Projective Reed–Muller type codes on

higher dimensional scrolls

Cícero Carvalho, Xavier Ramírez-

Mondragón, Victor G. L. Neumann &
Horacio Tapia-Recillas

Designs, Codes and Cryptography

An International Journal

ISSN 0925-1022

Des. Codes Cryptogr.

DOI 10.1007/s10623-018-00603-8

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
Science+Business Media, LLC, part of
Springer Nature. This e-offprint is for personal
use only and shall not be self-archived in
electronic repositories. If you wish to self-
archive your article, please use the accepted
manuscript version for posting on your own
website. You may further deposit the accepted
manuscript version in any repository,
provided it is only made publicly available 12
months after official publication or later and
provided acknowledgement is given to the
original source of publication and a link is
inserted to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at link.springer.com”.

1 23
 Author's personal copy
Designs, Codes and Cryptography
https://doi.org/10.1007/s10623-018-00603-8

Projective Reed–Muller type codes on higher dimensional

scrolls

Cícero Carvalho1 · Xavier Ramírez-Mondragón2 · Victor G. L. Neumann1 ·

Horacio Tapia-Recillas2

Received: 2 November 2017 / Revised: 12 November 2018 / Accepted: 20 December 2018

© Springer Science+Business Media, LLC, part of Springer Nature 2019

Abstract
In 1988 Lachaud introduced the class of projective Reed–Muller codes, defined by evaluating
the space of homogeneous polynomials of a fixed degree d on the points of Pn (Fq ). In this
paper we evaluate the same space of polynomials on the points of a higher dimensional scroll,
defined from a set of rational normal curves contained in complementary linear subspaces
of a projective space. We determine a formula for the dimension of the codes, and the exact
value of the dimension and the minimum distance in some special cases.

Keywords Projective variety codes · Evaluation codes · Reed–Muller type codes · Higher
dimensional scroll

Mathematics Subject Classification 11T71 · 13P25 · 94B60

Communicated by G. Korchmaros.

C. Carvalho and V. G. L. Neumann are partially supported by CNPq and FAPEMIG. X. Ramírez-Mondragón
is supported by CONACyT Scholarship No. 209918.

B Xavier Ramírez-Mondragón
[email protected]
Cícero Carvalho
[email protected]
Victor G. L. Neumann
[email protected]
Horacio Tapia-Recillas
[email protected]
1 Faculdade de Matemática, Universidade Federal de Uberlândia, Av. J. N. Ávila 2121, 38.408-902
Uberlândia, MG, Brazil
2 Departamento de Matemáticas, Universidad Autónoma Metropolitana, Av. San Rafael Atlixco 186,
Col. Vicentina, C.P. 09340, Delegación Iztapalapa, CD. Mexico, Mexico

123
 Author's personal copy
C. Carvalho et al.

1 Introduction

In 1988 Lachaud introduced the class of projective Reed–Muller codes (see [14]), defined
by evaluating the space of homogeneous polynomials of a fixed degree d on the points
of Pn (Fq ), the n-dimensional projective space over a finite field Fq with q elements.
For the purpose of the evaluation, the coordinates of the points in the projective space
are considered to be written in standard notation, i.e., the leftmost nonzero entry is
equal to 1. In [17] Sörensen found formulas for the dimension and minimum distance
of these codes and in [15] Rentería and Tapia-Recillas gave a different approach to
arrive at those formulas. Recently, there has been some interest in the problem of deter-
mining the generalized Hamming weights of projective Reed–Muller type codes (see
[2,3,10]).
The classical definition of the projective Reed–Muller code can be extended to arbitrary
subsets X of the projective space Pn (Fq ), and the codes thus produced are called Reed–
Muller type (projective) codes. Several instances where X is the set of Fq -rational points
of a projective variety defined over Fq have already been studied, for example, the cases
where X is a zero-dimensional complete intersection in Pn (Fq ) [9], the Segre variety [11],
smooth quadric surfaces and twisted Segre varieties [7], and the Veronese variety [16], among
others (see [5,6,12,15]). These investigations form a growing corpus of theoretical knowledge
which is crucial for future applications. Of course, they must be paired with studies on the
decoding of the proposed codes. Also, for practical applications, a code may have to wait
for the right moment, for example, Reed-Solomon codes were created in 1960 but started
to be widely used in applications only in the beginning of the 80’s, with the appearance of
CD’s.
In the present manuscript we continue the study of projective Reed–Muller type codes
set forward by the above works. In [5] the authors studied the case where X is a ratio-
nal normal scroll surface. They determined the dimension of the codes, a lower bound
for the minimum distance and its exact value in some cases. In this paper we consider
higher dimensional scrolls, defined from a set of rational normal curves contained in com-
plementary linear subspaces of a projective space. We also determine a formula for the
dimension of the codes, and the exact value of the dimension and the minimum dis-
tance in some special cases. To prove our results we generalized some of the methods
used in [5], and we present them in a more systematic form. A relationship between
the introduced codes and a direct product of (classical) projective Reed–Muller codes is
given, which explains the formula for the exact value of the minimum distance in some
cases.
The paper is organized as follows. In the next section, the higher dimensional scroll S
(following [13]) over the finite field Fq and the projective Reed–Muller type code C S (d)
obtained by evaluating homogeneous polynomials of degree d at its Fq -rational points are
introduced. In Sect. 3, an alternative construction for C S (d), which will be of fundamental
importance in the determination of its parameters is presented. Then, we proceed to deter-
mine a general formula for the dimension of C S (d) and its exact value in particular cases.
Examples of the studied codes are given where the parameters are completely determined.
In the last section, a connection between a special case of C S (d) and the direct product of
(classical) projective Reed–Muller codes is presented, which allows the determination of the
parameters of C S (d).

123
 Author's personal copy
Projective Reed–Muller type codes on higher dimensional scrolls

2 Higher dimensional scrolls and Reed–Muller type codes

In this section we present the definition of higher dimensional scrolls and the codes which
we will study in the rest of the manuscript. We refer the interested reader to [13] for results
and concepts on algebraic geometry, particularly in higher dimensional scrolls.
Let e ≥ 1 be an integer. A rational normal curve of degree e is the image of the map
ve : P1 (Fq ) → Pe (Fq )
(x0 : x1 )  → (x0e : x0e−1 x1 : · · · : x1e ).
Let e0 ≥ e1 ≥ e2 ≥ . . . ≥ en ≥ 1 be integers, and let

n
 = (e0 + 1) + (e1 + 1) + · · · + (en + 1) − 1 = ei + n.
i=0

A point in P (Fq ) will be denoted by

 
x0,0 : · · · : x0,e0 : x1,0 : · · · : x1,e1 : · · · : xn,0 : · · · : xn,en .
For i ∈ {0, . . . , n}, the set of points in P (Fq ) such that xs,t = 0 for all t ∈ {0, . . . , es } and
all s ∈ {0, . . . , n}\{i} is a linear subspace of dimension ei , which is denoted by Pei , and the
image of the map
u i : P1 (Fq ) →  Pei ⊂ P (Fq ) 
(b0 : b1 )  → 0 : · · · : b0ei : b0ei −1 b1 : · · · : b1ei : 0 · · · : 0 ,

is a rational normal curve of degree ei . For each (b0 : b1 ) ∈ P1 (Fq ), we define L (b0 :b1 ) as the
linear subspace of P (Fq ) spanned by the points u 0 (b0 : b1 ), . . . , u n (b0 : b1 ), i.e., L (b0 :b1 )
is the set of points of the form
 
a0 b0e0 : a0 b0e0 −1 b1 : · · · : a0 b1e0 : · · · : an b0en : an b0en −1 b1 : · · · : an b1en ,

where (a0 : · · · : an ) ∈ Pn (Fq ).

