Abstract
Firstly, we give a formula on the generalized Hamming weights of linear codes constructed generically by defining sets. Secondly, by choosing properly the defining set we obtain a class of cyclotomic linear codes and then present two alternative formulas for calculating their generalized Hamming weights. Lastly, we determine their weight distributions and generalized Hamming weights partially. Especially, we solve the generalized Hamming weights completely in one case.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Avoid common mistakes on your manuscript.
1 Introduction
Let \(q=p^{e}\) for a prime p. Denote by \( \mathbb {F}_{Q} =\mathbb {F}_{q^{m}} \) the finite field with Q elements and \(\mathbb {F}_{q^{m}}^{*}\) the multiplicative group of \(\mathbb {F}_{q^{m}}\).
If C is a k-dimensional \(\mathbb {F}_{q}\)-vector subspace of \(\mathbb {F}_{q}^{n},\) then it is called an [n, k, d] linear code with length n and minimum Hamming distance d over \(\mathbb {F}_{q}.\) Here the Hamming distance d(x, y) between two codewords \(x,y\in C\) is defined as the numbers of places in which x is different from y. And \(d=\min \{d(x,0)|x\in C, x\ne 0\}\) since C is linear. Denote by \(A_{i}\) the number of codewords with Hamming weight i in C. If \(|\{i:A_i\ne 0,1\le i\le n\}|=t, \) then C is called a t-weight code. The readers are referred to [12] for more details and general theory of linear codes.
A generic construction of linear code as below was proposed by Ding et al. [5, 6]. Let \( D= \{d_{1},d_{2}, \ldots ,d_{n}\}\) be a subset of \( \mathbb {F}_Q^{*}. \) Define a linear code \( C_{D} \) of length n over \( \mathbb {F}_{q} \) as follows:
where \( \mathrm {Tr}_{Q/q}\) is the standard trace map from \(\mathbb {F}_Q\) to \(\mathbb {F}_q\) and D is called the defining set. The method is used in a lot of references to get linear codes with a few weights [9, 17, 23, 24] by choosing properly defining sets.
For an [n, k, d] linear code C, we could extend Hamming weight to obtain the concept of the generalized Hamming weight(GHW) \( d_{r}(C)(0<r\le k)\) (see [15, 20]). It is defined as follows. Denote by \( [C,r]_{q} \) the set of the r-dimensional \(\mathbb {F}_{q}\)-vector subspaces of C. For \( V \in [C,r]_{q}, \) let Supp(V) be the set of positions i where there exists a codeword \( x = (x_{1}, x_{2}, \ldots , x_{n})\in V \) with \( x_{i} \ne 0. \) Then the rth generalized Hamming weight(GHW) \( d_{r}(C)\) of the linear code C is defined by
and \( \{d_{i}(C): 1\le i \le k\} \) is defined to be the weight hierarchy of C. In particular, the GHW \( d_{1}(C)\) is just the usual minimum distance d. Since the classic results of Wei in the paper [20] in 1991, many people researched into the generalized Hamming weight. A survey on known results on this topic up to 1995 was done in [19]. Afterwards there have been a number of studies on the generalized Hamming weight of some particular families of codes [1,2,3, 7, 11, 13, 14, 21, 22]. It is worth mentioning that the recent work in [22] gave a very instructive approach to calculating the GHWs of irreducible cyclic codes. Generally, it is not easy to determine the weight hierarchy.
The rest of this paper is organized as follows: in Sect. 2, we review basic concepts and results on Gauss sum and exponential sums which are needed in this paper; in Sect. 3, we follow the work of Ding et al. [8, 10] to construct a class of cyclotomic linear codes and give general formulas on \( d_{r}(C). \) Meanwhile, we determine their weight distribution under certain conditions; in Sect. 4, we give the conclusion of this paper.
2 Preliminaries
We assume that h is a positive divisor of \(Q-1\) and \(1<h<\sqrt{Q}+1.\) And \(\theta \) is a fixed primitive element of \(\mathbb {F}_{Q}=\mathbb {F}_{q^{m}}.\) We start with the additive character. Let \(b\in \mathbb {F}_{Q}\), the mapping
defines an additive character of \(\mathbb {F}_{Q}, \) where \(\zeta _{p}=e^{\frac{2\pi \sqrt{-1}}{p}}. \) Particularly, the character \(\chi _{1}\) is called the canonical additive character of \(\mathbb {F}_{Q}\). The multiplicative characters of \(\mathbb {F}_{Q}\) are defined by
For each additive \( \chi \) and multiplicative character \(\psi ,\) we define the Gauss sum \(G_{Q}(\psi , \chi )\) over \(\mathbb {F}_{Q}\) by
The reader can refer to [16] for more information about the explicit values of Gauss sums.
