
TL;DR
This paper extends the proof of the Nowicki conjecture, describing the algebra of constants for certain derivations, providing explicit generators, relations, and a Gr"obner basis, thus deepening understanding of invariant rings in polynomial algebras.
Contribution
It generalizes previous results by explicitly presenting the algebra of invariants for derivations with polynomial images, including generators, relations, and Gr"obner basis structure.
Findings
Explicit presentation of the algebra of invariants as a quotient of a polynomial ring.
Identification of a reduced Gr"obner basis for the defining ideal.
Extension of the Nowicki conjecture to more general polynomial derivations.
Abstract
Let be an integral domain over a field of characteristic 0. The derivation of is elementary if and , . Then the elements , , belong to the algebra of constants of and it is a natural question whether the -algebra is generated by all . In this paper we consider the special case of . If , , this is the Nowicki conjecture from 1994 which was confirmed in several papers based on different methods. The case , , , was handled by Khoury in the first proof of the Nowicki conjecture given by him in 2004. As a consequence of the proof of Kuroda in 2009 if , for any nonconstantβ¦
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.
Generalized Nowicki Conjecture
Vesselin Drensky
Institute of Mathematics and Informatics, Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria
Abstract.
Let be an integral domain over a field of characteristic 0. The derivation of is elementary if and , . Then the elements , , belong to the algebra of constants of and it is a natural question whether the -algebra is generated by all . In this paper we consider the special case of . If , , this is the Nowicki conjecture from 1994 which was confirmed in several papers based on different methods. The case , , , was handled by Khoury in the first proof of the Nowicki conjecture given by him in 2004. As a consequence of the proof of Kuroda in 2009 if , for any nonconstant polynomials , , then is generated by and . In the present paper we have found a presentation of the algebra
[TABLE]
[TABLE]
and a basis of as a vector space. As a corollary we have shown that the defining relations form the reduced GrΓΆbner basis of the ideal which they generate with respect to a specific admissible order. This is an analogue of the result of Makar-Limanov and the author in their proof of the Nowicki conjecture in 2009.
Key words and phrases:
algebras of constants; elementary derivations; Nowicki conjecture; GrΓΆbner bases; presentation of algebra.
2010 Mathematics Subject Classification:
13N15; 13P10; 13E15.
1. Introduction
In the present paper we consider only commutative algebras over a field of characteristic 0. A linear operator of an algebra is a derivation if it satisfies the Leibniz rule
[TABLE]
The kernel of is the algebra of constants of . Let be an integral domain over and let . The derivation of is elementary if and , . Then the determinants
[TABLE]
belong to and it is a natural question whether the -algebra is generated by the elements (1). In the sequel we assume that (and ). In the special case
[TABLE]
the finite generation of follows from a more general result of WeitzenbΓΆck [12] in 1932. In 1994 Nowicki [10] conjectured that for from (2) the algebra is generated by and
[TABLE]
This was confirmed in several papers based on different methods, see, e.g., [9] and [2] for details.
In the first proof of the Nowicki conjecture given in his Ph.D. thesis Khoury [6, 8] made one more step and established a result which gives an answer to a generalization of the Nowicki conjceture. If , , are positive integers, and the derivation of is defined by
[TABLE]
then the algebra of constants of is generated again by and
[TABLE]
The most general result in this direction belongs to Kuroda [9].
Theorem 1.1**.**
Let be an integral domain over and let be an elementary derivation of such that , , are algebraically independent over . If is flat over , then the -algebra is generated by , .
It seems that it is difficult to find further generalizations. Khoury [7] showed that the algebra of constants of the elementary derivation of , , for is finitely generated but cannot be generated by expressions which are linear in . Also, many of the modern counterexamples to the Fourteenth Hilbert problem are constructed in terms of elementary derivations, see, e.g., the surveys by Freudenburg [5] and Nowicki [11].