Definition 2.1 The higher dimensional scroll of type e0 , . . . , en is the set

Se0 ,...,en := L (b0 :b1 ) ⊂ P (Fq ).
(b0 :b1 )∈P1
If n = 1 the previous definition gives the rational normal scroll studied in [5] and in this
work some results presented there are generalized.
From now on, we will only write scroll when we refer to a higher dimensional scroll.
Everything we did so far in this section makes sense if we replace Fq by any field
K . If K is an algebraically closed field it is shown that Se0 ,...,en is an algebraic vari-
ety of dimension n + 1 which is the zero locus in P (K ) of the ideal of polynomials in
K [X 0,0 , . . . , X 0,e0 , . . . , X n,0 , . . . , X n,en ] generated by the 2 × 2 minors of the 2 × (l + 1)
matrix (see for instance pp. 92–93 and 105–109 of [13]),

X 0,0 . . . X 0,e0 −1 X 1,0 . . . X 1,e1 −1 . . . X n,0 . . . X n,en −1
M= .
X 0,1 . . . X 0,e0 X 1,1 . . . X 1,e1 . . . X n,1 . . . X n,en
Since the integers e0 , . . . , en are fixed throughout the paper, below we will denote Se0 ,...,en
simply by S. In [5] the authors use the fact that the rational normal scroll surface is a disjoint
union of lines. Here a generalization of this result is given which will be used to count the
number of rational points of S.

123
 Author's personal copy
C. Carvalho et al.

Proposition 2.2 If (b0 : b1 ) and (b̃0 : b̃1 ) are distinct points of P1 (Fq ) then L (b0 :b1 ) and
L (b̃0 :b̃1 ) are disjoint subspaces.

Proof Suppose (a0 : · · · : an ) and (ã0 : · · · : ãn ) exists in Pn (Fq ) such that
 
a0 b0e0 : a0 b0e0 −1 b1 : · · · : a0 b1e0 : · · · : an b0en : an b0en −1 b1 : · · · : an b1en
 
= ã0 b̃0e0 : ã0 b̃0e0 −1 b̃1 : · · · : ã0 b̃1e0 : · · · : ãn b̃0en : ãn b̃0en −1 b̃1 : · · · : ãn b̃1en .

Let j be the least subscript for which a j  = 0. Assume that b0  = 0, then the first nonzero
e
entry from left to right in the coordinates of the common point is a j b0 j , at position ( j, 0), so
ei e
in particular, for all i < j we have ãi b̃0 = 0 and so ãi = 0. Also, we must have ã j b̃0 j  = 0
so that ã j  = 0. Taking the ratios between entry ( j, 1) and ( j, 0) we get
e −1 e −1
a j b0 j b1 ã j b̃0 j b̃1
e = e ,
a j b0 j ã j b̃0 j

hence b1 /b0 = b̃1 /b̃0 .

Now assume that b0 = 0, then the first nonzero entry from left to right in the coordinates
e e
of the common point is a j b1 j , at position ( j, e j ), hence ã j b̃1 j  = 0 and for all i < j we have
e e
ãi b̃1 j = 0, so that ãi = 0. We also have ã j b̃0 j = 0, hence b̃0 = 0. This concludes the proof
that we must have (b0 : b1 ) = (b̃0 : b̃1 ). 

Corollary 2.3 S is the disjoint union of q + 1 linear subspaces of dimension n and |S| =
(q n + · · · + q + 1)(q + 1).

Proof For any (b0 : b1 ) ∈ P1 (Fq ) the points in the set {u 0 (b0 : b1 ), . . . , u n (b0 : b1 )} lie in
linear subspaces of P (Fq ) which are mutually disjoint, thus the set is linearly independent
and L (b0 :b1 ) is a linear space of dimension n. Consequently L (b0 :b1 ) contains q n + · · · + q + 1
rational points and hence |S| = (q n + · · · + q + 1)(q + 1). 

Let N := (q n + · · · + q + 1)(q + 1) and let P1 , . . . , PN be the rational points of the scroll.

We write
Fq [X] := Fq X 0,0 , . . . , X 0,e0 , X 1,0 , . . . , X 1,e1 , . . . , X n,0 , . . . , X n,en ,
and for a nonnegative integer d, Fq [X]d , will denote the Fq -vector space of polynomials in
Fq [X] of degree d.
Definition 2.4 Let
evd : Fq [X]d → FqN
f  → ( f (P1 ), . . . , f (PN ))
be the evaluation morphism, where the (rational) points of the scroll are written in standard
notation. Clearly evd is an Fq -linear transformation and its image, denoted by C S (d), is the
projective Reed–Muller type code associated to S.
Let I S be the ideal of Fq [X] generated by all homogeneous polynomials which vanish
on S and denote by I S (d) its degree d component, then clearly C S (d) is isomorphic, as an
Fq -vector space, to Fq [X]d /I S (d).
We endow the set of monomials in Fq [X] with the lexicographic order where
X 0,0 > X 0,1 > · · · > X 0,e0 > X 1,1 > · · · > X 1,e1 > · · · > X n,0 > · · · > X n,en .

123
 Author's personal copy
Projective Reed–Muller type codes on higher dimensional scrolls

For s, u ∈ {0, . . . , n}, t ∈ {0, . . . , es − 1}, v ∈ {1, . . . , eu } with either s < u or s = u and
t + 1 < v, let
g(s,t),(u,v) := X s,t X u,v − X s,t+1 X u,v−1 .
Let G be the set of all polynomials g(s,t),(u,v) which are the 2 × 2 minors of the matrix M
introduced above, and let I denote the ideal of Fq [X] generated by G. Clearly I ⊂ I S . It will
be shown that I  I S , and the following result is needed in order to do so.
Proposition 2.5 The set G is a Gröbner basis for I .

