Inverses of r-primitive k-normal elements over finite

fields
arXiv:2201.11334v1 [math.NT] 27 Jan 2022

Mamta Rania , Avnish K. Sharmaa , Sharwan K. Tiwarib,∗, Anupama

Panigrahia
a
Department of Mathematics, University of Delhi, New Delhi, 110007, India
b
Scientific Analysis Group, Defence Research & Development Organization, Metcalfe
House, Delhi, 110054, India

Abstract
Let r, n be positive integers, k be a non-negative integer and q be any prime
power such that r | q n − 1. An element α of the finite field Fqn is called an
r-primitive element, if its multiplicative order is (q n − 1)/r, and it is called a
k-normal element over Fq , if the greatest common divisor of the polynomials
P
mα (x) = ni=1 αq xn−i and xn −1 is of degree k. In this article, we define the
i−1

characteristic function for the set of k-normal elements, and with the help
of this, we establish a sufficient condition for the existence of an element
α in Fqn , such that α and α−1 both are simultaneously r-primitive and k-
normal over Fq . Moreover, for n > 6k, we show that there always exists an
r-primitive and k-normal element α such that α−1 is also r-primitive and
k-normal in all but finitely many fields Fqn over Fq , where q and n are such
that r | q n − 1 and there exists a k-degree polynomial g(x) | xn − 1 over Fq .
In particular, we discuss the existence of an element α in Fqn such that α
and α−1 both are simultaneously 1-primitive and 1-normal over Fq .
Keywords: Finite fields, r-Primitive elements, k-Normal elements, Additive
and multiplicative characters.
2020 MSC: 12E20, 11T23.

Corresponding author
∗

Email addresses: mamta11singla@gmail.com (Mamta Rani), avkush94@gmail.com

(Avnish K. Sharma), shrawant@gmail.com (Sharwan K. Tiwari),
anupama.panigrahi@gmail.com (Anupama Panigrahi)

Preprint submitted to Elsevier January 28, 2022

1. Introduction
Let Fqn be the field extension of degree n over Fq , where q be a prime
power and n ∈ N. We recall that, the multiplicative group F∗qn is cyclic, and
an element α ∈ F∗qn is called primitive, if its multiplicative order is q n − 1.
Let r be a divisor of q n − 1, then an element α ∈ F∗qn is called r-primitive,
if its multiplicative order is (q n − 1)/r. Clearly, a 1-primitive element in
Fqn is actually a primitive element. Moreover, the existence of r-primitive
elements is obvious, because the multiplicative order of the r th power of a
primitive element is (q n − 1)/r. Primitive elements have many applications
in the field of cryptography [5, 12, 2] due to their highest multiplicative
order. Moreover, r-primitive elements can also be considered as high ordered
elements for small values of r. Therefore, in many applications, r-primitive
elements may replace primitive elements.
Again, recall that, the finite field Fqn is also an Fq -vector space, and an
n−1
element α ∈ Fqn is said to be normal over Fq , if the set {α, αq , . . . , αq }
forms a basis (called normal basis) of Fqn over Fq . Normal bases have many
applications in the computational theory due to their efficient implementa-
tion in finite field arithmetic. For more detail on normal elements one may
refer [7] and the references therein.
From [11, Theorem 2.39], we know that an element α ∈ Fqn is normal over
n−2
Fq if and only if the polynomials mα (x) = αxn−1 + αq xn−2 + . . . + αq x +
n−1
αq and xn − 1 are relatively prime in Fqn [x]. Motivated by this definition,
Huczynska et al. [8] introduced the concept of k-normal elements and gave
the following definition.

Definition 1.1. An element α ∈ Fqn is said to be k-normal over Fq , if the

n−2 n−1
polynomials mα (x) = αxn−1 + αq xn−2 + . . . + αq x + αq and xn − 1 has
greatest common divisor of degree k over Fq .

From the above definition, it is clear that 0 ≤ k ≤ n − 1. We call an element

α ∈ Fqn , r-primitive k-normal, if it is both r-primitive and k-normal over
Fq . From the above definition, if k = 0, then α is normal over Fq . Moreover,
Huczynska et al. [8] characterized the k-normal elements in Fqn over Fq and
established bounds for the number of such elements, and proved the existence
of an element α which is both 1-primitive and 1-normal over Fq for n ≥ 6 if
q ≥ 11, and for n ≥ 3 if 3 ≤ q ≤ 9, where gcd(q, n) = 1. In 1987, Lenstra
and Schoof [10] proved the existence of primitive normal elements (i.e. 1-
primitive 0-normal elements) for arbitrary finite fields. In 2014, Anderson

2
and Mullen [13] conjectured that, for a prime p ≥ 5 and an integer n ≥ 3,
there always exists an element α ∈ Fpn of order (pn − 1)/r, which is k-normal
for (r, k) = (1, 1), (2, 0), (2, 1). In 2018, Reis and Thomson in [17] settled
their conjecture for (r, k) = (1, 1), for arbitrary q and n ≥ 3. Finally, in
2019, Kapetanakis and Reis [9] completely settled the above conjecture for
arbitrary prime power q.
Recently, Aguirre and Neumann [1] showed that, there exists a 1-primitive
2-normal element in Fqn over Fq if and only if n ≥ 5 and gcd(q 3 − q, n) 6= 1
or n = 4 and q ≡ 1 (mod 4). In this direction, we [14] generalized the above
existence problems to arbitrary r and k, and established a sufficient condition
for the existence of an element α in Fqn which is simultaneously r-primitive
and k-normal over Fq . Moreover, in particular, we gave the following complete
existence of 2-primitive 2-normal elements.
Theorem 1.1. Let q be a prime power and n be a positive integer. Then,
there always exists a 2-primitive 2-normal element in Fqn over Fq if and only
if q is odd, and either n ≥ 5 and gcd(q 3 −q, n) 6= 1 or n = 4 and q ≡ 1(mod 4).
Notice that, the inverse of an r-primitive element is also r-primitive in
Fqn for r | q n − 1. But this may not be true in case of k-normal elements.
For example, if k = 0, then there does not exist any normal element α ∈ F8
over F2 for which α−1 is also normal element in F23 over F2 . In [20], Tian
and Qi showed the existence of a primitive normal element α ∈ Fqn over Fq
whose inverse is also primitive normal over Fq for all q and n ≥ 32. Later,
Cohen and Huczynska [3] completely settled this problem by using sieving
technique and gave the following result.
Theorem 1.2 (Strong Primitive Normal Basis Theorem). There exists a
(1−) primitive (0−) normal element α ∈ Fqn ; n ≥ 2, such that α−1 is
also a primitive normal element over Fq unless (q, n) is one of the pair
(2, 3), (2, 4), (3, 4), (4, 3), (5, 4).
Motivated by the above theorem, in this article, we discuss the existence
of a pair (α, α−1) of r-primitive k-normal elements in Fqn over Fq . For this,
we first define the characteristic function in Section 3 for the set of k-normal
elements, which plays a crucial role in establishing a sufficient condition for
the existence of such pairs in Section 4. Moreover, we conclude that, for
n > 6k, there exists a pair (α, α−1 ) of r-primitive k-normal elements in all
but finitely many fields Fqn over Fq . In particular, we gave the following result
in Section 5.

3
Theorem 1.3. Let q be a prime power and n ≥ 5 be an integer. Then
(i) For n ≥ 7 and q ≥ 2, there always exists a pair (α, α−1) of primitive
1-normal elements in Fqn over Fq .
(ii) For n = 5, 6 and q ≥ 2 such that gcd(q, n) = 1, there always exists
a pair (α, α−1) of primitive 1-normal elements in Fqn over Fq with the
sole genuine exception (4, 5).

2. Preliminaries
In this section, we recall some definitions and lemmas that are crucial for
proving the results of this article.

2.1. Freeness
We begin with the following definition.
Definition 2.1 (e-free element). Let e | q n − 1 and β ∈ Fqn . Then β is said
to be e-free, if β is not a dth power in Fqn for any divisor d > 1 of e. In other
q n −1
words, β is e-free if and only if e and ord(β) are co-prime, where ord(β) is
the multiplicative order of β in F∗qn . Clearly, an element β ∈ Fqn is primitive
if and only if it is (q n − 1)-free.
P
Let h(x) = m i
i=0 ai x ∈ Fq [x] with am 6= 0. Then,Pthe additive group Fq n
forms an Fq [x]- module under the action h ◦ β = m qi
i=0 ai β . Notice that,
(xn − 1) ◦ β = 0 for all β ∈ Fqn , which leads to the following definition.
Definition 2.2. The Fq -order (denoted by Ordq (·)) of an element β ∈ Fqn ,
is the least degree monic divisor h(x) of xn − 1 such that h ◦ β = 0.
Similar to the e-free elements, we have the following definition for f -free
elements.
Definition 2.3. Let h(x) | xn − 1 and β ∈ Fqn . Then, β is called h-free, if
β 6= g ◦ γ for any γ ∈ Fqn and for any divisor g(x) of h(x) of positive degree.
xn −1
In other words, β is h-free if and only if h(x) and Ord q (β)
are co-prime.
n
Clearly, an element β ∈ Fqn is normal if and only if it is (x − 1)-free.
In this article, we will use the following definition of k-normal elements.
Definition 2.4. [8, Theorem 3.2] An element β ∈ Fqn is k-normal over Fq ,
if and only if the Fq -order of β is of degree n − k.

4
The following lemma provides a way to construct the k-normal elements from
a given normal element.

Lemma 2.1. [16, Lemma 3.1] Let β ∈ Fqn be a normal element over Fq
and g ∈ Fq [x] be a polynomial of degree k such that g divides xn − 1. Then
α = g ◦ β is k-normal.

2.2. Characters
Let G be a finite abelian group. A character χ of G is a homomorphism
from G into the multiplicative group of complex numbers of unit modulus.
The set of all such characters of G, denoted by G, b forms a multiplicative
group and G ∼ b A character χ is called the trivial character if χ(α) = 1
= G.
for all α ∈ G, otherwise it is a non-trivial character.