It is easy to see that Theorem 1.1 holds for the elementary derivation of , , and
[TABLE]
where is a nonconstant polynomial in the variable , . But there is an essential difference between the Nowicki conjecture and this result. The Nowicki conjecture is equivalent to a statement of classical invariant theory. When the polynomials are not linear, we cannot see how to restate the result in the language of invariant theory. In the present paper we apply the methods developed in our proof with Makar-Limanov [3] of the Nowicki conjecture and find a presentation of the algebra and a basis of as a vector space. As a consequence we show that our defining relations form the reduced GrΓΆbner basis of the ideal which they generate with respect to a specific admissible order.
2. Generators of the algebra of constants
We shall need the following easy lemma. We include the proof for self-containedness of the exposition.
Lemma 2.1**.**
Let , , be nonconstant polynomials. Then is a free -module.
Proof.
Let be the degree of . Every monomial can be written as a linear combination of polynomials , , , and hence as a -module is generated by the monomials
[TABLE]
The leading terms of the products
[TABLE]
with respect to the lexicographic order are equal to and are pairwise different. Hence the polynomials (3) are linearly independent which implies that is a free -module. β
Theorem 2.2**.**
Let , , be nonconstant polynomials in one variable and let be the derivation of defined by
[TABLE]
Then the algebra of constants of is generated by and the determinants from (1)
[TABLE]
Proof.
By [4, Corollary 6.6, p. 165] if a -module is finitely generated then it is flat if and only if it is a summand of a free -module. Hence by Lemma 2.1 is a flat -module. Obviously , , are algebraically independent over and Theorem 1.1 immediately gives that the algebra is generated by and the polynomials (4). β
Remark 2.3**.**
If some of the polynomials in Theorem 2.2 is a constant, then the description of is trivial. Let, for example, . We replace the variables by , where
[TABLE]
Then the definition of becomes
[TABLE]
Since , we obtain that .
3. The main result
In this section we follow our paper with Makar-Limanov [3] and use the methods developed there. Since we shall work with GrΓΆbner bases, we refer, e.g., to the book by Adams and Loustaunau [1] for a background on the topic. We fix the degrees of the polynomials and the set
[TABLE]
where the elements are defined in (4).
Lemma 3.1**.**
*The subsets and of satisfy the relations , where *
[TABLE]
[TABLE]
Proof.
The annihilating of the relations (5) and (6) in can be verified directly. Instead, the expansions of the determinants
[TABLE]
and
[TABLE]
relative to the first two rows and to the first row give, respectively, (5) and (6). β
Now we shall work in the polynomial algebra . By Theorem 2.2 there is a canonical epimorphism
[TABLE]
Since there will be no misunderstanding, we shall use the same symbols and for the generators of the polynomial algebra and their images under which generate the algebra of constants .
Our first goal is to show that the kernel of is generated by the relations and from (5) and (6). We associate with every the open interval . As in [3] we define an ordering of called degreeβinterval lengthβlexicographic order (DILL order). We order the monomials of first by the degree in and in , then by the total length of the intervals associated with the participating variables and finally lexicographically. If
[TABLE]
where , and if , with similar restrictions on , we define if
(i) (we compare the degrees of and in );
(ii) and (we compare the degrees of and in );
(iii) , and
[TABLE]
(we compare the total lengths of the intervals associated with and );
(iv) , ,
[TABLE]
and for the -tuples
[TABLE]
where if , for some (we compare lexicographically and ).
Obviously the DILL-order is admissible, i.e., it is linear, satisfies the descending chain condition, and if for two monomials and , then for all monomials . For we denote by the leading monomial of .
Lemma 3.2**.**
The set of normal words in with respect to DILL order and modulo the relations consists of the monomials
[TABLE]
such that
(i)* If for two different and in (8), then one of the intervals and is contained in the other;*
(ii)* If for some in (8), then .*
As a vector space the algebra is spanned by the images under of the products (8).
Proof.