Proof Let g(s,t),(u,v) , g(i, j),(k,l) ∈ G and observe that their leading monomials are X s,t X u,v
and X i, j X k,l , respectively. According to [8, pp. 103–104], in order to prove that G is a
Gröbner basis for I it suffices to show that the S-polynomial of g(s,t),(u,v) and g(i, j),(k,l) , in
the case where the leading monomials are not coprime, may be written as a1 g1 + · · · + am gm ,
with ai ∈ Fq [X], gi ∈ G and the leading monomial of ai gi being less or equal to the leading
monomial of the S-polynomial, for all i = 1, . . . , m.
Assume that X s,t < X i, j . Then we must have either X u,v = X i, j or X u,v = X k,l so that
X s,t X u,v and X i, j X k,l are not coprime. If X u,v = X i, j , then
S(g(s,t),(u,v) , g(u,v),(k,l) ) = X k,l g(s,t),(u,v) − X s,t g(u,v),(k,l)
= −X k,l X s,t+1 X u,v−1 + X s,t X u,v+1 X k,l−1
= −X s,t+1 (X u,v−1 X k,l − X u,v X k,l−1 )
+X k,l−1 (X s,t X u,v+1 − X s,t+1 X u,v )
= −X s,t+1 g(u,v−1),(k,l) + X k,l−1 g(s,t),(u,v+1) .
If X u,v = X k,l , then
S(g(s,t),(u,v) , g(i, j),(u,v) ) = X i, j g(s,t),(u,v) − X s,t g(i, j),(u,v)
= X s,t X i, j+1 X u,v−1 − X i, j X s,t+1 X u,v−1
= X u,v−1 g(s,t),(i, j+1) .
Assume now that X s,t = X i, j (and assume w.l.g. that X u,v > X k,l ). Then
S(g(s,t),(u,v) , g(s,t),(k,l) ) = −X s,t+1 (X u,v−1 X k,l − X u,v X k,l−1 ).
Since X u,v > X k,l , it follows that either u < k or u = k and v < l, so in either case
S(g(s,t),(u,v) , g(s,t),(k,l) ) = −X s,t+1 g(u,v−1),(k,l) . This concludes the proof of the Proposition.


A consequence of this Proposition is that I  I S . Indeed, consider the polynomial

X 0,0 X 1,0 − X 0,0 X 1,0 which clearly vanishes at all points of P (Fq ). Its leading monomial is
q q
q
X 0,0 X 1,0 and (since X 0,0 and X 1,0 only appear in the first row of M) one can easily check
that it is not a multiple of the leading monomial of any polynomial in G, hence it is not an
element of I .

3 An alternative construction of CS (d)

As observed above, the ideal I S contains properly the ideal I , and as far as we can tell, it is
not easily determined. Instead of determining I S , or I S (d) directly, we follow the ideas (and

123
 Author's personal copy
C. Carvalho et al.

generalize the methods) of [5] to find a quotient which is isomorphic to Fq [X]d /I S (d), to
compute the dimension of C S (d).
Motivated by the construction of the higher dimensional scroll, and the expression of the
entries of the points in the spaces L (b0 :b1 ) , we define the following morphism of algebras
ψ : Fq [X] → Fq [Y0 , . . . , Yn , Z 0 , Z 1 ],
e −j j
which takes X i, j to Yi Z 0i Z 1 for all i ∈ {0, . . . , n} and j ∈ {0, . . . , ei }.
We will write Fq [Y, Z] for Fq [Y0 , . . . , Yn , Z 0 , Z 1 ]. The following Lemma, whose (omit-
ted) proof consists of simple calculations, shows what the image under ψ of a monomial
looks like.
α α β
0,0
Lemma 3.1 Let X 0,0 · · · X n,e
n,en
n ∈ Fq [X]d be a monomial of degree d and let Y0 0 · · ·
β γ γ
Yn n Z 0 0 Z 1 1 be its image under ψ. Then:
ei
1. βi = j=0 αi, j for all 0 ≤ i ≤ n.
n
2. i=0 βi = d.
n ei
3. γ0 = i=0 j=0 (ei − j)αi, j .
n ei
4. γ1 = i=0 j=0 jαi, j .
n
5. γ0 + γ1 = i=0 ei βi .
Let B := ψ(Fq [X]) ⊂ Fq [Y, Z]. Then B has the graded algebra structure by taking the
β β γ γ n
degree of Y0 0 · · · Yn n Z 0 0 Z 1 1 as i=0 βi . Thus its homogeneous component of degree d,
Bd , is ψ(Fq [X]d ) and ψ is a (degree zero) morphism of graded algebras.

Proposition 3.2 Bd is a finite Fq -vector space which has


β γ γ

n 
n
Md := Y0 0 · · · Ynβn Z 0 0 Z 1 1 ∈ Fq [Y, Z] : βi = d, γ0 + γ1 = ei βi
i=0 i=0

as a basis.

Proof From the above Lemma it follows that Bd is a subspace of the space generated by Md .
n
To show the other inclusion, let β0 , . . . , βn be nonnegative integers such that i=0 βi = d.
β0 βn γ0 γ1
It suffices to prove that any monomial of the form Y0 · · · Yn Z 0 Z 1 , with γ0 + γ1 =
n
i=0 ei βi , is an element of Bd , and for this we will do an induction on γ0 .
When γ0 = 0 we have
  n
β0 β1 βn β eβ
ψ X 0,e X
0 1,e1
· · · X n,e n
= Y0 0 · · · Ynβn Z 00 Z 1 i=0 i i ∈ Bd .
n
Now assume that given 0 ≤ γ0 < i=0 ei βi and nonnegative integers β0 , . . . , βn such
n
that i=0 βi = d, a monomial exists
α α0,e α α
0,0
X 0,0 · · · X 0,e00 · · · X n,0
n,0
· · · X n,e n ,
n,en

such that
 α0,e

α0,0 αn,0 αn,en β γ γ
ψ X 0,0 · · · X 0,e00 · · · X n,0 · · · X n,e n = Y0 0 · · · Ynβn Z 0 0 Z 1 1 ,
n
where γ0 + γ1 = i=0 ei βi . Note that γ1 > 0 so αk,l > 0 for some 1 ≤ l ≤ ek and some
0 ≤ k ≤ n. Let  αk,l−1 = αk,l−1 + 1,  αk,l = αk,l − 1, and set 
αi, j = αi, j otherwise, where
i ∈ {0, . . . , n} and j ∈ {0, . . . , ei }. From the relations of Lemma 3.1 it follows that

123
 Author's personal copy
Projective Reed–Muller type codes on higher dimensional scrolls

 
α0,e


α0,0 
αn,0 
αn,en β γ +1 γ −1
ψ X 0,0 · · · X 0,e00 · · · X n,0 · · · X n,e n = Y0 0 · · · Ynβn Z 0 0 Z 1 1 ,

which completes the proof. 