For each \( \alpha \in \mathbb {F}_{Q}, \) an exponential sum \( S(\alpha ) \) is defined as follows.
For an integer i, define
It is easy to see \( C_{u}=C_{v} \) if and only if \( u\equiv v \ (\text {mod}\ h).\) These sets \( C_{i}\) and numbers \( \eta _{i} \) are called the cyclotomic classes and Gaussian periods (see [4]) of order h in \( \mathbb {F}_{Q}^{*}, \) respectively. By definition, it is not hard to get \( S(\theta ^{i})=h\eta _{i}+1.\)
The following lemma is about the explicit values of the exponential sum \( S(\alpha ). \) It will be used later.
Lemma 1
([18]) Assume \( m=2lk, h|(q^{k}+1). \) Then for any \( \alpha \in \mathbb {F}_{Q}^{*}, \)
where
Here we present three bounds on GHWs of linear codes. The reader may refer to the literature [19] for them.
Lemma 2
Let C be a linear code over \( \mathbb {F}_{q} \) with parameters [n, m]. For \(1\le r \le m, \)
-
1.
(Singleton type bound) \( r\le d_{r}(C)\le n-m+r. \) And C is called an r-MDS code if \(d_{r}(C) = n-m+r.\)
-
2.
(Griesmer-like bound)
$$\begin{aligned} d_{r}(C)\ge \sum _{i=0}^{r-1}\left\lceil \frac{d_{1}(C)}{q^{i}} \right\rceil . \end{aligned}$$ -
3.
(Plotkin-like bound)
$$\begin{aligned} d_{r}(C)\le \left\lfloor \frac{n(q^{r}-1)q^{m-r}}{q^{m}-1} \right\rfloor . \end{aligned}$$
3 Main results and proofs
First of all, we give a general formula for computing the GHWs of the linear code defined by the generic method in (1) with the defining set D.
Theorem 1
For each \( r (1\le r \le m), \) if the dimension of \( C_{D} \) is m, then \(d_{r}(C_{D})= n-\max \{|D \bigcap H|: H \in [\mathbb {F}_{Q},m-r]_{q}\}. \)
Proof
The proof is similar to that of Theorem 6 in [22]. But for the convenience of the reader, we provide the proof. Let \( \phi \) be a mapping from \( \mathbb {F}_{Q} \) to \( \mathbb {F}_{q}^{n}\) defined by
for each \(x\in \mathbb {F}_{Q}. \) Obviously, \(\phi \) is an \(\mathbb {F}_{q}\)-linear mapping and the image of \(\phi \) is \(C_{D}.\) And \(\phi \) is injective since the dimension of \( C_{D} \) is m. For an r-dimension subspace \( U_{r} \in [C_{D}, r]_{q}, \) denote by \(H_{r}\) the pre-image \(\phi ^{-1}(U_{r})\) in \(\mathbb {F}_{Q}.\) Also \(H_{r}\) is an r-dimension subspace of \(\mathbb {F}_{Q}.\) By definition, \( d_{r}(C_{D})= n-\max \{N(U_{r}): U_{r} \in [C_{D}, r]_{q}\}, \) where
Let \( \{ \beta _{1},\beta _{2},\ldots ,\beta _{r}\}\) be an \(\mathbb {F}_{q}\)-basis of \( H_{r}. \) Then
Let \( H^{\bot }=\{v\in \mathbb {F}_{Q}: \mathrm {Tr}_{Q/q}(uv)=0 \ \text {for any}\ u \in H\}.\) It is called the dual of H. We know that \( \dim _{\mathbb {F}_{q}}(H)+\dim _{\mathbb {F}_{q}}(H^{\bot })=m.\)
For \(y\in \mathbb {F}_{Q}, \)
By the above equation, we have
So the desired result follows from the fact that there is a bijection between \([\mathbb {F}_{Q},r]_{q}\) and \([\mathbb {F}_{Q},m-r]_{q}.\) We complete the proof.