The leading monomials with respect to the DILL order of and from (5) and (6) are, respectively,
[TABLE]
(i) Let divide the monomial , and let and if , then . If the intervals and have a nontrivial intersection and are not contained in each other, then . Hence the monomial is the leading term of . In this way is not a normal word and this proves (i).
(ii) If the monomial is a normal word and divides with , then is not divisible by a monomial from (9) and hence . This proves (ii).
Let us take some set of polynomials which are in the kernel of from (7). The images in of the normal words in with respect to span the vector space . By Lemma 3.1 the set of relations from (5) and (6) belong to the kernel of . Hence is spanned by the images of the normal words from (8). β
Lemma 3.3**.**
The images under the epimorphism from (7) of the normal words (8) from Lemma 3.2 form a basis of the vector space .
Proof.
Let be a normal word. Then
[TABLE]
We shall follow the proof of [3, Step 3 in the proof of Theorem 5]. We compare the monomials in lexicographically:
[TABLE]
if if the usual lexicographic order. If , then the leading monomials of and are and . Hence and the leading monomial of (10) is
[TABLE]
We shall show that we can recover the normal word from the leading monomial . Hence the monomials are pairwise different and the polynomials are linearly independent. By Lemma 3.2 this implies that form a basis of the vector space .
It is sufficient to compare those which are of the same degree in . As in [3] we use induction on the degree with respect to . Since , if , then and coincides with . Now let . We rewrite (11) in the form
[TABLE]
The variable participates in because participates in from (8) for some and . We choose the maximal such that and . Lemma 3.2 implies that does not belong to the ideal . Hence and the factor of is uniquely determined by the form (12) of . In this way for some normal word and
[TABLE]
By inductive arguments determines and this completes the proof. β
The following theorem is the main result of the paper.
Theorem 3.4**.**
Let be the derivation of defined by
[TABLE]
where are polynomials of positive degree. Then
(i)* The algebra of constants has the presentation*
[TABLE]
where
[TABLE]
and the ideal is generated by and from (5) and (6).
(ii)* The set is a reduced GrΓΆbner basis of the ideal with respect to the DILL order of .*
(iii)* The basis of the vector space consists of the images in of the normal words from Lemma 3.2.*
Proof.
By Lemma 3.1 belongs to the kernel of the epimorphism from (7). A subset of the ideal is its GrΓΆbner basis if and only if the set of normal words forms a basis of the factor algebra . Hence the statements (i), (ii), and (iii) follow immediately from Lemma 3.3. Since the leading terms of the relations and with respect to the DILL order do not divide the monomials participating in the other relations in and , we conclude that the GrΓΆbner basis is reduced. β
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] W.W. Adams, P. Loustaunau, An Introduction to GrΓΆbner Bases, Graduate Studies in Mathematics 3 , American Mathematical Society, Providence, RI, 1994.
- 2[2] V. Drensky, Another proof of the Nowicki conjecture, ar Xiv:1902.08758 [math.AC].
- 3[3] V. Drensky, L. Makar-Limanov, The conjecture of Nowicki on WeitzenbΓΆck derivations of polynomial algebras, J. Algebra Appl. 8 (2009), 41-51. DOI: 10.1142/S 0219498809003217.
- 4[4] D. Eisenbud, Commutative Algebra. With a View Toward Algebraic Geometry, Graduate Texts in Mathematics 150 , Springer-Verlag, Berlin, 1995.
- 5[5] G. Freudenburg, A survey of counterexamples to Hilbertβs fourteenth problem, Serdica Math. J. 27 (2001), No. 3, 171-192.
- 6[6] J. Khoury, Locally Nilpotent Derivations and Their Rings of Constants, Ph.D. Thesis, Univ. Ottawa, 2004.
- 7[7] J. Khoury, A note on elementary derivations, Serdica Math. J. 30 (2004), No. 4, 549-570.
- 8[8] J. Khoury, A Groebner basis approach to solve a conjecture of Nowicki, J. Symbolic Comput. 43 (2008), 908-922. DOI: 10.1016/j.jsc.2008.05.004.
