known that has a central regular element of degree three and , the twisted homogeneous coordinate ring of . When is PI, the automorphism of the elliptic curve has finite order and is a GK-dimension two domain that is module-finite over its center . It is proved in [43, Proposition 5.5(1)] that there exists a unique nonzero Poisson structure (up to scalars) on if we assume to be in the corresponding Poisson center. We can show that is strongly Poisson cancellative in this case. We observe that . Since is a Poisson domain of Krull dimension two, has trivial Poisson center by Corollary 5.6. Then, by passing to , an easy induction on degree of homogeneous elements in yields the result. Now let be the property of having no or at least two zero dimensional symplectic core skeletons in the symplectic core stratification of the corresponding maximal spectrum. Then [43, Theorem 1.3] shows that the -discriminant of is exactly given by . Hence is strongly Poisson cancellative by Theorem 7.16(2) and Example 7.12(1).
8. Some remarks and questions
In this last section, we provide some remarks and questions for future projects. First of all, we summarize the related concepts that are introduced in our paper to the PZCP.
[TABLE]
When is a Poisson algebra with trivial Poisson bracket, PZCP for becomes the ordinary ZCP for . This yields our first question.
Question 8.1**.**
Let be a Poisson algebra. If is (strongly, universally) cancellative as a commutative algebra, is always (strongly, universally) cancellative in the sense of Poisson algebras?
The inverse implication of the above question does not hold. In Example 3.3(1), the coordinate ring of the real sphere is not cancellative. But we can endow it with a Jacobian bracket given by the potential such that , , and . One can show the resulting Poisson structure on the coordinate ring yields the trivial Poisson center. Hence it is universally Poisson cancellative by Theorem 5.5. So this yields our second question.
Question 8.2**.**
Let be a commutative algebra that is not cancellative. Can we always endow with a Poisson bracket that makes it Poisson cancellative?
Our Theorem 4.2 is a Poisson analogue of the isomorphsm lemma for connected graded algebras generated in degree one [4, Theorem 1]. Note that the original isomorphism lemma for connected graded algebras has been generalized for graded path algebras [12]. We expect that our result can be extended to graded path Poisson algebras as well.
Question 8.3**.**
Let and be two -graded Poisson algebras that are finitely generated in degree 0 and 1. If as ungraded Poisson algebras, does as graded Poisson algebras?
In Example 7.17, we showed that any polynomial algebra of three variables with Jacobian bracket is strongly Poisson cancellative if the potential related to the Jacobian form is homogeneous with isolated singularities. We would like to know if the cancellation property holds more generally.
Question 8.4**.**
Is any polynomial Poisson algebra in three variables with nontrivial Jacobian bracket Poisson cancellative?
Moreover, every unimodular Poisson algebra on has Jacobian bracket [37, Theorem 5]. For noncommutative algebras, the ZCP was asked for Artin-Schelter regular algebras in [5, Question 0.2]. For a connected graded algebra, the (skew) Calabi-Yau property is equivalent to Artin-Schelter regularity [38, Lemma 1.2]. Note that unimodularity of Poisson algebras is an analogue of Calabi-Yau property for noncommutative algebras. We refer the interested reader to [29] for the notion of unimodularity and its connection to Calabi-Yau algebras. Therefore, we are interested in the PZCP for unimodular Poisson algebras.
Question 8.5**.**
Let be a unimodular complex Poisson polynomial algebra with nontrivial Poisson bracket. When is Poisson cancellative?
In Example 7.18, we show that for a three-dimensional Sklyanin algebra that is module-finite over its center, any nontrivial Poisson bracket on the center makes it strongly Poisson cancellative provided the canonical central regular element is in the Poisson center. A similar phenomena occurs with any four-dimensional Sklyanin algebra that is module-finite over its center . It is proved in [42, Proposition 7.3] that the Poisson structure on is uniquely determined as long as the two canonical central regular elements of are in the Poisson center of .
Question 8.6**.**
Let be a four-dimensional Sklyanin algebra that is module-finite over its center . Suppose has a nontrivial Poisson structure where the two canonical central regular elements of are in the Poisson center of . Is strongly Poisson cancellative?
For any Poisson algebra , there is a notion of Poisson universal enveloping algebra , whose representation category is Morita equivalent to the category of Poisson modules over . Now we recall the definition of from [34]. Let be a Poisson algebra and let and be two copies of the vector space endowed with two -linear isomorphisms and for any . Then the Poisson universal enveloping algebra is an associative algebra over , with an identity 1, generated by and with relations, for any ,
[TABLE]
Question 8.7**.**
Let be any affine Poisson algebra, and be its Poisson universal enveloping algebra. What is the relationship between being cancellative and being Poisson cancellative?
In practice many Poisson structures can be derived from the process of semiclassical limits, for instance see [14], which we will recall now. Suppose is a torsionfree -algebra such that is a commutative -algebra. Denote the specialization map . The algebra equipped with the Poisson bracket:
[TABLE]
is called the semiclassical limit of the family of (noncommutative) algebras , where .
Question 8.8**.**
Is the Poisson cancellation property of related to the cancellation property of the ?
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] Shreeram S. Abhyankar, William Heinzer, and Paul Eakin, On the uniqueness of the coefficient ring in a polynomial ring , J. Algebra 23 (1972), 310–342. MR 0306173
- 2[2] Pascal Kossivi Adjamagbo and Arno van den Essen, A proof of the equivalence of the Dixmier, Jacobian and Poisson conjectures , Acta Math. Vietnam. 32 (2007), no. 2-3, 205–214. MR 2368008
- 3[3] V. V. Bavula and D. A. Jordan, Isomorphism problems and groups of automorphisms for generalized Weyl algebras , Trans. Amer. Math. Soc. 353 (2001), no. 2, 769–794. MR 1804517
- 4[4] Jason Bell and James J. Zhang, An isomorphism lemma for graded rings , Proc. Amer. Math. Soc. 145 (2017), no. 3, 989–994. MR 3589298
- 5[5] by same author, Zariski cancellation problem for noncommutative algebras , Selecta Math. (N.S.) 23 (2017), no. 3, 1709–1737. MR 3663593
- 6[6] K. A. Brown and M. T. Yakimov, Azumaya loci and discriminant ideals of PI algebras , Adv. Math. 340 (2018), 1219–1255. MR 3886192
- 7[7] S. Ceken, J. H. Palmieri, Y.-H. Wang, and J.J. Zhang, The discriminant controls automorphism groups of noncommutative algebras , Adv. Math. 269 (2015), 551–584. MR 3281142
- 8[8] by same author, The discriminant criterion and automorphism groups of quantized algebras , Adv. Math. 286 (2016), 754–801. MR 3415697