From now on, we suppose \( h(q-1)\) is also a divisor of \(Q-1.\) In [10], C. Ding and H. Niederreiter presented two classes of cyclotomic linear codes of order 3 and determined their weight distributions. Inspired by their work, we construct linear codes by choosing the defining set to be
where \(d_{i}=\theta ^{h(i-1)}, n_{0}=\frac{q^{m}-1}{h(q-1)}, 0\le t_{1}<t_{2}<\cdots <t_{s} \le h-1, 1\le s\le h. \) The code \(C_{\overline{D}}\) is closely related to irreducible cyclic codes. Note that \(\{ d_{1}, d_{2}, \ldots , d_{n_{0}}\}\) is a complete set of coset representatives of the quotient group \(C_{0}/\mathbb {F}_{q}^{*}\) since \(h(q-1)\) divides \(Q-1.\) If \(s=1, t_{1}=0,\) then \(C_{\overline{D}}\) is a linear code punctured from the code \(C_{C_{0}}\) (see [8]). Here \(C_{C_{0}}\) is the code defined in (1) with the defining set \(D=C_{0}.\) It is known as the irreducible cyclic code. Thus we also call \(C_{\overline{D}}\) a cyclotomic linear code since \(\overline{D}\) has relation to the cyclotomic classes of order h in \( \mathbb {F}_{Q}^{*}. \)
In addition to Theorem 1, we give alternative formulas for calculating the GHWs of the cyclotomic linear code \(C_{\overline{D}}\).
Theorem 2
For each \( r(1\le r \le m),\) if \(\dim (C_{\overline{D}})=m,\) then \( d_{r}(C_{\overline{D}})= sn_{0}-N_{r}, \) where
-
1.
\( N_{r}=\frac{s(q^{m}-q^{r})}{hq^{r}(q-1)}+\frac{1}{hq^{r}(q-1)}\max \{ A_{H_{r}} : H_{r} \in [\mathbb {F}_Q, r]_{q}\}, \) and
\( A_{H_{r}} =\sum _{j=1}^{s}\sum _{\lambda =1}^{h-1}\sum _{\beta \in H_{r}^{*}}\overline{\varphi ^{\lambda }} (\beta \theta ^{t_{j}})G_{Q}(\varphi ^{\lambda }), \) or
-
2.
\( N_{r}=\frac{sn_{0}}{q^{r}}+\frac{1}{q^{r}(q-1)}\max \{ \sum _{i=0}^{h-1}|H_{r}\bigcap (\bigcup _{j=1}^{s}C_{i-t_{j}})|\eta _{i} : H_{r} \in [\mathbb {F}_Q, r]_{q}\}.\)
Proof
1. By definition, \( d_{r}(C_{\overline{D}})= sn_{0}-N_{r}, N_{r}=\max \{N(U_{r}):U_{r}\in [{C}_{\overline{D}},r]_{q}\}. \) Let \( \{ \beta _{1},\beta _{2},\ldots ,\beta _{r}\}\) be an \(\mathbb {F}_{q}\)-basis of \( H_{r}. \) Here \( \phi ( H_{r})=U_{r}.\) See the proof of Theorem 1 for the definitions of \(N(U_{r})\) and \(\phi .\) Set \( H_{r}^{*}=H_{r}\backslash \{0\}. \) Then
where \( \mathbb {F}^{*}_{q}\overline{D}=\{xy|x\in \mathbb {F}^{*}_{q}, \ y\in \overline{D}\}.\) So
For simplicity, we set \( A_{H_{r}} = \sum _{j=1}^{s}\sum _{\lambda =1}^{h-1}\overline{\varphi ^{\lambda }}(\theta ^{t_{j}})G_{Q}(\varphi ^{\lambda })\sum _{\beta \in H_{r}^{*}} \overline{\varphi ^{\lambda }}(\beta ). \) So
2. By the proof of Part 1, we have
By definition, \( \eta _{i}=\sum _{x\in C_{i}}\chi _{1}(x). \) So we have
Then the desired result follows and the proof is completed.