Corollary 3.3 The set


β γ γ

n
M := Y0 0 · · · Ynβn Z 0 0 Z 1 1 ∈ Fq [Y, Z] : γ0 + γ1 = ei βi ,
i=0

is a basis for B as an Fq -vector space.

From Corollary 2.3 we know that the cartesian product Pn (Fq ) × P1 (Fq ) and S have the
same number of rational points, and the construction of S together with Proposition 2.2,
suggest the bijection ϕ : Pn (Fq ) × P1 (Fq ) → S defined by
ϕ((a0 : · · · : an ), (b0 : b1 ))
 
= a0 b0e0 : a0 b0e0 −1 b1 : · · · : a0 b1e0 : · · · : an b0en : an b0en −1 b1 : · · · : an b1en ,

where the points of Pn (Fq ) and P1 (Fq ) are always written in standard notation. For i =
1, . . . , N , let Q i = ϕ −1 (Pi ), where P1 , . . . , PN is the set of points of S used to define
C S (d). Now define the Fq -linear evaluation map
 d : Bd →
ev FqN
f  → ( f (Q 1 ), . . . , f (Q N )),
β β γ γ
where the evaluation of a monomial Y0 0 · · · Yn n Z 0 0 Z 1 1 at such a pair is given as
β γ γ β γ γ
Y0 0 · · · Ynβn Z 0 0 Z 1 1 ((a0 : · · · : an ), (b0 : b1 )) = a0 0 · · · anβn b00 b11 .

Lemma 3.4 The diagram

ψ
Fq [X]d Bd

evd
d
ev

FqN

 d is C S (d).
is commutative and the image of ev

Proof Let Q = ((a0 : · · · : an ), (b0 : b1 )) ∈ Pn (Fq ) × P1 (Fq ) (with the points written
in standard notation) and let P = ϕ(Q) (also written in standard notation). Let Xα =
α0,0 α0,e αn,0 αn,en
X 0,0 · · · X 0,e00 · · · X n,0 · · · X n,e n ∈ Fq [X]d be a monomial of degree d, then
  α  α  α  α
Xα (P) = a0 b0e0 0,0 · · · a0 b1e0 0,e0 · · · an b0en n,0 · · · an b1en n,en
β γ γ
= a0 0 · · · anβn b00 b11 ,
ei n n ei
where βi = j=0 αi, j for all 0 ≤ i ≤ n, i=0 βi = d, γ0 = i=0 j=0 (ei − j)αi, j , and
n ei β βn γ0 γ1
γ1 = i=0 j=0 jαi, j . Hence Y0 0 · · · Yn Z 0 Z 1 ∈ Bd and
β γ γ β γ γ
a0 0 · · · anβn b00 b11 = Y0 0 · · · Ynβn Z 0 0 Z 1 1 (Q) = ψ(Xα )(Q).
This proves that the diagram is commutative and since ψ is surjective the images of evd and
 d coincide.
ev 

123
 Author's personal copy
C. Carvalho et al.

 d . As a consequence of the above result we have

Let Jd be the kernel of ev
Fq [X]d /I S (d) C S (d) Bd /Jd ,
and ψ −1 (Jd ) = I S (d). So an alternative construction of C S (d) is given, which will used
to determine its dimension by finding a basis for Bd /Jd . Clearly the classes in Bd /Jd of
the monomials in the set Md are a generating set for Bd /Jd . Considering when two such
monomials evaluate to the same value on the points Q 1 , . . . , Q N , we are led to the following

Definition 3.5 Let Jd be the Fq -vector subspace of Bd generated by the binomials of the
type
β γ γ β̃ γ̃ γ̃
Y0 0 · · · Ynβn Z 0 0 Z 1 1 − Y0 0 · · · Ynβ̃n Z 0 0 Z 1 1 ,
β β γ γ β̃ β̃ γ̃ γ̃
where Y0 0 · · · Yn n Z 0 0 Z 1 1 , Y0 0 · · · Yn n Z 0 0 Z 1 1 ∈ Md , and for all i ∈ {0, . . . , n} we have

βi ≡ β̃i (mod q − 1), γi ≡ γ̃i (mod q − 1),

βi = 0 ⇐⇒ β̃i = 0, γi = 0 ⇐⇒ γ̃i = 0.
Clearly Jd ⊂ Jd , and it will be shown that equality holds. We define a total order in the
set Md by stating that
β γ γ β β γ γ
Y0 0 · · · Ynβn Z 0 0 Z 1 1 ≺ Y0 0 · · · Yn n Z 0 0 Z 1 1 ,
if the first nonzero entry from left to right in (β0 − β 0 , . . . , βn − β n , γ0 − γ 0 ) is negative.
β β γ γ
In what follows we will denote the class of a monomial Y0 0 · · · Yn n Z 0 0 Z 1 1 in the quotient
β β γ γ
space Bd / Jd by [Y0 0 · · · Yn n Z 0 0 Z 1 1 ].
β β γ γ
Lemma 3.6 Let Y0 0 · · · Yn n Z 0 0 Z 1 1 ∈ Bd , let j := max{0 ≤ i < n : βi  = 0}. Then the least
β β γ γ β̃ β̃ γ̃ γ̃
monomial in the class is Y0 0 · · · Yn n Z 0 0 Z 1 1 where
[Y0 0 · · · Yn n Z 0 0 Z 1 1 ]
⎧
⎪ 0 if βi = 0, i ∈ {0, . . . , n},
⎪
⎪
⎨ a with 1 ≤ a ≤ q − 1, a ≡ βi (mod q − 1) if 0 ≤ i < j and βi  = 0,
β̃i = j−1
⎪
⎪
⎪d −
⎩ β̃s if i = j,
s=0
⎧
⎪
⎪0 if γ0 = 0,
⎪
⎪
⎪
⎪ 
j
⎪
⎨ b with 1 ≤ b ≤ q − 1, b ≡ γ0 (mod q − 1) if 0 < γ0 < βi ei ,
γ̃0 =
⎪
⎪
i=0
⎪
⎪ 
j 
j
⎪
⎪ β̃i ei if γ0 = βi ei ,
⎪
⎩
i=0 i=0


j
and γ̃1 := β̃i ei − γ̃0 .
i=0

Proof From the definition of β̃0 , . . . , β̃n , γ̃0 and γ̃1 , it follows that the monomial
β̃ β̃ γ̃ γ̃
Y0 0 · · · Yn n Z 0 0 Z 1 1 is in Bd and that
β γ γ β̃ γ̃ γ̃
Y0 0 · · · Ynβn Z 0 0 Z 1 1 − Y0 0 · · · Ynβ̃n Z 0 0 Z 1 1 ∈ Jd .