Lemma 2.2. [11, Theorem 5.4] Let χ be any non-trivial character of a finite
abelian group G and α ∈ G be any non-trivial element, then
X X
χ(α) = 0 and χ(α) = 0.
α∈G b
χ∈G

Let ψ denote the additive character for the additive group Fqn , and χ
denote the multiplicative character for the multiplicative group F∗qn . The
additive character ψ0 defined by ψ0 (β) = e2πiTr(β)/p , for all β ∈ Fqn , where p
is the characteristic of Fqn and Tr is the absolute trace function from Fqn to
Fp , is called the canonical additive character of Fqn . Moreover, every additive
character ψβ for β ∈ Fqn can be expressed in terms of the canonical additive
character ψ0 as ψβ (γ) = ψ0 (βγ), for all γ ∈ Fqn .
For any ψ ∈ F b qn is an Fq [x]-module under
bqn , α ∈ Fqn and g(x) ∈ Fq [x], F
the action ψ ◦ g(α) = ψ(g ◦ α). The Fq -order of an additive character ψ ∈ F bqn ,
denoted by Ordq (ψ), is the least degree monic divisor g(x) of xn −1 such that
ψ ◦ g is the trivial character and there are precisely ΦqP (g) characters of Fq -
∗
order g(x), where Φq (g) = |(Fq [x]/< g >) |. Moreover, h|g Φq (h) = q deg(g) .
We shall need the following lemma to prove our sufficient condition.
Q
Lemma 2.3. [6, Theorem 5.5] Let F (x) = kj=1 fj (x)nj be a rational function
in Fqn (x), where fj (x) ∈ Fqn [x] are irreducible polynomials and nj are non-
zero integers. Let χ be a multiplicative character of order d of F∗qn . Suppose

5
that the rational function F (x) is not of the form L(x)d for any L(x) ∈ F(x),
where F is the algebraic closure of Fqn . Then we have
  !
 X  X k
 
 χ(F (α)) ≤ deg(fj ) − 1 q n/2 .
 
α∈Fqn j=1
F (α)6=0,∞

Lemma 2.4. [6, Theorem 5.6] Let F (x), G(x) ∈ Fqn (x) be rational func-
Q
tions. Write F (x) = kj=1 fj (x)nj , where fj (x) ∈ Fqn [x] are irreducible
Pk
polynomials and nj are non-zero integers. Let d1 = j=1 deg(fj (x)),
d2 = max{deg(G(x)), 0}, d3 be the degree of the denominator of G(x) and
d4 be the sum of degrees of those irreducible polynomials dividing the denom-
inator of G(x), but distinct from fj ; 1 ≤ j ≤ k. Let χ be the multiplicative
character of F∗qn and ψ be the non-trivial additive character of Fqn . Suppose
n
that G(x) 6= L(x)q − L(x) in F(x), where F is the algebraic closure of Fqn .
Then we have
 
 X 
 
 χ(F (α))ψ(G(α)) ≤ (d1 + d2 + d3 + d4 − 1)q n/2 .
 
α∈Fqn ,F (α)6=0,∞
G(α)6=∞

2.3. Some characteristic functions

For the set of e-free elements of F∗qn , we have the following characteristic
function ρe : F∗qn → {0, 1}.
φ(e) X µ(d) X
ρe (β) = χd (β), (1)
e φ(d) χ
d|e d

where µ is the Möbius function and the internal sum runs over all the mul-
tiplicative characters χd of order d. Similar to the characteristic function
for the set of e-free elements, we have the following characteristic function
Υg : Fqn → {0, 1} for the set of g-free elements of Fqn .
Φq (g) X µ′ (h) X
Υg (α) = deg(g) ψh (α), (2)
q Φq (h) ψ
h|g h

where µ′ is the analog of the Möbius function, which is defined as µ′ (h) =

(−1)t , if h(x) is a product of t distinct monic irreducible polynomials, oth-
erwise 0, and the internal sum runs over the additive characters ψh of the
Fq -order h(x).

6
3. Characterization of r-primitive and k-normal elements
3.1. Characteristic function for the set of r-primitive elements
Let r | q n − 1 be a positive integer, and write r = upb11 pb22 · · · pbss , where u
and (q n − 1)/u are co-prime and pj ’s are distinct primes such that bj ≥ 1 and
b +1
pjj | q n − 1, for all j = 1, 2, . . . , s. In this article, we denote the product
of distinct irreducible factors of an integer or a polynomial m by rad(m).
b b +1
Now, let R = rad(q n − 1)/rad(r), and set δj := pjj and λj := pjj for
all j = 1, 2, . . . , s. In [4], Cohen and Kapetanakis showed that an element
α ∈ Fqn is r-primitive if and only if α is R-free, α = β r for some β ∈ Fqn , but
α 6= β λj for any β ∈ Fqn and any j = 1, 2, . . . , s, and using this fact, they
defined the characteristic function for r-primitive elements. In [14], following
them, we defined the following characteristic function Γdr for the set Qdr of
elements α ∈ Fqn such that α is d-free for some divisor d of R, α = β r for
some β ∈ Fqn , but α 6= β λj for any β ∈ Fqn and any j = 1, 2, . . . , s.

φ(d) X X µ(d1 )  Y
s  X
Γdr (α) = ℓp ,e (χd1 χd2 χe1 · · · χes )(α),
rd φ(d1 ) j=1 j j χd1 ,χd2 ,
d1 |d,d2 |u ej |λj ,
1≤j≤s χe1 ,...,χes

(3)

1 − 1/pj if ej 6= λj ,
where ℓpj ,ej = . In particular, for d = R, the above
−1/pj if ej = λj
characteristic function provides the characteristic function given by Cohen
and Kapetanakis [4] for the set QR r of r-primitive elements.

3.2. Characteristic function for the set of k-normal elements

For the existence of those elements in a finite field Fqn that simultaneously
satisfy certain properties, such as primitivity, normality, trace, norm, etc., a
general approach is to use the characteristic functions for the set of elements
with those properties. In this article, we also use this approach to discuss
the existence of a pair (α, α−1) of r-primitive k-normal elements in Fqn over
Fq . For this, we shall need the characteristic function for the set of k-normal
elements that we define in this subsection.
Let Sk be the collection of all k-normal elements in Fqn over Fq and
Pk be the collection of polynomials of degree k that divides xn − 1. By
Lemma 2.1, if β is a normal element of Fqn over Fq , then for any polynomial
g ∈ Pk , g ◦ β is k-normal over Fq . So, for every g ∈ Pk , we consider a set

7
Sg,k := {α ∈ Sk | α = g ◦ β for some element β ∈ Fqn } of k-normal elements.
We have the following lemma.
′
T
Lemma S 3.1. If g, g are two distinct polynomials in P k , then S g,k Sg′ ,k = ∅,
and g∈Pk Sg,k = Sk .
T
Proof. Suppose that, Sg,k Sg′ ,k 6= ∅, then there exist elements β and β ′ in
Fqn such that g ◦ β = g ′ ◦ β ′ , and by uniqueness
T of the Fq -order of an element,
′
we get g = gS, a contradiction. Hence, Sg,k Sg′ ,k = ∅.
Clearly, g∈Pk Sg,k ⊆ Sk . Now, let α ∈ Sk , then α = h ◦ β, for some
normal element β and a polynomial h ∈ Fq [x] of degree at most n − 1. Let
f be the Fq -order of α, then f ◦ α = f h ◦ β = 0, and hence, xn − 1 | f h,
n
i.e. h = (xn − 1)h′ /f , for some h′ ∈ Fq [x]. Thus, α = x f−1 ◦ (h′ ◦ β), which
n
implies α = g ◦ γ ∈ Sg,k , where g = x f−1 ∈ Pk and γ = h′ ◦ β ∈ Fqn . Hence,
S
g∈Pk Sg,k = Sk .

From the above lemma, it is clear that, the set {Sg,k | g ∈ Pk } provides
a partition of Sk . Therefore, the characteristic function for the set Sk of
k-normal elements is equal to the sum of the characteristic functions for the
sets Sg,k ; g ∈ Pk . Hence, it is enough to define the characteristic function for
the set Sg,k .
Let g ∈ Pk be an arbitrary but fixed polynomial. Write g =
b1 b2
πf1 f2 · · · ftbt , where π and (xn −1)/π are co-prime, and fi ’s are distinct irre-
ducible polynomials such that bi ≥ 1 and fibi +1 | xn − 1, for all i = 1, 2, . . . , t.
Further, let G = rad(xn − 1)/rad(g), and set ∆i := fibi and Λi := fibi +1 for
all i = 1, 2, . . . , t. Now, we prove the following lemma.

Lemma 3.2. Let k be a non-negative integer and g ∈ Pk . Then, the following

are equivalent.

(i) α ∈ Sg,k .

(ii) α is G-free, α = g ◦ β for some β ∈ Fqn , but α 6= Λi ◦ β for any β ∈ Fqn

and for any i = 1, 2, . . . , t.

Proof. First, let α ∈ Sg,k . Then by definition of Sg,k , α = g ◦ β for some

β ∈ Fqn , and F
q -order ofα is (xn − 1)/g. Notice that, G and g are co-prime,
xn −1
therefore, gcd G, Ord q (α)
= 1, which implies α is G-free. Now, suppose that
xn −1
α = Λi ◦ β for some β ∈ Fqn and some i = 1, 2, . . . , t. Then, Λi
◦ α = 0,

8
 n n
which implies x g−1 | x Λ−1i
, i.e. Λi | g, which is not possible. Hence, (i) implies
(ii).
Now, let α is G-free, α = g ◦ β for some β ∈ Fqn , but α 6= Λi ◦ β for any
β ∈ Fqn and for any i = 1, 2, . . . , t. Then, α = g ◦ (h ◦ γ) for some normal
element γ ∈ Fqn and a polynomial h ∈ Fq [x] of degree at most n − 1, and
xn −1 ′
Ordq (α) = g gcd(h,(x n −1)/g) . Suppose that, h be an irreducible factor of the
n
gcd(h, (xn − 1)/g), then h′ | h and h′ | x g−1 . This implies, either h′ | G or
n
Λi | gh. In any case, we get a contradiction. This means Ordq (α) = x g−1 , i.e.
α is a k-normal element. Hence, (ii) implies (i).
Now, we define the characteristic function Ψg for the set of elements of
the form g ◦ β. Consider the set Mg of the elements of the form g ◦ β, where
g | xn − 1. Clearly, Mg is a subgroup of the additive group Fqn . Let A ⊆ F bqn
bqn
be the annihilator of Mg , i.e. the collection of all additive characters ψ ∈ F
such that ψ(α) = 1 for all α ∈ Mg . Clearly, the set A consists of all the
additive characters of Fq -order h dividing g, and from [11, Theorem 5.6], it
is isomorphic to the group of the characters of the quotient group Fqn /Mg ,
= (F\
i.e. A ∼ \
q n /Mg ). Therefore, we can define a character Yh on (Fq n /Mg ) by
Yh (α + Mg ) = ψh (α), where ψh ∈ A. Then,
XX  deg(g)
q , if α ∈ Mg ,
Yh (α + Mg ) = .
Y
0 , otherwise
h|g h