Remarks
-
(1)
If \( s=h,\) then \( C_{\overline{D}}\) is a \([\frac{q^{m}-1}{q-1},m,q^{m-1}]\) code, the nonzero elements of which all have weights \(q^{m-1}.\) It is a simplex code. By Theorem 1 or Theorem 2(2), it is easy to get \( d_{r}(C_{\overline{D}})=\frac{q^{m}-q^{m-r}}{q-1}. \)
-
(2)
By the construction of \(\overline{D},\) for any two elements \(\alpha \) and \(\beta \) in \( \overline{D},\) we have \((\frac{\alpha }{\beta })^{q-1}\ne 1.\) This means that \(\frac{\alpha }{\beta }\notin \mathbb {F}^{*}_{q}.\) So \(\max \{|\overline{D} \bigcap H|: H \in [\mathbb {F}_{Q},1]_{q}\}=1. \) By Theorem 1, if \( \dim (C_{\overline{D}})=m,\) then \( d_{m-1}(C_{\overline{D}})=\frac{s(q^{m}-1)}{h(q-1)}-1. \) By the Singleton type bound in Lemma 2, \( C_{\overline{D}} \) is an \((m-1)\)-MDS code [19] over \(\mathbb {F}_q. \) Especially, if \( m=2, \) then the code \( C_{\overline{D}} \) is an \( [\frac{s(q+1)}{h},2,\frac{s(q+1)}{h}-1]\) MDS code [12] over \(\mathbb {F}_q. \)
-
(3)
Generally, it is difficult to establish linkage between the additive properties and the multiplicative ones of a field. So Theorems 1 and 2 indicate that it is difficult to give the explicit values of the generalized Hamming weights of \(C_{\overline{D}}\) in other cases.
Next under certain conditions, we give the weight distributions of the cyclotomic linear codes \(C_{\overline{D}}\) in the following theorem.
Theorem 3
Assume \( m=2lk \) and \( h|(q^{k}+1). \) Then the code \(C_{\overline{D}}\) is an \([\frac{s(Q-1)}{h(q-1)},m]\) linear code over \( \mathbb {F}_{q} \) with the weight distribution in Table 1. And the dual code \(C_{\overline{D}}^{\perp }\) of \(C_{\overline{D}}\) is an \([\frac{s(Q-1)}{h(q-1)}, \frac{s(Q-1)}{h(q-1)}-m, d^{\perp }]\) linear code with minimum distance \( d^{\perp }\ge 3.\)
Proof
For \(x \in \mathbb {F}_{q}^{*}, \) let \(c_{x}=(\mathrm {Tr}_{Q/q}(xd))_{d\in \overline{D}}\) and \(w(c_{x})\) denote the Hamming weight of the codeword \(c_{x},\) then we have
By Lemma 1, we have
As for the parameters of the dual code, it is enough to prove \( d^{\perp }\ge 3. \) It is easy to show that any two elements in \(\overline{D}\) are linearly independent over \(\mathbb {F}_{q}.\) Then the desired results follow and we complete the proof.
Example 1
Let \((q,m,l,k,h,s)=(3,4,1,2,5,3)\) and \((t_{1},t_{2},t_{3})=(1,2,3).\) Then, the corresponding code \(C_{\overline{D}}\) has parameters [24, 4, 15], weight enumerator \(1+48x^{15}+32x^{18}\) and its dual code has parameters [24, 20, 3].
Example 2
Let \((q,m,l,k,h,s)=(5,4,2,1,6,2)\) and \((t_{1},t_{2})=(0,1).\) Then, the corresponding code \(C_{\overline{D}}\) has parameters [52, 4, 40], weight enumerator \(1+416x^{40}+208x^{45}\) and its dual code has parameters [52, 48, 3].