123
 Author's personal copy
Projective Reed–Muller type codes on higher dimensional scrolls

Also from the definitions above, it is clear that, for j + 1 ≤ i ≤ n we have β̃i = 0, because
βi = 0 for i ≥ j + 1. Moreover, for 0 ≤ i ≤ j − 1, the value of β̃i is the least possible.
β β γ γ
Thus, if Y0δ0 · · · Ynδn Z 0θ0 Z 1θ1 ∈ [Y0 0 · · · Yn n Z 0 0 Z 1 1 ] and δi > β̃i for some i ∈ {0, . . . , j − 1},
we have
β̃ γ̃ γ̃
Y0 0 · · · Ynβ̃n Z 0 0 Z 1 1 ≺ Y0δ0 · · · Ynδn Z 0θ0 Z 1θ1 .

If δi = β̃i for i ∈ {0, . . . , j − 1}, by observing that δi = 0 for i ∈ { j + 1, . . . , n},

j−1
δ j = d − s=0 β̃s = β̃ j . Since γ̃0 also has the least possible value we must have
β̃ γ̃ γ̃
Y0 0 · · · Ynβ̃n Z 0 0 Z 1 1  Y0δ0 · · · Ynδn Z 0θ0 Z 1θ1 ,
completing the proof. 

We now collect the monomials which are minimal in their classes.

β β γ γ
Definition 3.7 Let (B)d be the set of monomials Y0 0 · · · Y j j Z 0 0 Z 1 1 ∈ Bd such that
β β γ γ β β γ γ
Y0 0 · · · Y j j Z 0 0 Z 1 1 is the minimal element of [Y0 0 · · · Y j j Z 0 0 Z 1 1 ].
Observe that given a polynomial g ∈ Bd there exists a polynomial 
g in the Fq -space
g ∈ Jd .
 (B)d  generated by (B)d such that g − 
Proposition 3.8 Let h be an Fq -linear combination of elements of (B)d such that h(Q) = 0
for all Q ∈ Pn (Fq ) × P1 (Fq ). Then h = 0.

Proof From [15] we know that the vanishing ideal of Pn (Fq ) is

q q
In = Yi Y j − Yi Y j : 0 ≤ i < j ≤ n ⊂ Fq [Y0 , . . . , Yn ],
and moreover, that the set of the given generators of this ideal is a Gröbner basis for In with
respect to the lexicographic order where Y0 > · · · > Yn . From a well-known result from
Gröbner bases ([1, Prop. 6.52]) it follows that the classes in Fq [Y0 , . . . , Yn ]/In of the set
β β
(In ), consisting of all monomials of the form Y0 0 · · · Y j j , where j ∈ {0, . . . , n}, β j  = 0,
βi = 0 if i > j, and 0 ≤ βi ≤ q − 1 if 0 ≤ i < j, is a basis for Fq [Y0 , . . . , Yn ]/In as an
Fq -vector space. The set (In ) is formed by all monomials which are not multiples of any
of the leading monomials of the generators of In , and it is easy to verify that
 
β β
(In )d = Y0 0 · · · Ynβn ∈ (In ) : deg(Y0 0 · · · Ynβn ) = d ,

is a basis for the quotient Fq [Y0 , . . . , Yn ]d /In (d), where in Fq [Y0 , . . . , Yn ]d and In (d) there
are only homogeneous polynomials of degree d.
q q
Let us write I1 = Z 0 Z 1 − Z 0 Z 1  ⊂ Fq [Z 0 , Z 1 ], so that
 γ γ 
(I1 )d = Z 0 0 Z 1 1 : γ0 + γ1 = d and either γ1 = 0 or 0 ≤ γ0 ≤ q − 1 ,
n
and let ρ(β1 , . . . , βn ) = i=0 ei βi . Then from Lemma 3.6 it follows that the elements of
β β γ γ β β
(B)d are exactly the monomials of the form Y0 0 · · · Yn n Z 0 0 Z 1 1 with Y0 0 · · · Yn n ∈ (In )d
γ0 γ1
and Z 0 Z 1 ∈ (I1 )ρ(β1 ,...,βn ) .
We abbreviate the sequences (β0 , . . . , βn ) and (γ0 , γ1 ) by β and γ respectively, and also
β β γ γ
write Yβ for Y0 0 · · · Yn n and Zγ for Z 0 0 Z 1 1 . Let

h= c(β;γ ) Yβ Zγ ,
Yβ Zγ ∈ (B)d

123
 Author's personal copy
C. Carvalho et al.

be an Fq -linear combination of elements of (B)d such that h(Q) = 0 for all Q ∈ Pn (Fq ) ×
P1 (Fq ), and let (b0 : b1 ) ∈ P1 (Fq ). Then
⎛ ⎞
 ⎜  ⎟
⎜ γ γ ⎟
h (b0 :b1 ) := ⎜ c(β;γ ) b00 b11 ⎟ Yβ ,
⎝ ⎠
Yβ ∈ (In )d γ such that
Zγ ∈ (I1 )ρ(β)

vanishes at all points of Pn (Fq ), hence it is an element of In (d). As (In )d is a basis for the
quotient Fq [Y0 , . . . , Yn ]d /In (d),
 γ γ
c(β;γ ) Z 0 0 Z 1 1
γ such that
Zγ ∈ (I1 )ρ(β)

vanishes at all points of P1 (Fq ), for all β such that Yβ ∈ (In )d . Similarly, since (I1 )ρ(β)
is a basis for the quotient Fq [Z 0 , Z 1 ]d /I1 (d), c(β;γ ) = 0 whenever Zγ ∈ (I1 )ρ(β) and
Yβ ∈ (In )d , or equivalently, whenever Yβ Zγ ∈ (B)d . 

There are two important consequences of the above result.

Proposition 3.9 Jd = Jd .

Proof Clearly, Jd ⊂ Jd . Let g ∈ Jd and let g ∈  (B)d  be such that g − g ∈ Jd . Then
g(Q) = g (Q) for all Q ∈ Pn (Fq ) × P1 (Fq ), so 
g (Q) = 0 for all Q ∈ Pn (Fq ) × P1 (Fq ).
g = 0 and hence g ∈ Jd .
From the above result it follows that  

Proposition 3.10 The classes in Bd /Jd of the monomials in (B)d form a basis for Bd /Jd
as an Fq -vector space.

Proof Clearly the classes of the monomials of (B)d generate Bd /Jd as an Fq -vector space,
and from Proposition 3.8 it follows that the set of these classes is linearly independent. 