Thus, we can define a characteristic function Ψg : Fqn → {0, 1} for the set
Mg as follows.
1 XX
Ψg (α) := deg(g) ψh (α). (4)
q ψ h|g h

Now, we are ready to define a characteristic function QG g for the set Sg,k .
From the characteristic functions (2) and (4) and from Lemma 3.2, we have
t 
Y 
QGg (α) = Υ G (α)Ψ g (α) 1 − Ψ Λi (α) .
i=1
Qt
Since, g = π Q i=1 ∆i , and π, ∆i ; i = 1, 2, . . . , t are mutually co-prime,
Ψg (α) = Ψπ (α) ti=1 Ψ∆i (α), and hence,
t
Y  
QG
g (α) = ΥG (α)Ψπ (α) Ψ∆i (α) 1 − ΨΛi (α) .
i=1

9
Moreover, if α = Λi ◦ β for some β, then α = ∆i ◦ γ for some γ in Fqn , and
Ψ∆i ΨΛi = ΨΛi for all i = 1, 2, . . . , t. Hence,
t 
Y 
QG
g (α) = ΥG (α)Ψπ (α) Ψ∆i (α) − ΨΛi (α) . (5)
i=1

Now,
1 XX 1 XX
Ψ∆i (α) − ΨΛi (α) = ψh (α) − ψh (α)
q deg(∆i ) q deg(Λi )
h|∆i ψh h|Λi ψh
1 XX
= ℓ′fi ,h ψh (α),
q deg(∆i )
h|Λi ψh


1 − 1/q deg(fi ) if h 6= Λi ,
where ℓ′fi ,h = . Notice that, |ℓ′fi ,h |≤ ℓ′fi ,1 for all
−1/q deg(fi ) if h = Λi
i = 1, 2, . . . , t. Now, substituting the values of ΥG (α), Ψπ (α) and Ψ∆i (α) −
ΨΛi (α) in Equation (5), we get

Φq (G) X X µ′ (g1 )  Y
t  X
QG
g (α) = ℓ′fi ,hi (ψg1 ψg2 ψh1 · · · ψht )(α).
q deg(G)+deg(g) Φq (g1 )
i=1 ψg1 ,ψg2 ,
g1 |G, hi |Λi ,
g2 |π 1≤i≤t ψhi ; 1≤i≤t
(6)
Thus, we define P the characteristic function Q for the set Sk of all k-normal
elements as Q = g∈Pk QG g.
H
Now, for a divisor H of G, we define a collection Tg,k of elements α of Fqn
such that α is H-free, α = g ◦ β for some β ∈ Fqn , but α 6= Λi ◦ β for any
β ∈ Fqn and for any i = 1, 2, . . . , t. We define the characteristic function QH
g
H
for the set Tg,k similar to that for the set Sg,k as follows

Φq (H) X X µ′ (g1 )  Y
t  X
QH
g (α) = ℓ′fi ,hi (ψg1 ψg2 ψh1 · · · ψht )(α).
q deg(H)+deg(g) Φq (g1 )
i=1 ψg1 ,ψg2 ,
g1 |H, hi |Λi ,
g2 |π 1≤i≤t ψhi ; 1≤i≤t
(7)
H
Observe that, in particular, if H = G, then = Sg,k and =Tg,k QH
g QG
g.
Henceforth, the notation used in this section shall have the same meaning
throughout the article unless otherwise stated.

10
4. Existence of an r-primitive k-normal pair (α, α−1 )
In this section, our aim is to show the existence of a pair (α, α−1) of
r-primitive k-normal elements in Fqn over Fq . From Lemma 2.1, it is clear
that, if γ ∈ Fqn is a normal element and g ∈ Pk , then α = g ◦ γ is a k-normal
element over Fq . Therefore, for the existence of a desired pair, we only need
to check the r-primitivity of α and the k-normality of α−1 . This means for
the existence of a desired pair, it is enough to find an (xn − 1)-free element
γ, such that α = g ◦ γ ∈ QR r and α
−1 G
∈ Tg,k , for some arbitrary but fixed
polynomial g ∈ Pk . Instead of this, more generally, we will show the existence
of an h-free element γ, for some h | xn − 1, such that α = g ◦ γ ∈ Qdr and
α−1 ∈ Tg,kH
, where d | R and H | G. Let Nr,k,g (h, d, H) denote the number of
such elements γ, then we prove the following result.
Theorem 4.1. Let r > 0 and k ≥ 0 be integers. Further, let q be a prime
power and n be a positive integer such that r | q n − 1 and there exists a
polynomial g ∈ Pk . Then, Nr,k,g (h, d, H) > 0, if

q n/2−ϑ > 2r rad(r)W (h)W (d)W (H),

where, ϑ = 2k, if gcd(q, n) = 1, and ϑ = 3k, otherwise.

P P
Proof. Let g = ki=0 ai xi ∈ Pk with ak = 1, and define g ◦ x = ki=0 ai xq .
i

Further, let Z be the set of zeroes of the polynomial g ◦ x. Clearly,

Z consists
P of all the elements β, whose Fq -order divides g, and hence,
′ k
|Z|= g ′ |g q (g ) = q . Now, by using the definitions of the characteris-
Φ
tic functions Υh , Γdr and QH g , we have
X
Nr,k,g (h, d, H) = Υh (β)Γdr (g ◦ β)QH −1
g ((g ◦ β) )
β∈Fqn \Z
X X X µ ′  X
=H (h , d1 , g1 )ℓ (ej )sj=1 , (hi )ti=1 ×
Φ
h′ |h d1 |d,d2 |u, g1 |H,g2 |π, ψh′
ej |λj ;1≤j≤s hi ,|Λi ;1≤i≤t
X X X
ψh′ (β)χD (g ◦ β)ψF ((g ◦ β)−1 ),
χd1 ,χd2 , ψg ,ψg , β∈F n \Z
1 2 q
χej ;1≤j≤s ψh ; 1≤i≤t
i
(8)
Φq (h)φ(d)Φq (H) µ ′ µ′ (h′ )µ(d1 )µ′ (g1 )
where H = deg(h)+deg(H)+deg(g)
, (h , d ,
1 1g ) = ′ )φ(d )Φ (g )
,
 rdq
 Q Φ Φ q (h 1 q 1
Q
ℓ (ej )sj=1 , (hi )ti=1 = sj=1 ℓpj ,ej ti=1 ℓ′fi ,hi , χD = χd1 χd2 χe1 · · · χes is a multi-

11
plicative character of order D = d1 d2 e1 · · · es , since d1 , d2, e1 , . . . , es are mutu-
ally co-prime, and similarly, ψF = ψg1 ψg2 ψh1 · · · ψht is the additive character
of Fq -order F = g1 g2 h1 · · · ht .
Now, let ψ0 be the canonical additive character of Fqn , then there exist
y1 , y2 ∈ Fqn , such that ψh′ (β) = ψ0 (y1 β) and ψF (g ◦ β) = ψ0 (y2 (g ◦ β)) for all
β ∈ Fqn . Thus,
X X
ψh′ (β)χD (g ◦ β)ψF ((g ◦ β)−1 ) = χD (g ◦ β)ψ0 (y1 β + y2 (g ◦ β)−1 ).
β∈Fqn \Z β∈Fqn \Z

P 
 
Now, we estimate  β∈Fqn \Z χD (g ◦ β)ψ0 (y1β + y2 (g ◦ β)−1 ) using Lemma
2.4, for which we first show that the rational function y1 x + y2 (g ◦ x)−1 is
n
not of the form L(x)q − L(x) for any rational function L(x) = LL12 (in the
simplest form) over the algebraic closure F of Fqn , unless y1 = y2 = 0.
n
Assume that, y1 x + y2 (g ◦ x)−1 = ( LL21 )q − LL21 for some L1 , L2 ∈ F[x]
n P P n n
i.e. Lq2 (y1 ki=0 ai xq +1 + y2 ) = ki=0 ai xq (Lq1 − L1 Lq2 −1 ). First, let y2 6= 0,
i i

Pk n
then i=0 ai xq = Lq2 , which is not possible because k < n. So, y2 must be
i

n n n
zero. Additionally, let y1 6= 0, then Lq2 y1 x = Lq1 − L1 Lq2 −1 , which implies
L2 is a constant polynomial and q n | 1, which is again not possible. Hence,
y1 = y2 = 0. Therefore, if one of the y1 or y2 is non-zero, then by Lemma
2.4,  X 
 −1 
 χD (g ◦ β)ψ0 (y1 β + y2 (g ◦ β) ) ≤ 2q n/2+k . (I)
β∈Fqn \Z

Now, let y1 = y2 = 0, then ψh′ and ψF are trivial additive characters,

i.e. h′ = 1 = F , and the latter one is possible if and only if each factor of
F = g1 g2 h1 · · · ht is 1, since g1 , g2 , h1 , . . . , ht are mutually co-prime. Hence,
we have
X X
χD (g ◦ β)ψ0 (y1 β + y2 (g ◦ β)−1 ) = χD (g ◦ β).
β∈Fqn \Z β∈Fqn \Z

Notice that, if the polynomial g ◦ x = L(x)D for some L(x) ∈ F[x], then
D | gcd(q k , q n − 1), i.e D = 1. Hence, by Lemma 2.3, for D 6= 1, we get
 X 
 