Corollary 1
Assume \( m=2lk \) and \( h|(q^{k}+1). \) If \( (l,2)=1, \) then
Proof
By Lemma 1, we have \( \eta _{i}=\frac{(h-1)\sqrt{Q}-1}{h} \) if \( i=h_{0}, \) otherwise \( \eta _{i}=\frac{-\sqrt{Q}-1}{h}. \) So by Theorem 2(2), we get \( \sum _{j=1}^{s}\sum _{i=0}^{h-1}|H_{r}\bigcap C_{i-t_{j}}|\eta _{i} \)
If \( (l,2)=1, \) then \(\mathbb {F}_{q^{lk}}\subset C_{0}.\) Notice that \( C_{i}=\theta ^{i}C_{0} \) and \( \theta ^{i}H_{r}\) is also an r-dimension subspace. So we have
for each r with \( 1\le r \le \frac{m}{2}. \) By Theorem 2, we get the first part of the corollary. If \( \frac{m}{2}\le r \le m, \) then \( 0\le m-r \le \frac{m}{2}. \) So we know that there is an \((m-r)\)-dimensional subspace \(H_{m-r}\subset \mathbb {F}_{q^{lk}}\subset C_{0}.\) Therefore, \( \max \{|(\bigcup _{j=1}^{s}C_{t_{j}}) \bigcap H|: H \in [\mathbb {F}_{Q},m-r]_{q}\} =q^{m-r}-1. \) Note that \( \mathbb {F}^{*}_{q}\overline{D}=\bigcup _{j=1}^{s}C_{t_{j}}\) and \(\frac{\alpha }{\beta }\notin \mathbb {F}^{*}_{q}\) for any two elements \(\alpha \) and \(\beta \) in \(\overline{D}.\) So \(|(\bigcup _{j=1}^{s}C_{t_{j}}) \bigcap H|=(q-1)|H\bigcap \overline{D}|\) for any subspace H. Therefore, \( \max \{|\overline{D} \bigcap H|: H \in [\mathbb {F}_{Q},m-r]_{q}\} =\frac{q^{m-r}-1}{q-1}. \) By Theorem 1, we get the second part of this corollary. The proof is completed.
Example 3
For the code in Example 1, its weight hierarchy is \(d_{1}=15, d_{2}=20, d_{3}=23, d_{4}=24.\)
Corollary 2
Also assume \( m=2lk \) and \( h|(q^{k}+1). \) If \( l=2^{u}l'\) with \( u>0, (l',2)=1,\) and \( s<h, \) then
Proof
Also by Lemma 1, we have \( \eta _{i}=-\frac{(h-1)\sqrt{Q}+1}{h} \) if \( i=h_{0}, \) otherwise \( \eta _{i}=\frac{\sqrt{Q}-1}{h}. \) For an r-dimensional subspace \(H_{r},\)
By assumption, we have \(\mathbb {F}_{q^{l'k}}\subset C_{0}.\) Notice that \(\bigcup _{j=1}^{s}C_{h_{0}-t_{j}}\ne \overline{D}\) since \(s<h.\) So for each r with \( 1\le r \le l'k, \) we have
Then the desired result of the first part follows directly from Theorem 2(2). For \( m-l'k\le r \le m, \) we know that there is an \((m-r)\)-dimensional subspace \(H_{m-r}\subset \mathbb {F}_{q^{l'k}}\subset C_{0}.\) So \( \max \{|(\bigcup _{j=1}^{s}C_{t_{j}}) \bigcap H|: H \in [\mathbb {F}_{Q},m-r]_{q}\} =q^{m-r}-1\) and \( \max \{|\overline{D} \bigcap H|: H \in [\mathbb {F}_{Q},m-r]_{q}\} =\frac{q^{m-r}-1}{q-1}. \) By Theorems 1 and 3, we get the second part of this corollary. The proof is completed.
Example 4
For the code in Example 2, its weight hierarchy is \(d_{1}=40, d_{2}=48, d_{3}=51, d_{4}=52.\)
The above four examples have been verified by Magma.
4 Concluding remarks
In this paper, we gave a formula for computing the generalized Hamming weights of linear code \(C_{D}, \) which is constructed by the generic method proposed by Ding et al. By choosing properly the defining set, we presented a class of cyclotomic linear codes \( C_{\overline{D}}. \) We gave two alternative formulas about their generalized Hamming weights in terms of Gauss sums and Gaussian periods. Under certain conditions, we solved the weight distribution of \( C_{\overline{D}} \) and proved that it is a two-weight linear code. We determined completely the generalized Hamming weights of \( C_{\overline{D}} \) in one case.
