The characteristic quasi-polynomials of the arrangements of root systems and mid-hyperplane arrangements
Hidehiko Kamiya, Akimichi Takemura, Hiroaki Terao

TL;DR
This paper explores the properties of characteristic quasi-polynomials associated with arrangements of root systems and mid-hyperplanes, providing formulas and results for specific cases and highlighting their mathematical significance.
Contribution
It introduces general properties of characteristic quasi-polynomials and derives explicit formulas for root system arrangements and low-dimensional mid-hyperplane arrangements.
Findings
A formula for the generating function of root system arrangements.
Explicit determination of generating functions for mid-hyperplane arrangements in dimensions less than six.
Insights into the structure and computation of characteristic quasi-polynomials.
Abstract
Let be a positive integer. In our recent paper, we proved that the cardinality of the complement of an integral arrangement, after the modulo reduction, is a quasi-polynomial of , which we call the characteristic quasi-polynomial. In this paper, we study general properties of the characteristic quasi-polynomial as well as discuss two important examples: the arrangements of reflecting hyperplanes arising from irreducible root systems and the mid-hyperplane arrangements. In the root system case, we present a beautiful formula for the generating function of the characteristic quasi-polynomial which has been essentially obtained by Ch. Athanasiadis and by A. Blass and B. Sagan. On the other hand, it is hard to find the generating function of the characteristic quasi-polynomial in the mid-hyperplane arrangement case. We determine them when the dimension is less than six.
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.
Taxonomy
TopicsAdvanced Combinatorial Mathematics · Advanced Mathematical Identities · Mathematical functions and polynomials
The characteristic quasi-polynomials of the arrangements of
root systems and mid-hyperplane arrangements
Hidehiko Kamiya 111 Faculty of Economics, Okayama University
Akimichi Takemura 222 Graduate School of Information Science and Technology, University of Tokyo
Hiroaki Terao 333 Department of Mathematics, Hokkaido University
(September 2007)
Abstract
Let be a positive integer. In [8], we proved that the cardinality of the complement of an integral arrangement, after the modulo reduction, is a quasi-polynomial of , which we call the characteristic quasi-polynomial. In this paper, we study general properties of the characteristic quasi-polynomial as well as discuss two important examples: the arrangements of reflecting hyperplanes arising from irreducible root systems and the mid-hyperplane arrangements. In the root system case, we present a beautiful formula for the generating function of the characteristic quasi-polynomial which has been essentially obtained by Ch. Athanasiadis [2] and by A. Blass and B. Sagan [3]. On the other hand, it is hard to find the generating function of the characteristic quasi-polynomial in the mid-hyperplane arrangement case. We determine them when the dimension is less than six.
Key words: characteristic quasi-polynomial, elementary divisor, hyperplane arrangement, root system, mid-hyperplane arrangement.
1 Introduction
Let be an arbitrary integral matrix without zero columns. For each positive integer , denote and . Consider the set
[TABLE]
and its cardinality . In our recent paper [8], we showed that there exists a monic quasi-polynomial (periodic polynomial) with integral coefficients of degree such that
[TABLE]
Note that the set is the complement of an arrangement of hyperplanes in the following sense: Let be the columns of . Each set
[TABLE]
can be called a “hyperplane” in by a slight abuse of terminology. Then
[TABLE]
For a sufficiently large prime number , is known [2] to be equal to the characteristic polynomial [9, Def. 2.52] of the real arrangement consisting of the following hyperplanes (ignoring possible repetitions):
[TABLE]
It is thus natural to call the quasi-polynomial the characteristic quasi-polynomial of as in [8]. Let us define its **generating function **
[TABLE]
We understand that for and hence the summation is in effect for .
In this paper, we study the characteristic quasi-polynomial or equivalently its generating function In Section 2, we discuss general properties of the characteristic quasi-polynomials and their generating functions. In the subsequent chapters, we deal with two kinds of specific arrangements defined over : the arrangements of reflecting hyperplanes arising from irreducible root systems (Section 3) and the mid-hyperplane arrangements (Section 4). Let be an irreducible root system of rank and . We assume that an integral matrix satisfies
[TABLE]
where is a set of positive roots and is the set of simple roots associated with . In other words, is a coefficient matrix of with respect to the basis . Define the characteristic quasi-polynomial and the generating function for each irreducible root system . Then and depend only upon . In Section 3, we present a beautiful formula for the generating function for every irreducible root system . This formula has been essentially proved by Ch. Athanasiadis in [2] and A. Blass and B. Sagan in [3]. (See [6] also.) In Theorem 3.1, we will state the formula in our language and include a proof following [2, 3, 6] for completeness. In Section 4, we will give a formula for the generating function when is equal to the coefficient matrix for the mid-hyperplane arrangement of dimension less than six.