 χD (g ◦ β) ≤ (q k − 1)q n/2 < 2q n/2+k . (II)
β∈Fqn \Z

12
 Finally, let h′ = F = 1 and D = 1, then
X
χD (g ◦ β)ψ0 (y1 β + y2 (g ◦ β)−1) = q n − q k . (III)
β∈Fqn \Z

Using the estimates, obtained in (I), (II) and (III), in Equation (8), we get
   
 
Nr,k,g (h, d, H)−(q n − q k )ℓ (1)sj=1 , (1)ti=1 H ≤
X X X µ   
 ′  s t 
H  (h , d , g )
1 1  ℓ (e )
j j=1 , (hi i=1 ×
)
′
Φ
h |h d1 |d,d2 |u, g1 |H,g2 |π,
ej |λj ;1≤j≤s hi ,|Λi ;1≤i≤t
| {z }
(all are not 1 simultaneously)
X X X
2q n/2+k .
′
ψh χd1 ,χd2 , ψg ,ψg ,
1 2
χej ;1≤j≤s ψh ; 1≤i≤t
i

    
 
Since ℓ (ej )sj=1 , (hi )ti=1  ≤ ℓ (1)sj=1 , (1)ti=1 , then
   
 
Nr,k,g (h, d, H) − (q n − q k )ℓ (1)sj=1, (1)ti=1 H ≤
   Ys Pt 
n/2+k s t
2q ℓ (1)j=1, (1)i=1 H u λj q deg(π)+ i=1 deg(Λi ) W (h)W (d)W (H) − 1 .
j=1

From the above equation, we get

Nr,k,g (h, d, H)
  ≥ (q n − q k ) − 2q n/2+k ×
s t
ℓ (1)j=1 , (1)i=1 H
 Ys
Pt 
u λj q deg(π)+ i=1 deg(Λi ) W (h)W (d)W (H) − 1 .
j=1

Clearly, Nr,k,g (h, d, H) > 0, if

s
Y Pt
n/2−k
q > 2u λj q deg(π)+ i=1 deg(Λi )
W (h)W (d)W (H). (9)
j=1

Now, if gcd(q, n) = 1, then xn − 1 has no repeated factor overPFq , and hence,

t
g = π and deg(Λi ) = 0 for each i = 1, 2, . . . , t. Thus, q deg(π)+ i=1 deg(Λi ) = q k ,

13
 Q
s Qs
otherwise ≤ q 2k . Moreover, u λj = r j=1 pj ≤ r rad(r). Hence, from
j=1
Equation (9), Nr,k,g (h, d, H) > 0, if q n/2−ϑ > 2r rad(r)W (h)W (d)W (H),
where, ϑ = 2k, if gcd(q, n) = 1, and ϑ = 3k, otherwise.
The following corollary provides the desired existence of r-primitive k-
normal pairs (α, α−1 ) in Fqn over Fq .

Corollary 4.1.1. With r and k as in Theorem 4.1, there exists a pair

(α, α−1) of r-primitive k-normal elements in Fqn over Fq , i.e. Nr,k,g (xn −
1, R, G) > 0, if

q n/2−ϑ > 2r rad(r)W (xn − 1)W (R)W (G), (10)

where, ϑ = 2k, if gcd(q, n) = 1, and ϑ = 3k, otherwise.

Proof. Let h = xn − 1, d = R and H = G in the above theorem, then the

result follows immediately.
Observe that, Inequalities (10) can never hold for n ≤ 2ϑ, and as the
value of k increases, the existence of a pair (α, α−1 ) of r-primitive k-normal
elements in Fqn over Fq becomes more challenging for all n. However, we can
show that such a pair exists in all but finitely many fields Fqn over Fq , for
any fixed integers r and k. For this, we need the following two lemmas.

Lemma 4.1. [10, Lemma 2.9] Let q be a prime power and n be a positive
1
integer. Then, we have W (xn − 1) ≤ 2 2 (n+gcd(n,q−1)) . In particular, W (xn −
1) ≤ 2n and W (xn − 1) = 2n if and only if n | (q − 1). Furthermore,
W (xn − 1) ≤ 23n/4 if n ∤ (q − 1), since in this case, gcd(n, q − 1) ≤ n/2.

Lemma 4.2. [3, Lemma 3.7] For any M ∈ N and a positive real number ν,
Q
t
2
W (M) ≤ Cν M 1/ν , where Cν = 1/ν and p1 , p2 , . . . , pt are the primes ≤ 2
ν
i=1 pi
that divide M.

Following the idea of [18, Proposition 3.1], we have the following result.

Proposition 4.1. Let r ≥ 1 and k ≥ 0 be fixed integers. Let q be a prime

power and n > 2ϑ, where ϑ is as defined in Theorem 4.1, such that r | q n − 1
and there exists a polynomial g ∈ Pk . Then, there exists a pair (α, α−1 ) of
r-primitive k-normal elements in all but finitely many fields Fqn over Fq .

14
Proof. Notice that, W (R) ≤ W (q n − 1) and W (G) ≤ W (xn − 1). Hence,
from Lemmas 4.1 and 4.2, Inequality (10) holds if
q n/2−ϑ > 2r rad(r)Cν q n/ν 22n . (11)
Set A = 2r rad(r)Cν and ν = max {4ϑ + 2, 93.46}. Then, taking logarithm on
both sides, we get the following equivalent inequality.
log A + 2nlog 2
log q > .
n/2 − ϑ − n/ν
The right hand side of the above inequality is a decreasing function of n, if
n
2
− ϑ − nν > 0, i.e. ν−2
ν
> 2ϑ
n
, which holds for n > 2ϑ. Therefore, there exists
a number q0 such that the above inequality holds for q ≥ q0 and n ≥ 2ϑ + 1.
Now, for q < q0 , we consider the following inequality, which is equivalent
to Inequality (11).
log A + ϑ log q
n> .
(1/2 − 1/ν)log q − 2 log 2
Clearly, the denominator of the right hand side of the above inequality is
positive, if q > 16. Hence, corresponding to each 16 < q < q0 , there exists a
natural number nq such that the above inequality holds for n ≥ nq .
Now, for 3 ≤ q ≤ 16, we use the bound W (xn − 1) < 2n/3+cq (from [10,
Lemmas 2.9, 2.11]), where cq is a constant corresponding to each q. Then,
Inequality (11) holds, if
log A + ϑ log q + 2cq log 2
n> .
(1/2 − 1/ν)log q − (2 log 2)/3
The above inequality holds for n ≥ nq , where nq is a natural number corre-
sponding to each 3 ≤ q ≤ 16.
n−4
Finally, let q = 2, and from [10, Lemma 2.11], we have W (xn − 1) < 2 5
for all n except n ∈ {1, 2, 3, 4, 5, 7, 9, 15, 21}, and Inequality (11) holds, if
log A + ϑ log q − (8/5) log 2
n> .
(1/2 − 1/ν)log q − (2 log 2)/5
There exists a natural number n2 such that the last inequality holds for
n ≥ n2 .
Hence, from the above discussion, we conclude that, there exists a pair
(α, α−1) of r-primitive k-normal elements in all but finitely many Fqn over
Fq .

15
4.1. The sieving technique
In this section, we give a sieving technique to improve Inequality (10) by
following the methods of Cohen and Huczynska [3]. The proofs of the next
two lemmas are similar to that of [14, Lemmas 5.1, 5.2], hence omitted.
Lemma 4.3. Let r | q n − 1, and let k be a non-negative integer and g ∈ Pk .
Further, let d be a divisor of R and {p1 , p2 , . . . , pl } be the set of remaining
distinct primes dividing R. Furthermore, let h, H be the divisors of xn −1, G
respectively, and {h1 , h2 , . . . , hm }, {H1 , H2 , . . . , Hm′ } be the sets of remaining
distinct irreducible factors of xn − 1, G respectively. Then
m
X l
X
n
Nr,k,g (x − 1, R, G) ≥ Nr,k,g (hhi , d, H) + Nr,k,g (h, dpi , H)+
i=1 i=1
m′
X
Nr,k,g (h, d, HHi ) − (m + l + m′ − 1)Nr,k,g (h, d, H).
i=1

Lemma 4.4. With the notations of the above lemma, define

m l m ′
X 1 X 1 X 1 m + l + m′ − 1
D := 1 − − − and S := + 2.
i=1
q deg(hi ) p
i=1 i i=1
q deg(Hi ) D

Suppose D > 0, then Nr,k,g (xn − 1, R, G) > 0, if

q n/2−ϑ > 2r rad(r)W (h)W (d)W (H)S, (12)

where, ϑ = 2k, if gcd(q, n) = 1, and ϑ = 3k, otherwise.

5. Existence of a primitive 1-normal pair (α, α−1 )

In [3], Cohen and Huczynska proved the existence of a pair (α, α−1 ) of
primitive normal elements in Fqn over Fq . Here, we present an application of
the sufficient condition and the sieving inequality, obtained in the previous
section, by showing the existence of a pair (α, α−1) of primitive 1-normal
elements in Fqn over Fq . All the non-trivial calculations wherever needed in
this section are done using SageMath [19].
From Inequality (10), there exist pairs (α, α−1) of primitive 1-normal
elements in Fqn over Fq , if

q n/2−ϑ > 2W (xn − 1)W (R)W (G), (13)

16
where ϑ = 2, if gcd(q, n) = 1, otherwise, ϑ = 3. Notice that, if a pair (q, n)
satisfies Inequality (13) in the case of ϑ = 3, then it satisfies the same in the
case of ϑ = 2 as well. Moreover, Inequality (13) can never hold for n ≤ 4,
if ϑ = 2, and for n ≤ 6, if ϑ = 3. Therefore, first we assume that ϑ = 3
and settle the cases n ≥ 14 in Subsection 5.1, and then discuss the cases
5 ≤ n ≤ 13 in Subsection 5.2.