4 On the dimension of CS (d)

From Proposition 3.10 it follows that that dim C S (d) = | (B)d |, and in this section we
present some formulas to determine | (B)d |. Recall that the elements of (B)d are exactly
β β γ γ
the monomials of the form Y0 0 · · · Yn n Z 0 0 Z 1 1 where for j ∈ {0, . . . , n}, β j  = 0, βi = 0
j j
if i > j; 0 ≤ βi ≤ q − 1 if 0 ≤ i < j and i=0 βi = d, and also γ0 + γ1 = i=0 ei βi
with either γ1 = 0 or 0 ≤ γ0 ≤ q − 1. For a fixed j ∈ {0, . . . , n}, the maximal value for
j
γ0 + γ1 is e0 (d − 1) + e j , which is the value of i=0 ei βi when β0 = d, in the case j = 0,
or β0 = d − 1, β j = 1 and βi = 0 if 0 < i < j, in the case 1 ≤ j ≤ n. Indeed if β̃0 , . . . , β̃ j
are such that i=0 β̃i = d, β˜j  = 0, and β˜k  = 0 for some 0 < k < j, then β̃0 < d − 1 and
j


j 
j
e0 (β̃0 + 1) + ek (β̃k − 1) + ei β̃i − ei β̃i = e0 − ek ≥ 0
i=1,i =k i=0

123
 Author's personal copy
Projective Reed–Muller type codes on higher dimensional scrolls

j
which shows that the maximal value of i=0 ei βi is assumed when βk = 0 for 0 < k < j.
j
Similarly, it can be shown that we can assume β j = 1. We write i=0 ei βi as
⎛ ⎞
j 
j−1 
j−1 
j−1
ei βi = ei βi + e j ⎝d − βi ⎠ = m i, j βi + e j d,
i=0 i=0 i=0 i=0

where m i, j := ei − e j for 0 ≤ i < j ≤ n, and accordingly write e0 (d − 1) + e j =

(e0 − e j )(d − 1) + e j d = m 0, j (d − 1) + e j d.
For j ∈ {1, . . . , n}, let Ad ( j, s) be the number of integer solutions of the system

β0 + · · · + β j−1 + β j =d
(1)
m 0, j β0 + · · · + m j−1, j β j−1 = s

where 0 ≤ s ≤ m 0, j (d − 1) with the restrictions β j > 0 and 0 ≤ βi < q for 0 ≤ i < j, and
for j ∈ {0, . . . , n} let,
β β γ γ
(B)d, j = {Y0 0 · · · Y j j Z 0 0 Z 1 1 ∈ (B)d | β j > 0 and βi = 0 for j < i ≤ n}.
 
Observe that | (B)d | = nj=0  (B)d, j . From Lemma 3.6 it is easy to see that | (B)d,0 | =
min(q −1, e0 d −1)+2. In the following proposition a formula for | (B)d, j | when 1 ≤ j ≤ n
is presented.

Proposition 4.1 For j ∈ {1, . . . , n},

q−e j d m 0, j (d−1)
 
| (B)d, j | = (e j d + s + 1)Ad ( j, s) + (q + 1) Ad ( j, s).
s=0 s=q−e j d+1

j
Proof If β0 , . . . , β j are nonnegative integers such that i=0 β j = d, β j > 0 and 0 ≤
β β γ γ
βi < q for 0 ≤ i < j, then Y0 0 · · · Y j j Z 0 0 Z 1 1 ∈ (B)d, j for all γ0 ∈ {0, . . . , min(q −
j j j
1, i=0 ei βi − 1)} ∪ { i=0 ei βi } (with γ1 = i=0 ei βi − γ0 ).
j j−1 j
As above, we write i=0 ei βi = s + e j d, where s = i=0 m i, j βi . If i=0 ei βi > q,
(equivalently, if s > q − e j d), then there are exactly q + 1 different choices for γ0 , namely
j
{0, . . . , q − 1, s + e j d}. If i=0 ei βi ≤ q, (equivalently, if s ≤ q − e j d), then there are
j
i=0 ei βi + 1 different choices for γ0 , namely {0, . . . , s + e j d}. 

The next result presents a method to determine Ad ( j, s).

Lemma 4.2 For j ∈ {1, . . . , n} and s ∈ {0, . . . , m 0, j (d − 1)},

1 ∂ d+s g
Ad ( j, s) = (0, 0),
d!s! ∂ d x ∂ s y
where
j−1
x 1 − x q y qm i, j
g(x, y) = · .
1−x 1 − x y m i, j
i=0

123
 Author's personal copy
C. Carvalho et al.

Proof We use ideas from generating function theory to find the number of solutions of the
system (1) above. For basic results on generating functions, we refer for example to Chapter
six of [19]. Let j ∈ {1, . . . , n} and consider the series g(x, y) defined as
j−1
 
g(x, y) = (x + x 2 + · · · ) 1 + x y m i, j + · · · + (x y m i, j )q−1
i=0
j−1
x 1 − x q y qm i, j
= · .
1−x 1 − x y m i, j
i=0

For 0 ≤ i < j the exponents of x and y in the term 1 + x y m i, j + · · · + (x y m i, j )q−1 can be

thought of as representing the possible choices for the values of βi and m i, j βi , respectively,
which appear in the second equation of the system (1) for a fixed s ∈ {0, . . . , m 0, j (d − 1)}.
Similarly, the exponents of x in the term (x + x 2 + · · · ) can be though as representing the
possible choices for the value of β j , which appear in the first equation of system (1). Hence,
the coefficient of x d y s in this expression yields the number of solutions of (1) and can be
given as
1 ∂ d+s g
(0, 0).
d!s! ∂ d x ∂ s y

Now the dimension of C S (d) is determined in some cases.
q −1
Proposition 4.3 If d > , then
en
 n j  
j j + d − 1 − iq
dim(C S (d)) = (q + 1) (−1)i .
i d − 1 − iq
j=0 i=0

Proof From Proposition

  it follows that dim(C S (d)) = | (B)d | = min(q − 1, e0 d −
4.1
1) + 2 + nj=1  (B)d, j . For all j ∈ {0, . . . , n}, e j d ≥ en d > q − 1, and in particular
min(q − 1, e0 d − 1) + 2 = q + 1. When 1 ≤ j ≤ n, q − e j d < 1 and if q − e j d = 0, for
s = 0 we have e j d + s + 1 = q + 1, so from Proposition 4.1 it follows that
m 0, j (d−1)

| (B)d, j | = (q + 1) Ad ( j, s).
s=0
m (d−1)
Ad ( j, s) counts the number of integer solutions of the equation
0, j
Observe that s=0
β0 + · · · + β j = d with the restrictions β j > 0 and 0 ≤ βi < q for 0 ≤ i < j. This is
equivalent to determining the coefficient of x d in the expression
(1 + x + x 2 + · · · + x q−1 ) j (x + x 2 + · · · )
 
1 − xq j 1
=x
1−x 1−x
 j+1
1
= x(1 − x )q j
1−x
⎡ ⎤% &
j  ∞ 
⎣ j iq ⎦ k+ j k
=x (−1) i
x x
i k
i=0 k=0

123
 Author's personal copy
Projective Reed–Muller type codes on higher dimensional scrolls

which is given by


j  
j j + d − 1 − iq
(−1)i .
i d − 1 − iq
i=0

Since this summation is equal to 1 when j = 0, the result is obtained by summing up these
expressions from j = 0 to j = n and multiplying the result by q + 1. 