We are aided by the computer package PARI/GP [10].
2 Results on the characteristic quasi-polynomial of an integral matrix
Let be the characteristic quasi-polynomial of an integral matrix without zero columns. Fix a nonempty and define an matrix consisting of the columns of corresponding to the set . Let be the elementary divisors of numbered so that , where . Write , and define the lcm period of by
[TABLE]
Then it is known ([8, Theorem 2.4]) that the lcm period is a period of .
It is further shown in [8] that the constituents of the quasi-polynomial are the same for all ’s with the same value of . Let be a positive integer which divides , and define a monic polynomial with integral coefficients of degree by
[TABLE]
Put
[TABLE]
Then the following formula was essentially proved in our previous paper [8].
Theorem 2.1**.**
For each with , the polynomial is given by
[TABLE]
where for , we understand that and that .
Proof. Obtained from [8, (10)] and the inclusion-exclusion principle. ∎
Theorem 2.2** ([8] Theorem 2.5).**
The polynomial
[TABLE]
is equal to the ordinary characteristic polynomial [9, Def. 2.52] of the real arrangement consisting of the hyperplanes (ignoring possible repetitions)
Corollary 2.3**.**
Suppose both divide , and assume the following condition holds true for some positive integer : for all with . Then
[TABLE]
In particular, we have if for all with .
Proof. We apply Theorems 2.1 and 2.2. It is enough to show for with . We can choose a subset such that . Then Since [8, Lemma 2.3], This shows
∎
Corollary 2.4**.**
Suppose that and both divide and that . In addition, we assume the following condition holds true for some positive integer :
[TABLE]
for all with . Then
[TABLE]
Proof. Suppose with . It is enough to show
[TABLE]
We can choose a subset such that . Then either by (2). Since ,
[TABLE]
This shows that either We finally have
[TABLE]
∎
Corollary 2.5**.**
Suppose that and both divide and that . If are prime powers or one for all , we have .
Proof. Easily follows from Corollary 2.4. ∎
For the rest of this section we discuss general properties of the generating functions of characteristic quasi-polynomials. Let which is a primitive ’s root of unity. By (1)
[TABLE]
Note that . Therefore from the orthogonality relations among powers of , i.e., by the Fourier inversion, for each can be recovered from by
[TABLE]
This relation will be used in Example 3.5 below.
Taking a common denominator, we can express as a rational function
[TABLE]
In the numerator the powers , , correspond to . Therefore as in (3) for each we can extract these powers as
[TABLE]
Let
[TABLE]
Then
[TABLE]
Define polynomials by
[TABLE]
Then we obtain
[TABLE]
Now we present the following proposition, to which we give a proof because we were not able to find an appropriate reference in literature.
Proposition 2.6**.**
Define by (5). Let be their -th derivatives at . Then
[TABLE]
Proof. For notational simplity write and let
[TABLE]
Then the inverse Fourier transform is
[TABLE]
The -th derivative of this at is
[TABLE]
It follows that does not depend on if and only if
[TABLE]
By the use of Eulerian numbers (see Chapter III of [1]) we can write
[TABLE]
Note that for . Therefore differentiating (9) with respect to , we have for and for . Thus . By summing up (5) we have
[TABLE]
Since the Eulerian numbers are positive integers, it is not hard to see that the right hand side is a polynomial of degree with positive integer coefficients. Thus is not zero for . ∎
Proposition 2.6 and (6) imply that
[TABLE]
Furthermore note that lower order derivatives of at determine coefficients of higher degree terms in . Therefore in terms of the generating function the relations in Corollaries 2.3 and 2.4 can be written as follows:
[TABLE]
[TABLE]
3 Arrangements of root systems
Let be an -dimensional Euclidean space and be the affine space underlying . Let be an irreducible root system in of rank and . Suppose that is a set of positive roots and is the set of simple roots associated with . Denote the coefficient matrix (with an arbitrary order of columns) of the positive roots with respect to by , which is an matrix:
[TABLE]
In this section we give an explicit formula for the generating function . The formula was essentially proved by Ch. Athanasiadis [2] and by A. Blass and B. Sagan [3]. (See [6] also.) Let
[TABLE]
be the highest root. Then it is well-known that is equal to the Coxeter number; see [4, Ch. VI, §1, 11. Prop. 31].
Theorem 3.1**.**
[TABLE]
Before proving this formula after [2, 3, 6], we introduce basic concepts. Let be the basis for which is dual to the basis : . Define a free abelian group
[TABLE]
of rank . Let
[TABLE]
for and . Then is an arrangement of (infinitely many) affine hyperplanes in . The reflection with respect to is denoted by :
[TABLE]
The affine Weyl group is the group generated by . Each connected component of is called an alcove. The closure of an alcove is a fundamental domain of the group acting on [4, Ch. VI, §2, 1]. Consider a special alcove
[TABLE]
which is an open simplex with vertices [math] and the .