5.1. The cases n ≥ 14

From Lemmas 4.1 and 4.2, we have W (R) ≤ W (q n − 1) < Cν q n/ν , W (xn −
1) ≤ 2n and W (G) ≤ 2n−1, and hence, Inequality (13) holds, if
q n/2−3 > Cν q n/ν 22n . (14)
First, we study the cases q > 3 and n ≥ 14 in the following Lemma and then
the cases q = 2, 3 and n ≥ 14 in Lemma 5.3.
Lemma 5.1. Let q > 3 be a prime power and n ≥ 14 be a positive integer.
Then, there always exists a pair (α, α−1) of primitive 1-normal elements in
Fqn over Fq .
Proof. For n ≥ 14, if we choose ν = 7.6, Inequality (14) holds for q ≥ 214183.
Now, for q < 214183, we rewrite Inequality (14) as follows
log (Cν ) + 3 log (q)
n> , (15)
(1/2 − 1/ν)log q − 2 log 2
provided ν > log 2q−4
log q
log 2
and q > 16. Notice that, the value of ν is large for
small values of q, and it is difficult to calculate Cν in our computer. So, for
simplicity, we assume 37 ≤ q < 214183, and for each such q, we find a natural
number nq (see Table 1) such that Inequality (15) holds for n ≥ nq .
2
Now, for 3 < q < 37, we use the bound W (xn − 1) ≤ 2n/3+2(q −1)/3 , which
we obtain from [10, Lemma 2.9], and find that, Inequality (14) holds, if
2
log (24(q −1)/3 Cν )
n> , (16)
(1/2 − 1/ν)log q − (2 log 2)/3
provided ν > 3 log6q−4
log q
log 2
and q > 2. For these values of q, we obtain a natural
number nq (see Table 2) such that Inequality (16) holds for n ≥ nq .
Now, for n < nq and q’s listed in Tables 1 and 2, we test Inequality (13)
with ϑ = 3, and list the values of q and n in Table 3, which fail to satisfy
this inequality.

17
 Table 1: Values of nq for 37 ≤ q < 214183 using W (xn − 1) ≤ 2n .

ν q nq ν q nq ν q nq
7.6 ≥ 214183 14 7.8 307, 311, 313, 317 34 8.3 107 62
7.5 61747 to 214182 15 7.8 277, 281, 283, 289, 293 35 8.3 103 64
7.5 24428 to 61746 16 7.8 257, 263, 269, 271 36 8.3 101 65
7.5 11926 to 24427 17 7.8 243, 251, 256 37 8.4 97 67
7.5 6738 to 11925 18 7.8 229, 233, 239, 241 38 8.5 89 73
7.5 4231 to 6737 19 7.9 223, 227 39 8.5 83 79
7.5 2875 to 4230 20 7.9 211 40 8.6 81 81
7.5 2074 to 2874 21 7.9 197, 199 41 8.6 79 83
7.5 1569 to 2073 22 7.9 191, 193 42 8.7 73 92
7.5 1233 to 1568 23 7.9 179, 181 43 8.8 71 96
7.5 999 to 1232 24 7.9 173 44 8.8 67 104
7.5 829 to 998 25 8 167, 169 45 8.9 64 111
7.5 703 to 828 26 8 157, 163 47 9 61 120
7.6 606 to 702 27 8 149, 151 48 9.1 59 128
7.6 531 to 605 28 8.1 137, 139 51 9.3 53 157
7.6 471 to 530 29 8.1 128, 131 53 9.6 49 188
7.6 422 to 470 30 8.1 125, 127 54 9.7 47 210
7.6 382 to 421 31 8.1 121 56 10.1 43 272
7.7 348 to 381 32 8.2 113 59 10.3 41 320
7.7 331, 337, 343, 347 33 8.3 109 61 10.8 37 489

2
Table 2: Values of nq for 3 < q < 37 using W (xn − 1) ≤ 2n/3+2(q −1)/3
.

ν 11.8 11.7 11.6 11.4 11.2 11 10.6 10.3

q 32 31 29 27 25 23 19 17
nq 1019 973 883 797 714 636 493 428
ν 10.1 9.7 9.4 9 8.8 8.7 8.5 9
q 16 13 11 9 8 7 5 4
nq 397 312 263 222 206 195 214 351

Now, for the values of q and n, listed in Table 3, we test the SageMath
procedure TEST_SIEVE(q,n,ϑ) with ϑ = 3, in which we choose the suitable
values of h, d and H, and test Inequality (12) (whose pseudocode is given in
Appendix A), which returns True for all the pairs (q, n) except (4, 15), (5,
16), (5, 24), (8, 14), (9, 16), (16, 15), (17, 16), (19, 18). For these remaining
pairs of (q, n), we explicitly search for a pair (α, α−1 ) of primitive 1-normal el-
ements in Fqn over Fq using SageMath procedure DIRECT_SEARCH(q,n) (pseu-
docode is given in Appendix A), and find that such a pair always exists.
Now, for the cases q = 2, 3, first we recall some results. Let q be a prime
power and n = q i · n′ be a positive integer with gcd(n′ , q) = 1; i ≥ 0. Further,
′
let e be the multiplicative order of q modulo n′ , then xn − 1 factors into
irreducible polynomials of degree ≤ e. Let ρ(q, n′ ) be the ratio of the number

18
 Table 3: Pairs (q, n) fail to satisfy Inequality (13) with ϑ = 3.

n q
14 4, 5, 8, 9, 11, 13, 23, 25, 27, 29, 41, 43, 64, 71, 113, 125, 127, 169, 197, 211, 239, 281, 337,
379, 421, 449, 463, 491, 547, 617, 631, 659, 673, 701, 729, 743, 883, 911, 953
15 4, 7, 8, 11, 13, 16, 19, 29, 31, 41, 49, 61, 64, 71, 121, 151, 181, 211, 241, 256, 271, 331, 361,
421, 571, 631, 751, 841
16 5, 7, 9, 11, 13, 17, 19, 23, 25, 27, 29, 31, 41, 49, 73, 81, 89, 97, 113, 193, 241, 257, 289, 337,
401
17 4, 16, 103, 256
18 4, 5, 7, 13, 17, 19, 25, 31, 37, 43, 73, 109, 127, 163, 181, 199, 289, 361
20 7, 9, 11, 13, 19, 29, 31, 41, 61, 81, 101, 121
21 4, 8, 13, 16, 43, 64, 169
22 23, 67, 89
23 47
24 5, 7, 11, 13, 17, 19, 25, 37, 49, 73, 97, 121
26 27, 53
28 13, 29
30 4, 7, 11, 19, 31, 61
31 32
32 17
36 5, 19, 37
40 9, 11, 41
42 43
45 4
48 5, 7
63 5

′
of irreducible factors of xn − 1 of degree < e to n′ , then nρ(q, n) = n′ ρ(q, n′ ),
and we have the following lemma.

Lemma 5.2. [3, Lemma 7.1] Assume that n > 4 (p ∤ n), then the following
holds:

(i) Suppose q = 2, then ρ(2, 5) = 1/5; ρ(2, 9) = 2/9; ρ(2, 21) = 4/21;
otherwise ρ(2, n) ≤ 1/6.

(ii) Suppose q = 3, then ρ(3, 16) = 5/16; otherwise ρ(3, n) ≤ 1/4.

′
If we choose h as the product of all irreducible factors of xn − 1 of degree
< e, H = h/(x − 1) and d = R in Lemma 4.4, then following Lemma 10 of
[15], we get that, D > 1 − 2/e > 0 and S ≤ 2n′ ≤ 2n, provided e > 2 (i.e.
n′ ∤ q 2 − 1). Using these results, we prove the following lemma.

Lemma 5.3. Let q = 2, 3 and n ≥ 14 be a positive integer. Then, there

always exists a pair (α, α−1) of primitive 1-normal elements in Fqn over Fq .

19
Proof. Let n = q i · n′ with gcd(q, n′ ) = 1 and i ≥ 0. First, assume that
n′ ∤ q 2 − 1, then e > 2 and n′ > 4. If we choose d = R, h as the product
′
of all irreducible factors of xn − 1 of degree < e, and H = h/(x − 1), then
from the above discussion and Inequality (12), there exists a pair (α, α−1) of
primitive 1-normal elements in Fqn over Fq , if

q n/2−3 > 2Cν q n/ν W (h)W (H)S. (17)

′ ′
Notice that, W (H) = W (h)/2 and W (h) = 2n ρ(q,n ) . From Lemma 5.2
n′ ρ(q, n′ ) ≤ n′ /6 ≤ n/6, when q = 2, and n′ ρ(q, n′ ) ≤ n′ /4 ≤ n/4, when
q = 3. Thus, the above inequality holds, if
 n/2−3
q > Cν q n/ν 2n/3 2n, for q = 2,
q n/2−3 > Cν q n/ν 2n/2 2n, for q = 3.

The above inequalities hold for n ≥ 557, when q = 2 and ν = 8.7, and for
n ≥ 265, when q = 3 and ν = 8.3. For the remaining q and n, we test
Inequality (13), and find that this fails for n = 14, 15, 17, 18, 20, 21, 22, 27,
28, 30, 31, 33, 35, 42, 45, 63, when q = 2, and for n = 14, 15, 16, 20, 22, 26,
28, 32, 40, when q = 3.
Now, assume that n′ | q 2 − 1, and select d = R, h as the product of all the
′
linear factors of xn − 1, and H = h/(x − 1) in Lemma 4.4. Then, for q = 2,
′
there exists exactly one linear and at most 1 quadratic factor of xn − 1 over
F2 . Therefore, in this case, D ≥ 1/2 and S ≤ 4. Similarly, for q = 3, there
′
exists at most 2 linear and at most 2 quadratic factors of xn − 1 over F3 .
Thus, in this case, D ≥ 1/3 and S ≤ 17. If we take ν = 6, then Inequality
(17) holds for n ≥ 33 in both the cases. For 14 ≤ n < 33, we test Inequality
(13), which does not hold for n = 16, when q = 2, and for n = 18, 24, when
q = 3.
Finally, for q = 2, 3 and the remaining values of n, we apply the SageMath
procedure TEST_SIEVE(q,n,ϑ) with ϑ = 3, which returns False only for
q = 2 and n = 14, 15, 16, 18, 20, 21, 24, 30, and for q = 3 and n = 14,
16. For these remaining q and n, we explicitly search for a pair (α, α−1 )
of primitive 1-normal elements in Fqn over Fq using SageMath procedure
DIRECT_SEARCH(q,n), and find that such a pair always exists.

5.2. The cases 5 ≤ n ≤ 13

Observe that, Inequality (13) is applicable for the cases, n = 5, 6, only
if gcd(n, q) = 1. Therefore, first we settle the cases 7 ≤ n ≤ 13 for all

20
q ≥ 2, and then the cases n = 5, 6 with the assumption that gcd(q, n) = 1.
Before moving ahead, first we state the following lemma, which is a direct
consequence of Lemma 4.4.