References
Bras-Amorós, M., Lee, K., Vico-Oton, A.: New lower bounds on the generalized Hamming weights of AG codes. IEEE Trans. Inf. Theory 60(10), 5930–5937 (2014)
Barbero, A.I., Munuera, C.: The weight hierarchy of Hermitian codes. SIAM J. Discrete Math. 13(1), 79–104 (2000)
Cheng, J., Chao, C.-C.: On generalized Hamming weights of binary primitive BCH codes with minimum distance one less than a power of two. IEEE Trans. Inf. Theory 43(1), 294–298 (1997)
Ding, C.: Codes from Difference Sets. World Scientific, Singapore (2015)
Ding, C.: Linear codes from some 2-designs. IEEE Trans. Inf. Theory 61(6), 3265–3275 (2015)
Ding, K., Ding, C.: Bianry linear codes with three weights. IEEE Commun. Lett. 18(11), 1879–1882 (2014)
Delgado, M., Farrn, J.I., Garca-Snchez, P.A., Llena, D.: On the weight hierarchy of codes coming from semigroups with two generators. IEEE Trans. Inf. Theory 60(1), 282–295 (2014)
Ding, C., Luo, J., Niederreiter, H.: Two-weight codes punctured from irreducible cyclic codes. In: Li, Y., et al. (Eds.), Proceedings of the First Worshop on Coding and Cryptography. World Scientific, Singapore, pp. 119–124 (2008)
Ding, C., Li, C., Li, N., Zhou, Z.: Three-weight cyclic codes and their weight distributions. Discrete Math. 339(2), 415–427 (2016)
Ding, C., Niederreiter, H.: Cyclotomic linear codes of order 3. IEEE Trans. Inf. Theory 53(6), 2274–2277 (2007)
Heijnen, P., Pellikaan, R.: Generalized Hamming weights of q-ary ReedCMuller codes. IEEE Trans. Inf. Theory 44(1), 181–196 (1998)
Huffman, W.C., Pless, V.: Fundamentals of Error-Correcting Codes. Cambridge University Press, Cambridge (2003)
Jian, G., Feng, R., Wu, H.: Generalized Hamming weights of three classes of linear codes. Finite Fields Appl. 45, 341–354 (2017)
Janwa, H., Lal, A.K.: On the generalized Hamming weights of cyclic codes. IEEE Trans. Inf. Theory 43(1), 299–308 (1997)
Kløve, T.: The weight distribution of linear codes over \(GF(q^{l})\) having generator matrix over \(GF(q>)\). Discrete Math. 23(2), 159–168 (1978)
Lidl, R., Niederreiter, H.: Finite Fields. Cambridge University Press, New York (1997)
Li, C., Yue, Q., Li, F.: Hamming weights of the duals of cyclic codes with two zeros. IEEE Trans. Inf. Theory 60(7), 3895–3902 (2014)
Moisio, M.J.: A note on evaluations of some exponential sums. Acta Arith. 93, 117–119 (2000)
Tsfasman, M.A., Vladut, S.G.: Geometric approach to higher weights. IEEE Trans. Inf. Theory 41(6), 1564–1588 (1995)
Wei, V.K.: Generalized Hamming weights for linear codes. IEEE Trans. Inf. Theory 37(5), 1412–1418 (1991)
Xiong, M., Li, S., Ge, G.: The weight hierarchy of some reducible cyclic codes. IEEE Trans. Inform. Theory 62(7), 4071–4080 (2016)
Yang, M., Li, J., Feng, K., Lin, D.: Generalized Hamming weights of irreducible cyclic codes. IEEE Trans. Inf. Theory 61(9), 4905–4913 (2015)
Yang, S., Yao, Z.: Complete weight enumerators of a family of three-weight linear codes. Des. Codes Cryptogr. 82(3), 663–674 (2017)
Zhou, Z., Li, N., Fan, C., Helleseth, T.: Linear codes with two or three weights from quadratic bent functions. Des. Codes Cryptogr. 81(2), 283–295 (2016)
Acknowledgements
I explicitly acknowledge anonymous reviewers for their valuable suggestions and comments, which have helped improve the quality of the paper. I am also extremely grateful to the editors for their careful considerations and kind help.
Author information
Authors and Affiliations
Corresponding author
Additional information
This research is supported in part by National Natural Science Foundation of China (61602342).
Rights and permissions
About this article
Cite this article
Li, F. A class of cyclotomic linear codes and their generalized Hamming weights. AAECC 29, 501–511 (2018). https://doi.org/10.1007/s00200-018-0355-1
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00200-018-0355-1