Proof of Theorem 3.1. ([2, 3, 6]) Let be the parallelepiped defined by
[TABLE]
Then is, by definition, a union of alcoves. The number of alcoves in is equal to
[TABLE]
see [4, Ch. VI, §2, 4. (5)]. Compute
[TABLE]
On the other hand,
[TABLE]
This function is known as the Ehrhart quasi-polynomial of the open simplex bounded by the coordinate hyperplanes and the hyperplane ; see [11, page 235ff]. Thus
[TABLE]
This completes the proof.
Corollary 3.2**.**
The minimum period of the characteristic quasi-polynomial for an irreducible root system is equal to .
Proof.
The assertion holds true because by Proposition 4.4.1 of [11] the minimum period of the Ehrhart quasi-polynomial of the open simplex bounded by the coordinate hyperplanes and the hyperplane is equal to . ∎
The minimum periods for all irreducible root systems are shown in the following table:
[TABLE]
Remark 3.3**.**
With PARI/GP we checked that for every irreducible root system the minimum period coincides with the lcm period .
Corollary 3.4**.**
Let be a positive integer. For an irreducible root system with its Coxeter number , if and only if .
Proof.
The lowest non-zero term of is equal to . ∎
Example 3.5**.**
* By Theorem 3.1 and the table above, we have the generating functions of type :*
[TABLE]
Thus the characteristic quasi-polynomials of is the same as the characteristic quasi-polynomials of . Since the minimum periods of these three root systems are all equal to two, there exist four polynomials satisfying
[TABLE]
[TABLE]
Compute
[TABLE]
and
[TABLE]
This implies
[TABLE]
Since is equal to the ordinary characteristic polynomial of by Theorem 2.2, we obtain
[TABLE]
Actually we may derive the following characteristic quasi-polynomials from the generating functions and :
[TABLE]
Remark 3.6**.**
We may also prove by constructing a one-to-one correspondence between and .
Example 3.7**.**
Let be a root system of type . We use PLATE V in [4] to get the matrix :
[TABLE]
Thus and . By Theorem 3.1, we have the generating function of type :
[TABLE]
By expanding this formal power series we have the characteristic quasi-polynomial of :
[TABLE]
We have computed , , and the constituents of the quasi-polynomial are consistent with Corollaries 2.3 and 2.4.
Remark. R. Suter [12] gave essentially the same calculation for every irreducible root system.
4 Mid-hyperplane arrangement
The mid-hyperplane arrangement was defined in [7] to find the number of “ranking patterns” generated by the unidimensional unfolding model in mathematical psychology ([5]). This arrangement is defined as follows.
Let be an integer. We define two kinds of hyperplanes as follows:
[TABLE]
where I_{4}:=\{(i,j,k,l):1\leq i<j\leq m,\ i<k<l\leq m,\ \text{jkl}\}. Then the mid-hyperplane arrangement is defined as
[TABLE]
Now, let be the coefficient matrix of , where .
4.1 Characteristic quasi-polynomial
and generating function of
When , we have , and thus . The characteristic quasi-polynomial is
[TABLE]
From this characteristic quasi-polynomial, we obtain
[TABLE]
4.2 Characteristic quasi-polynomial
and generating function of
When , we have , and thus . With the help of the relations
[TABLE]
(see Corollary 2.4), we obtained the constituents of as
[TABLE]
The generationg function is
[TABLE]
Remark 4.1**.**
For , we have , and thus . It was computationally infeasible for us to obtain , so we cannot tell whether this is the minimum period or not.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] M. Aigner, Combinatorial Theory , Springer, Berlin, 1979.
- 2[2] C. A. Athanasiadis, Characteristic polynomials of subspace arrangements and finite fields, Adv. Math. 122 (1996), 193–233.
- 3[3] A. Blass and B. Sagan, Characteristic and Ehrhart polynomials, J. Algebraic Combin. 7 (1998), 115–126.
- 4[4] N. Bourbaki, Lie Groups and Lie Algebras: Chapters 4-6, Springer-Verlag, Berlin-Heidelberg-New York, 2002
- 5[5] C. H. Coombs, A Theory of Data, John Wiley & Sons, New York, 1964.
- 6[6] M. Haiman, Conjectures on the quotient ring of diagonal invariants, J. Algebraic Combin. 3 (1994), 17–76.
- 7[7] H. Kamiya, P. Orlik, A. Takemura and H. Terao, Arrangements and ranking patterns, Ann. Comb. 10 (2006), 219-235.
- 8[8] H. Kamiya, A. Takemura and H. Terao, Periodicity of hyperplane arrangements with integral coefficients modulo positive integers, to appear in J. Alg. Combin. doi:10.1007/s 10801-007-0091-2.