Lemma 5.4. Let d be a divisor of q n − 1 and n0 be a fixed positive integer

such that all the prime divisors of (q n − 1)/d, co-prime to d, are of the form
n0 s + 1 for some positive integer s. Further, let p1 , p2Q
, . . . , pl be the first l
primes of the form n0 s + 1 such that the product Pl = li=1 pi ≤ (q n − 1)/d,
Pl
and denote the sum of their reciprocals by Sl = i=1 1/pi . If h = 1 and
n n
H = 1 are the divisors of x − 1 and (x − 1)/(x − 1), respectively, then
D ≥ 1 − Sl − (2n − 1)/q and S ≤ (l + 2n − 2)/D + 2. Moreover, if D > 0
and q n/2−ϑ > 2W (d)S, then there exists a pair (α, α−1) of primitive 1-normal
elements in Fqn over Fq , where D, S and ϑ are the same as defined in Lemma
4.4 and Inequality (13).

Lemma 5.5. Let 7 ≤ n ≤ 13 and q be a prime power such that gcd(q, n) = 1.

Then, there always exists a pair (α, α−1) of primitive 1-normal elements in
Fqn over Fq .

Proof. Since, gcd(q, n) = 1, Inequality (13) holds, if

q n/2−2 > Cν q n/ν 22n . (18)

For each 7 ≤ n ≤ 13 and suitable choices of ν, the above inequality holds

for q ≥ Mn , where the values of Mn are listed in Table 4. Now, for each

Table 4: Values of Mn such that Inequality (18) holds for q ≥ Mn .

Mn 3.60 × 1012 2.10 × 108 2.60 × 106 222280 46892 16049 7331
n 7 8 9 10 11 12 13
ν 7.8 7.4 7.2 7.1 7.1 7.1 7.1

10 ≤ n ≤ 13, and q < Mn , by calculating the exact values of W (xn −1), W (R)
and W (G), and testing Inequality (13) with ϑ = 2, we obtain that, there are
136 pairs of (q, n), which fail to satisfy this inequality. For these pairs,
we use the SageMath procedure TEST_SIEVE(q,n,ϑ) with ϑ = 2, which
returns False for the pairs (3, 10), (11, 10), (2, 11), (3, 11), (5, 12), (7, 12),
and (13, 12). Finally, for these pairs, we explicitly search for a pair (α, α−1 )
of primitive 1-normal elements in Fqn over Fq using SageMath procedure
DIRECT_SEARCH(q,n), and find that such a pair always exists.

21
 Now, let n = 7, then for q < M7 , we use Lemma 5.4 with d = q −1, h = 1,
7
P6 p idividing (q − 1)/d, co-prime to
H = 1 and ϑ = 2. In this case, the primes
d, are of the form 7s + 1, and Pl ≤ i=0 M7 for l ≤ 29. Further, if q ≥ 65,
then D > 0.6608 and S ≤ 64.042, and there exists a desired pair (α, α−1 ),
if q 3/2 > 2Cν q 1/ν S, which is true for q ≥ 149 and ν = 2.8. Proceeding in
this way for 65 ≤ q < 149, we get that, the inequality q 3/2 > 2Cν q 1/ν S holds
for all 68 ≤ q ≤ 149. Now, for q < 68, we use the SageMath procedure
TEST_SIEVE(q,n,ϑ) with ϑ = 2, which returns False for q = 2, 3, 4, 5, 8,
9, 11, 13, and n = 7. In the end, for these pairs, we explicitly search for a
pair (α, α−1) of primitive 1-normal elements in Fqn over Fq using SageMath
procedure DIRECT_SEARCH(q,n), and find that such a pair always exists.
Finally, for the cases n = 8, 9, and q < Mn , again we use Lemma 5.4
with d = q 4 − 1, q 3 − 1, respectively. In these cases, the primes p dividing
(q n − 1)/d, co-prime to d, are of the form 8s + 1, when n = 8, and of the
form 9s + 1, when n = 9. Now, following the same line as in the case n = 7,
we have reduced the bounds on q from M8 to 203, and from M9 to 41.Now,
for q < 68, we use the SageMath procedure TEST_SIEVE(q,n,ϑ) with ϑ = 2,
which returns False for q = 3, 5, 7, 9, 11, 13, 17 and n = 8, and q =
2, 4, 7, 19, and n = 9. Finally, for these pairs, we explicitly search for a
pair (α, α−1) of primitive 1-normal elements in Fqn over Fq using SageMath
procedure DIRECT_SEARCH(q,n), and find that such a pair always exists.
Lemma 5.6. Let 7 ≤ n ≤ 13 and q be a prime power such that gcd(q, n) 6= 1.
Then, there always exists a pair (α, α−1) of primitive 1-normal elements in
Fqn over Fq .
Proof. Since, gcd(q, n) 6= 1, q = pt , for each prime p dividing n and some
positive integer t, and Inequality (13) holds, if

pnt/2−3 > Cν pnt/ν 22n (19)

For each 7 ≤ n ≤ 13, the above inequality holds for t ≥ tn,p and the
suitable choices of ν, listed in Table 5. For the cases 9 ≤ n ≤ 13, and
Table 5: Values of tn,p such that Inequality (19) holds for t ≥ tn,p .

tn,p 3797 223 46 42, 19 9 24, 15 6

(n, p) (7, 7) (8, 2) (9, 3) (10, 2), (10, 5) (11, 11) (12, 2), (12, 3) (13, 13)
ν 15.7 10.3 9 8 7.5 7.5 7

t < tn,p , by calculating the exact values of W (xn − 1), W (R) and W (G),

22
and testing Inequality (13) with ϑ = 3, we obtain that, there are 25 pairs of
(q, n), which fail to satisfy this inequality. For these pairs, we use the Sage-
Math procedure TEST_SIEVE(q,n,ϑ) with ϑ = 3, which returns False for
the pairs (3, 9), (9, 9), (2, 10), (4, 10), (16, 10), (5, 10), (2, 12), (4, 12), (3, 12)
and (9, 12). Finally, for these pairs, we explicitly search for a pair (α, α−1 )
of primitive 1-normal elements in Fqn over Fq using SageMath procedure
DIRECT_SEARCH(q,n), and find that such a pair always exists.
Now, let n = 7, q = 7t and t < t7,7 . Clearly, x7 − 1 has only one linear
factor over Fq and 7 ∤ q 7 −1, therefore, using Lemma 4.4 with d = q −1, h = 1
and H = 1, we get that there exists a pair (α, α−1) of desired properties, if
7t/2 > 2Cν 7t/ν S. Since, the primes dividing (q 7 − 1)/d are of the form 7s + 1,
so let Pl and Sl be the product and sum ofPreciprocals of first l primes of the
form 7s + 1 respectively, such that Pl ≤ 6i=0 7it , then l ≤ 3891, and from
Lemma 4.4, D ≥ 1 − Sl − 1/7t and S ≤ l/D + 2. Now, For t ≥ 8, we get
D > 0.7661 and S ≤ 5080.82, and inequality 7t/2 > 2Cν 7t/ν S holds for t ≥ 20
and ν = 5. Proceeding in this way for 8 ≤ t < 20, we get that, the inequality
7t/2 > 2Cν 7t/ν S always holds for t ≥ 11. In the end, for 1 ≤ t ≤ 10, we
first perform the SageMath procedure TEST_SIEVE(q,n,ϑ) with ϑ = 3, which
returns False only for t = 1, 2, 3, 4, and then for these values, we explicitly
search for a pair (α, α−1) of primitive 1-normal elements in Fqn over Fq using
SageMath procedure DIRECT_SEARCH(q,n), and find that such a pair always
exists.
Now, in the case n = 8, we have q = 2t and x8 − 1 = (x − 1)8 over Fq . If
d = q 4 −1 and p is a prime divisor of q 4 + 1 co-prime to d, then q 4 6≡ 1(mod p)
and q 4 ≡ −1(mod p), which implies that, the multiplicative order of q modulo
p is 8. Thus, p is of the form 8s+1. Again, we use Lemma 4.4 with d = q 4 −1,
h = H = 1. Let Pl and Sl be the product and sum of reciprocals of first l
primes of the form 8s + 1 respectively, such that Pl ≤ 24t + 1, then l ≤ 90 for
t < t8,2 , and from Lemma 4.4, D ≥ 1 − Sl − 1/2t and S ≤ l/D + 2. Now, For
t ≥ 8, we get D > 0.764 and S ≤ 119.75, and inequality 2t > 2Cν 24t/ν S holds
for t ≥ 35 and ν = 7. Proceeding in this way for 8 ≤ t < 35, we get that, the
inequality 2t > 2Cν q 4t/ν S always holds for t ≥ 30. In the end, for 1 ≤ t ≤ 29,
we obtain that, there are 14 values of t (1 ≤ t ≤ 12; t = 15, 18) which fails
to satisfy Inequality (13). For these values, we use the SageMath procedure
TEST_SIEVE(q,n,ϑ) with ϑ = 3, which returns False for 1 ≤ t ≤ 6. Finally,
we explicitly search for a pair (α, α−1 ) of primitive 1-normal elements in Fqn
over Fq using SageMath procedure DIRECT_SEARCH(q,n), and find that such
a pair always exists.

23
Lemma 5.7. Let n = 5, 6, and q be a prime power such that gcd(q, n) = 1.
Then, there always exists a pair (α, α−1) of primitive 1-normal elements in
Fqn over Fq with the sole genuine exception (4, 5).
Proof. First, we test Inequality (18) in both the cases n = 5, 6, and get that,
this holds for q ≥ Mn , where M5 = 6.97 × 10323 , if we set ν = 11.9, and
M6 = 4.74 × 1027 , if we set ν = 8.7. Now, for q < Mn , in each case, we use
Lemma 5.4 with d = q − 1 for n = 5, and d = q 2 − 1 for n = 6. Notice that,
primes dividing (q n − 1)/d, co-prime to d, are of the form 5s + 1 for n = 5,
and 3s+1 for n = 6. Then, for n = 5 and q ≥ 105 , we get l ≤ 360, D > 0.648,
S ≤ 569.9, and for n = 6 and q ≥ 104 , we get l ≤ 48, D > 0.429, S ≤ 137.11.
With these, we test the inequalities q 1/2 > 2Cν q 1/ν S for n = 5; ν = 5.5, and
q > 2Cν q 2/ν S for n = 6; ν = 5.2, and get that, there always exists a pair
(α, α−1) of desired properties for q ≥ 4.395 × 1013 and n = 5, and q ≥ 646882
and n = 6. By repeating the process for remaining values of q in each case,
we significantly reduce the bound on q from M5 to 4.96 × 109 , and from M6
to 94531.
Now, for n = 5; 6 and q < 4.96 × 109 ; 94531, respectively, we use Lemma
5.4 with d = gcd(2·3·5, q n −1). Notice that, the primes dividing (q n −1)/d co-
prime to d are of the form 2s + 1 in each case. Then, for 105 ≤ q < 4.96 × 109
and n = 5, we get l ≤ 28, D > 0.175, S ≤ 207.05, and the inequality
q 1/2 > 24 S holds for q ≥ 10973928. Similarly, for 103 ≤ q < 94531 and
n = 6, we get l ≤ 19, D > 0.253, S ≤ 116.58, and the inequality q > 24 S
holds for q ≥ 1866. Proceeding in this way, we get that, there always exists a
pair (α, α−1) of desired properties for q ≥ 3839356 and n = 5, and q ≥ 1109
and n = 6. For the remaining values of q in each case, we run the SageMath
procedure TEST_SIEVE(q,n,ϑ) with ϑ = 2 and get that, there are 871 values of
q (listed in Table A.6) in the case of n = 5, and 47 values of q (listed in Table
A.6) in the case of n = 6, for which this procedure returns False. Finally,
we perform the procedure DIRECT_SEARCH(q,n) for the remaining values of
t, and find that, there always exists the desired pairs (α, α−1) with the sole
genuine exception (4, 5).
Summing up Lemmas 5.1, 5.3, 5.5, 5.6, and 5.7, we conclude our Theorem
1.3.