Proposition 4.4 If d > n(q − 1), then

dim(C S (d)) = (q + 1)(q n + q n−1 + · · · + 1).

Proof For d > n(q − 1), d ≥ n(q − 1) + 1 ≥ (q − 1) + 1 = q. Since d ≤ en d, en d > q − 1,

a similar argument as in the previous proposition gives

m 0, j (d−1)

n 
dim(C S (d)) = (q + 1) + (q + 1) Ad ( j, s).
j=1 s=0

Since d > q, the number of integer solutions of the equation β0 + · · · + β j = d with the
restrictions β j > 0 and 0 ≤ βi < q for 0 ≤ i < j is q j , therefore

m 0, j (d−1)

Ad ( j, s) = q j .
s=0

We end this section by giving some numerical examples of the parameters of projective
Reed–Muller type codes over a scroll. These examples were obtained by using the com-
putational software Magma [4]. They illustrate the case when ei  = e j when i  = j, for
i, j ∈ {0, . . . , n}. In the next section we will determine the parameters of Cs (d) for all
relevant values of d in the case when e0 = · · · = en .

Example 4.5 Consider the scroll S := S3,2,1 defined over the field with 4 elements F4 . The
following table shows the basic parameters of the code C S (d) as d varies from 1 to 7.

d Length Dimension Minimum distance

1 105 9 32
2 105 27 12
3 105 49 8
4 105 75 4
5 105 90 3
6 105 100 2
7 105 105 1

123
 Author's personal copy
C. Carvalho et al.

5 The parameters of CS (d) in a special case

Let θ be the isomorphism between the Fq -vector space M N1 ×N2 (Fq ) of N1 × N2 matrices
and FqN1 N 2 , given by
⎛ ⎞
a11 a12 · · · a1N2
⎜ a21 a22 · · · a2N2 ⎟  
⎜ ⎟
⎜ .. .. .. ⎟  → a11 , a12 , . . . , a1N2 , . . . , a N1 1 , a N1 2 , . . . , a N1 N2
⎝ . . ··· . ⎠
a N1 1 a22 · · · a N1 N2
We recall the concept of direct product of codes, as presented in [20, p. 36], slightly
modified for our purposes.
n
Definition 5.1 Let Ci ⊆ Fq i be a linear code, where n i is a nonnegative integer for i = 1, 2,
and let Mn 1 ×n 2 (Fq ) be the linear space of n 1 × n 2 matrices with entries in Fq . The direct
product of C1 and C2 , denoted by C1 ⊗C2 , is defined as the image, under θ , of all matrices
in Mn 1 ×n 2 (Fq ) the rows of which belong to C2 and the columns of which belong to C1 .
Proposition 5.2 Let Ci ⊆ Fqn i be a linear code which has dimension ki and minimum distance
δi for i = 1, 2. Then,
1. C1 ⊗C2 has length n 1 n 2 , dimension k1 k2 and minimum distance δ1 δ2 .
2. If {u i : 1 ≤ i ≤ k1 } is a basis for C1 and {v j : 1 ≤ j ≤ k2 } is a basis for C2 , then
{θ (u iT v j ) : 1 ≤ i ≤ k1 , 1 ≤ j ≤ k2 } is a basis for C1 ⊗C2 , where u iT is an n 1 × 1 matrix
and v j is a 1 × n 2 matrix.
Proof The first part is proved in [20], Theorems 2.5.2 and 2.5.3. The second part is a conse-
quence of Lemma 2.3 in [18]. 
These results will be used to determine the true minimum distance of C S (d) in a special
case (cf. [5, Thm. 5.8]). Before doing so, we briefly recall the basic data on projective Reed–
Muller codes. Let d and m be positive integers and let Nm = q m + · · · + q + 1. Denoting the
points of Pm (Fq ) by U1 , . . . , U Nm , we recall that the projective Reed–Muller code (of order
d, over Pm (Fq )) is the image PRM(d, m) of the evaluation morphism

ϕd,m : Fq [Y0 , . . . , Ym ]d →  FqNm  

f  → f (U1 ), . . . , f U Nm ,
defined over the space of homogeneous polynomials of degree d, and where, for evaluation
purposes, the points of Pm (Fq ) are written in standard notation. From [17] (see also [15])
we know that if d ≥ m(q − 1) then ϕd,m is surjective, and that for 1 ≤ d ≤ m(q − 1) the
dimension and minimum distance δPRM (d, m) of PRM(d, m) are as follows:
m  j  
j j + d − 1 − iq
dim(PRM(d, m)) = (−1)i ,
i d − 1 − iq
j=0 i=0

δPRM (d, m) = (q − r )q m−k−1 ,

a  d − 1 = k(q − 1) + r and 0 ≤ r < q − 1. In the formula for the dimension we assume

where
b = 0 if b < 0, so for example, if d ≤ q then
m  
j +d −1 d +m
dim(PRM(d, m)) = = ,
d −1 d
j=0

123
 Author's personal copy
Projective Reed–Muller type codes on higher dimensional scrolls

a formula first established by Lachaud (see [14]).

Let d1 = dim(PRM(d, m)). From [17] we also know that there are monomials
Yβ 1 , . . . , Yβ d1 in Fq [Y0 , . . . , Ym ]d such that {ϕd,m (Yβ 1 ), . . . , ϕd,m (Yβ d1 )} is a basis for
PRM(d, m).
Proposition 5.3 Assume e0 = e1 = · · · = en = e. Then,
C S (d) ∼
= PRM(d, n) ⊗ PRM(de, 1).
Proof Let N = (q +1)(q n +· · ·+q +1). We know from Lemma 3.4 that C S (d) = ev  d (Bd ) ⊂
β β γ γ n
FqN , so C S (d) is generated by the vectors ev
 d (Y0 0 · · · Yn n Z 0 0 Z 1 1 ) where i=0 βi = d and,
since e0 = e1 = · · · = en = e, γ0 + γ1 = de. As mentioned above, from [17] we know that,
setting d1 = dim(PRM(d, n)), there are monomials Yβ 1 , . . . , Yβ d1 in Fq [Y0 , . . . , Yn ]d and
elements h 1 , . . . , h d1 ∈ Fq such that

β

d1
Y0 0 · · · Ynβn (a0 , . . . , an ) = h i Yβ i (a0 , . . . , an ),
i=1

for all (a0 : · · · : an ) ∈ Pn (Fq ). In the same way, setting d2 = dim(PRM(de, 1)) monomials
Zγ 1 , . . . , Zγ d2 exists in Fq [Z 0 , Z 1 ]de , and elements g1 , . . . , gd2 ∈ Fq such that