5.3. A note on the cases n = 1, 2, 3, 4

From the definition of k-normal elements in Fqn over Fq , it is clear that
0 ≤ k ≤ n − 1, therefore, it is sensible to talk about the existence of a pair

24
(α, α−1) of primitive 1-normal elements in Fqn over Fq for n > 1. Now, for the
case n = 2, if α is a 1-normal element in Fqn over Fq , then either αq − α = 0
or αq + α = 0, which implies ord(α) ≤ 2(q − 1) < q 2 − 1. Thus, there does
not exist any pair (α, α−1) of primitive 1-normal elements in Fq2 over Fq .
Further, we show that, such a pair does not exists in the case of n = 3 also.

Lemma 5.8. Let n = 3 and q ≥ 2 be any prime power. Then, there does
not exist any pair (α, α−1) of primitive 1-normal elements in Fqn over Fq .

Proof. Suppose that, there exists a pair (α, α−1) of primitive 1-normal ele-
ments in Fq3 over Fq . Since, x3 − 1 = (x − 1)3 or (x − 1)(x2 + x + 1) or
(x − 1)(x − a)(x − a−1 ) for some a ∈ Fq \ {0, 1}. Therefore, we have the
following cases on the Fq -orders of α and α−1 .
Case 1. If Ordq (α) = Ordq (α−1 ) = x2 + x + 1, then, the trace of α and
α−1 over Fq is equal to zero. Consequently, the primitive polynomial of α
will be of the form x3 + a for some a ∈ Fq , and hence α3 ∈ Fq . This means
q 3 − 1 | 3(q − 1), which is never possible for any q > 1.
Case 2. If (x − 1) | gcd(Ordq (α), Ordq (α−1 )), then we get that, (αq − α)q−1
2
and (α−q − α−1 )q−1 both belong to Fq . This implies αq −1 ∈ Fq , i.e. q 3 − 1 |
(q 2 − 1)(q − 1), which is not possible.
Case 3. If Ordq (α) = (x − 1)(x − a) and Ordq (α−1 ) = (x − a)(x − a−1 ),
then we get that, (αq − aα)q−1 = 1 and (α−q − a−1 α−1 )q−1 ∈ Fq . This implies
2
αq −1 ∈ Fq , i.e. q 3 − 1 | (q 2 − 1)(q − 1), which is not possible.
Case 4. If Ordq (α) = (x − 1)(x − a−1 ) and Ordq (α−1 ) = (x − a)(x − a),then
by interchanging the role of α and α−1 in Case 3, we get that, q 3 − 1 |
(q 2 − 1)(q − 1), which is not possible.
This completes the proof of the lemma.
We also provide a necessary condition for the existence of a pair (α, α−1 )
of primitive 1-normal elements in Fq4 over Fq .

Lemma 5.9. If there exists a pair (α, α−1) of primitive 1-normal elements
in Fq4 over Fq , then q ≡ 1 (mod 4).

Proof. Let’s assume that q 6≡ 1 (mod 4), then the polynomial x4 −1 factorizes
either as (x − 1)4 , or as (x − 1)(x + 1)(x2 + 1) over Fq . Further, let (α, α−1 )
be a pair of 1-normal elements in Fq4 over Fq . Then, we have the following
cases.

25
 2 2
Case 1: (x−1) | gcd(Ordq (α), Ordq (α−1 )), then αq −α and α−q −α−1 both
2
will belong to Fq . This implies αq +1 ∈ Fq , i.e. ord(α) ≤ (q 2 + 1)(q − 1) <
q 4 − 1.
2 2
Case 2: (x+1) | gcd(Ordq (α), Ordq (α−1 )), then (αq −α)2 and (α−q −α−1 )2
2
both will belong to Fq . This implies α2(q +1) ∈ Fq , i.e. ord(α) ≤ 2(q 2 + 1)(q −
1) < q 4 − 1.
2
Case 3: gcd(Ordq (α), Ordq (α−1 )) = x2 + 1, then either αq − α and
2 2 2
(α−q − α−1 )2 belong to Fq , or (αq − α)2 and α−q − α−1 will belong to Fq
simultaneously. In either case, we get that ord(α) ≤ 2(q 2 + 1)(q − 1) < q 4 − 1.
In each case, we get that ord(α) < q 4 − 1, which implies α is never a
primitive element in Fq . Thus, a pair (α, α−1) of primitive 1-normal elements
in Fq4 over Fq can exist, only if q ≡ 1 (mod 4).
For the cases, n = 4 and q = 4t + 1; n = 5 and q = 5t ; and n = 6 and q =
2t , 3t , where t is a positive integer, we performed the extensive experiments,
and obtained that, there always exists a pair (α, α−1) of primitive 1-normal
in Fqn over Fq unless n = 6 and q = 2, 4. On the basis of these experiments,
we conjecture the following.

Conjecture 5.1. Let t be a positive integer. There always exists a pair

(α, α−1) of primitive 1-normal in Fqn over Fq if n = 4 and q = 4t + 1; n = 5
and q = 5t ; and n = 6 and q = 2t , 3t with the sole genuine exceptions (2, 6),
(4, 6).

6. Acknowledgements
This work was funded by Council of Scientific & Industrial Research,
under Grant F. No. 09/045(1674)/2019-EMR-I and University Grant Com-
mission, under Grant Ref. No. 1042/CSIR-UGC NET DEC-2018.

References
[1] Aguirre, J.J., Neumann, V.G., 2021. Existence of primitive 2-normal
elements in finite fields. Finite Fields and Their Applications 73, 101864.

[2] Blum, M., Micali, S., 1984. How to generate cryptographically strong
sequences of pseudorandom bits. SIAM journal on Computing 13, 850–
864.

26
 [3] Cohen, S.D., Huczynska, S., 2010. The strong primitive
normal basis theorem. Acta Arith. 143, 299–332. URL:
https://doi.org/10.4064/aa143-4-1, doi:10.4064/aa143-4-1.

[4] Cohen, S.D., Kapetanakis, G., 2021. Finite field extensions with the line
or translate property for-primitive elements. Journal of the Australian
Mathematical Society 111, 313–319.

[5] Diffie, W., Hellman, M., 1976. New directions in cryptography. IEEE
transactions on Information Theory 22, 644–654.

[6] Fu, L., Wan, D., 2014. A class of incomplete character sums. Quarterly
Journal of Mathematics 65, 1195–1211.

[7] Gao, S., 1999. Elements of provable high orders in finite fields. Proceed-
ings of the American Mathematical Society 127, 1615–1623.

[8] Huczynska, S., Mullen, G.L., Panario, D., Thomson, D., 2013. Existence
and properties of k-normal elements over finite fields. Finite Fields and
Their Applications 24, 170–183.

[9] Kapetanakis, G., Reis, L., 2019. Variations of the primitive normal basis
theorem. Designs, Codes and Cryptography 87, 1459–1480.

[10] Lenstra, Jr., H.W., Schoof, R.J., 1987. Primitive normal

bases for finite fields. Math. Comp. 48, 217–231. URL:
https://doi.org/10.2307/2007886, doi:10.2307/2007886.

[11] Lidl, R., Niederreiter, H., 1997. Finite fields. volume 20. Second ed.,
Cambridge University Press, Cambridge.

[12] Meletiou, G., Mullen, G.L., 1992. A note on discrete logarithms in

finite fields. Applicable Algebra in Engineering, Communication and
Computing 3, 75–78.

[13] Mullen, G.L., 2016. Some open problems arising from my recent fi-
nite field research, in: Contemporary developments in finite fields and
applications. World Scientific, pp. 254–269.

[14] Rani, M., Sharma, A.K., Tiwari, S.K., 2021a. On r-primitive k-normal
elements over finite fields. Communicated .

27
[15] Rani, M., Sharma, A.K., Tiwari, S.K., Gupta, I., 2021b. On the exis-
tence of pairs of primitive normal elements over finite fields. São Paulo
Journal of Mathematical Sciences , 1–18.

[16] Reis, L., 2019. Existence results on k-normal elements over finite fields.
Revista Matemática Iberoamericana 35, 805–822.

[17] Reis, L., Thomson, D., 2018. Existence of primitive 1-normal elements
in finite fields. Finite Fields and Their Applications 51, 238–269.

[18] Sharma, A.K., Rani, M., Tiwari, S.K., 2022. Primitive normal pairs
with prescribed norm and trace. Finite Fields and Their Applications
78, 101976.

[19] The Sage Developers, 2020. SageMath, the Sage Mathematics Software
System (Version 9.0). URL: https://www.sagemath.org.

[20] Tian, T., Qi, W.F., 2006. Primitive normal element and its inverse in
finite fields. Acta Math. Sinica (Chin. Ser.) 49, 657–668.