γ γ

d2
Z 0 0 Z 1 1 (b0 , b1 ) = g j Zγ j (b0 , b1 ),
j=1

so C S (d) is generated by the vectors ev

d (Yβ i Zγ j )
where i = 1, . . . , d1 and j = 1, . . . , d2 .
On the other hand, from Proposition 5.2 and the data on projective Reed–Muller codes,
PRM(d, n) ⊗ PRM(de, 1) ⊂ P Nn N1 (Fq ), where Nn = q n + · · · + q + 1, N1 = q + 1, and
{θ (ϕd,n (Yβ i )T ϕde,1 (Zγ j )) : i = 1, . . . , d1 ; j = 1, . . . , d2 },
is a basis for PRM(d, n) ⊗ PRM(de, 1).
 d (Yβ i Zγ j ) or
For i ∈ {1, . . . , d1 } and j ∈ {1, . . . , d2 } the entries in either vector ev
β T γj
θ (ϕd,n (Y ) ϕde,1 (Z )) are of the same form, namely
i

β γ γ
a0 0 · · · anβn b00 b11 ,
each entry carrying the data of one of the (q n +· · ·+q+1)(q+1) rational points of the cartesian
product Pn (Fq ) × P1 (Fq ), the only difference between the vectors being (possibly) on the
order under which the points appear. This proves that C S (d) ∼ = PRM(d, n) ⊗ PRM(de, 1).

We denote by δ(C S (d)) the minimum distance of C S (d).
Corollary 5.4 Assume that e0 = e1 = · · · = en = e, then the dimension and the minimum
distance of C S (d) are as follows:
If d ≥ n(q − 1) then,
dim(C S (d)) = (q n + · · · + q + 1)(q + 1) and δ(C S (d)) = 1.
If (q − 1)/e < d ≤ n(q − 1) then,

n 
j  
i j j + d − 1 − iq
dim(C S (d)) = (q + 1) (−1) and
i d − 1 − iq
j=0 i=0

δ(C S (d)) = (q − r )q n−k−1 , where d − 1 = k(q − 1) + r and 0 ≤ r < q − 1.

123
 Author's personal copy
C. Carvalho et al.

If 1 ≤ d ≤ (q − 1)/e then,

n+d
dim(C S (d)) =(de + 1) and
d
δ(C S (d)) =(q − d + 1)(q − de + 1)q n−1 .
Proof This is a consequence of the two Propositions above and the data on the projective
Reed–Muller codes. We note that if d ≥ n(q − 1) then de ≥ (q − 1) so PRM(d, n) =
q n +···+q+1 q+1
Fq and PRM(de, 1) = Fq . If (q − 1)/e < d ≤ n(q − 1), de > q − 1 so
q+1
PRM(de, 1) = Fq . And if 1 ≤ d ≤ (q − 1)/e, then de < q − 1 and a fortiori d < q − 1,
so Lachaud’s formula can be applied for the dimension of PRM(d, n) and PRM(de, 1). 

Acknowledgements The second author is grateful to Prof. F. Zaldivar for pointing out the concept of a higher
dimensional scroll. We thank the referees for a careful reading and their comments which improved the
manuscript.

References
1. Becker T., Weispfenning W.: Gröbner Bases. Springer, New York (1993).
2. Beelen, P., Datta, M., Ghorpade, S.R.: A combinatorial approach to the number of solutions of systems
of homogeneous polynomial equations over finite fields. arXiv:1807.01683v2 (2018)
3. Beelen P., Datta M., Ghorpade S.R.: Maximum number of common zeros of homogeneous polynomials
over finite fields. Proc. Am. Math. Soc. 146(4), 1451–1468 (2018).
4. Bosma W., Cannon J., Playoust C.: The Magma algebra system. I. The user language. J. Symb. Comput.
24, 235–265 (1997).
5. Carvalho C., Neumann V.G.L.: Projective Reed-Muller type codes on rational normal scrolls. Finite Fields
Appl. 37, 85–107 (2016).
6. Carvalho C., Neumann V.G.L., Lopez H.: Projective nested cartesian codes. Bull. Braz. Math. Soc. 48,
283–302 (2017).
7. Couvreur A., Duursma I.: Evaluation codes from smooth quadric surfaces and twisted Segre varieties.
Des. Codes Cryptogr. 66, 291–303 (2013).
8. Cox D., Little J., O’Shea D.: Ideals, Varieties, and Algorithms, 3rd edn. Springer, New York (2007).
9. Duursma I., Rentería C., Tapia-Recillas H.: Reed Muller codes on complete intersections. Appl. Algebra
Eng. Commun. Comput. (AAECC) 11, 455–462 (2001).
10. González-Sarabia, M., Martínez-Bernal, J., Villarreal, R.H., Vívares, C.E.: Generalized minimum distance
functions. J. Algebr. Comb. https://doi.org/10.1007/s10801-018-0855-x (2018)
11. González-Sarabia M., Rentería C., Tapia Recillas H.: Reed-Muller type codes over the segre variety.
Finite Fields Appl. 8, 511–518 (2002).
12. González-Sarabia M., Rentería C.: The dual code of some Reed-Muller-type codes. Appl. Algebra Eng.
Commun. Comput. (AAECC) 14, 329–333 (2004).
13. Harris, J.: Algebraic Geometry: A First Course, 3rd. edn, Springer, GTM, no. 133 (1995)
14. Lachaud G.: Projective Reed-Muller codes. In: Cohen G., Godlewski P. (eds.) Coding Theory and Appli-
cations. Lecture Notes in Computer Science, vol. 311. Springer, Berlin (1988).
15. Rentería C., Tapia-Recillas H.: Reed-Muller codes: an ideal theory approach. Commun. Algebra 25(2),
401–413 (1997).
16. Rentería, C., Tapia-Recillas, H.: Reed-Muller type codes on the Veronese Variety over finite fields. In:
Buchmann, J., Hoholdt, T., Stichtenoth, H., Tapia-Recillas, H. (eds.) Coding Theory, Cryptography and
Related Areas, ISBN 3-540-66248-0, pp. 237–243 , Springer, New York (2000)
17. Sörensen A.B.: Projective Reed-Muller codes. IEEE Trans. Inform. Theory 37(6), 1567–1576 (1991).
18. Tochimani A., Pinto M.V., Villarreal R.H.: Direct products in projective Segre codes. Finite Fields Appl.
39, 96–110 (2016).
19. Tucker A.: Applied Combinatorics, 6th edn. Wiley, New York (2012).
20. van Lint, J.H.: Coding Theory, Lect. Notes Math., 2nd edn, vol. 201. Springer, Berlin (1973)

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

123