28
Appendix A.

Table A.6: Pairs (q, n) for which TEST_SIEVE(q,n) returns False in the cases n = 5, 6.

n q
5 2, 3, 4, 7, 8, 9, 11, 13, 16, 17, 19, 23, 27, 29, 31, 32, 37, 41, 43, 47, 49, 53, 59, 61, 64, 67, 71, 73, 79, 81, 83,
89, 97, 101, 103, 107, 109, 113, 121, 127, 128, 131, 137, 139, 149, 151, 157, 163, 167, 169, 173, 179, 181,
191, 193, 197, 199, 211, 223, 227, 229, 233, 239, 241, 243, 251, 256, 257, 263, 269, 271, 277, 281, 283, 289,
293, 307, 311, 313, 317, 331, 337, 343, 347, 349, 353, 359, 361, 367, 373, 379, 383, 389, 397, 401, 409, 419,
421, 431, 433, 439, 443, 449, 457, 461, 463, 467, 479, 487, 491, 499, 503, 509, 512, 521, 523, 529, 541, 547,
557, 563, 569, 571, 577, 587, 593, 599, 601, 607, 613, 617, 619, 631, 641, 643, 647, 653, 659, 661, 673, 677,
683, 691, 701, 709, 719, 727, 729, 733, 739, 743, 751, 757, 761, 769, 773, 787, 797, 809, 811, 821, 823, 827,
829, 839, 841, 853, 857, 859, 863, 877, 881, 883, 887, 907, 911, 919, 929, 937, 941, 947, 953, 961, 967, 971,
977, 991, 997, 1009, 1013, 1019, 1021, 1024, 1031, 1033, 1039, 1049, 1051, 1061, 1063, 1069, 1087, 1091,
1093, 1103, 1109, 1117, 1123, 1129, 1151, 1153, 1163, 1171, 1181, 1193, 1201, 1213, 1229, 1231, 1237, 1249,
1259, 1279, 1289, 1291, 1301, 1303, 1321, 1327, 1331, 1361, 1367, 1369, 1373, 1381, 1399, 1409, 1423, 1429,
1439, 1447, 1451, 1453, 1459, 1471, 1481, 1483, 1489, 1499, 1511, 1531, 1543, 1549, 1559, 1567, 1571, 1579,
1583, 1597, 1601, 1609, 1619, 1621, 1627, 1667, 1669, 1681, 1693, 1697, 1699, 1709, 1721, 1723, 1741, 1747,
1753, 1759, 1777, 1783, 1789, 1801, 1811, 1831, 1849, 1861, 1867, 1871, 1873, 1879, 1901, 1931, 1933, 1949,
1951, 1993, 1999, 2003, 2011, 2017, 2029, 2053, 2081, 2083, 2087, 2089, 2111, 2113, 2129, 2131, 2137, 2141,
2143, 2161, 2179, 2197, 2203, 2209, 2221, 2239, 2251, 2269, 2281, 2293, 2297, 2311, 2341, 2347, 2351, 2371,
2377, 2381, 2389, 2401, 2411, 2437, 2441, 2467, 2473, 2503, 2521, 2531, 2539, 2551, 2557, 2591, 2593, 2621,
2647, 2659, 2671, 2683, 2689, 2699, 2707, 2711, 2713, 2719, 2731, 2741, 2749, 2767, 2791, 2797, 2801, 2803,
2809, 2851, 2857, 2861, 2887, 2927, 2953, 2971, 3001, 3011, 3019, 3037, 3041, 3049, 3061, 3067, 3079, 3109,
3121, 3169, 3181, 3191, 3217, 3221, 3251, 3259, 3271, 3301, 3319, 3331, 3361, 3371, 3391, 3433, 3457, 3461,
3463, 3469, 3481, 3491, 3499, 3511, 3529, 3541, 3547, 3571, 3581, 3613, 3631, 3671, 3691, 3697, 3701, 3721,
3727, 3733, 3739, 3761, 3821, 3823, 3851, 3853, 3877, 3881, 3907, 3911, 3919, 3931, 4001, 4003, 4019, 4021,
4051, 4057, 4096, 4111, 4129, 4159, 4201, 4211, 4219, 4229, 4231, 4241, 4243, 4261, 4271, 4327, 4391, 4421,
4441, 4447, 4451, 4481, 4489, 4519, 4561, 4591, 4603, 4621, 4651, 4691, 4751, 4759, 4789, 4801, 4831, 4861,
4871, 4909, 4931, 4951, 4957, 4999, 5011, 5021, 5041, 5101, 5119, 5167, 5171, 5179, 5209, 5261, 5281, 5329,
5351, 5381, 5419, 5431, 5441, 5449, 5471, 5501, 5503, 5521, 5531, 5581, 5591, 5641, 5659, 5701, 5741, 5743,
5779, 5791, 5801, 5821, 5839, 5851, 5861, 5881, 5923, 5981, 6007, 6011, 6043, 6091, 6121, 6131, 6151, 6163,
6211, 6221, 6241, 6271, 6301, 6361, 6421, 6427, 6451, 6469, 6481, 6521, 6561, 6571, 6581, 6637, 6661, 6679,
6691, 6761, 6763, 6781, 6791, 6841, 6871, 6889, 6961, 6991, 7001, 7039, 7121, 7151, 7177, 7309, 7321, 7351,
7411, 7417, 7451, 7481, 7541, 7549, 7561, 7591, 7621, 7681, 7741, 7841, 7921, 7951, 8011, 8101, 8161, 8171,
8191, 8221, 8231, 8311, 8317, 8419, 8431, 8461, 8501, 8521, 8581, 8641, 8731, 8737, 8761, 8779, 8821, 8941,
8971, 9001, 9091, 9151, 9161, 9181, 9199, 9241, 9311, 9391, 9409, 9421, 9491, 9511, 9521, 9601, 9619, 9631,
9661, 9721, 9769, 9781, 9811, 9829, 9871, 9901, 9931, 10111, 10141, 10151, 10201, 10321, 10459, 10501,
10531, 10609, 10651, 10711, 10771, 10831, 10861, 10891, 11047, 11071, 11131, 11161, 11251, 11257, 11311,
11411, 11491, 11551, 11621, 11701, 11719, 11731, 11821, 11881, 11941, 11971, 12211, 12241, 12301, 12391,
12421, 12433, 12451, 12511, 12541, 12601, 12721, 12769, 12781, 12841, 13171, 13291, 13381, 13399, 13411,
13441, 13567, 13681, 13711, 13831, 13921, 14071, 14221, 14251, 14281, 14341, 14401, 14431, 14551, 14641,
14731, 14771, 14821, 14851, 15031, 15121, 15271, 15331, 15361, 15391, 15451, 15511, 15541, 15601, 15661,
15901, 15991, 16111, 16141, 16231, 16339, 16381, 16651, 16831, 16921, 17011, 17161, 17191, 17341, 17431,
17491, 17581, 17791, 17851, 17911, 18061, 18121, 18181, 18301, 18451, 18481, 18661, 18691, 19081, 19141,
19231, 19321, 19381, 19441, 19471, 19501, 19531, 19801, 19891, 20011, 20101, 20161, 20431, 20551, 20641,
20749, 21001, 21121, 21211, 21481, 21751, 21841, 21871, 21961, 22051, 22111, 22201, 22291, 22441, 22531,
22621, 22741, 22801, 22861, 23011, 23071, 23131, 23311, 23371, 23431, 23761, 23971, 24061, 24091, 24121,
24151, 24181, 24391, 24421, 24481, 24571, 24781, 24841, 25111, 25261, 25411, 25621, 25741, 26041, 26251,
26701, 26821, 26881, 27031, 27091, 27361, 27691, 27751, 27901, 28051, 28081, 28351, 28561, 28711, 28771,
29131, 29191, 29401, 29581, 29611, 29641, 29671, 29881, 30211, 30241, 30391, 30661, 30871, 31081, 31321,
31531, 31981, 32191, 32341, 32371, 32761, 32971, 33091, 33181, 33391, 33811, 34171, 34651, 35281, 35311,
35491, 36061, 36481, 36541, 37171, 37591, 37951, 38611, 38851, 39901, 40111, 40531, 41611, 42331, 42841,
43261, 43891, 44521, 46411, 47881, 51871, 53551, 57121, 60901, 62791, 63211, 65521, 70981
6 5, 7, 11, 13, 17, 19, 23, 25, 29, 31, 37, 41, 43, 47, 49, 53, 59, 61, 67, 71, 73, 79, 83, 89, 97, 101, 103, 107,
109, 113, 121, 127, 131, 137, 139, 149, 151, 157, 163, 169, 179, 181, 191, 199, 211, 229, 241.

29
Algorithm 1 Testing the sieving inequality for the existence of primitive 1-
normal pair (α, α−1 )
Input: q, n, ϑ
Output: True, if q n/2−ϑ > 2W (h)W (d)W (H)S for some h | xn − 1, d | R and H | G otherwise, False.
1: procedure test_sieve(q,n,ϑ)
rad(xn −1)
2: Construct the set A of divisors of rad(q n − 1) and set B of divisors of x−1
.
3: for d in A do
4: for H in B do
5: h=H n
6: Construct the set L1 of prime divisors rad(qd −1)
n
7: Construct the set L2 of irreducible factors of rad(x −1)
H(x−1)
and the set L3 = L2 ∪ {x − 1}
P 1 P 1 P 1
8: Compute D = 1 − p 1
− deg(p2 )
q
−
qdeg(p3 )
p1 ∈L1 p2 ∈L2 p3 ∈L3
9: if D > 0 then
|L1 |+|L2 |+|L3 |−1
10: compute S = D
+2
11: if q n/2−ϑ > 2W (d)W (H)2 S then ⊲ Sieving inequality
12: return True
13: end if
14: end if
15: end for
16: end for
17: return False
18: end procedure

Algorithm 2 Explicit search for a primitive 1-normal pair (α, α−1 )

Input: q, n
Output: True, if a pair (α, α−1 ) of primitive 1-normal elements exists in Fqn over Fq , otherwise, False.
1: procedure direct_search(q,n)
2: for β in Fqn do
3: α = βq − β
4: if α 6= 0 then
P q i n−i−1
5: construct mα (x) and mα−1 (x) ⊲ mα (x) = n−1 i=0 α x
6: if deg(gcd(mα (x), xn − 1)) = 1 & deg(gcd(mα−1 (x), xn − 1)) = 1 & ord(α) = q n − 1 then
7: return True
8: end if
9: end if
10: end for
11: return False
12: end procedure

30