Non-splitting bi-unitary perfect polynomials over F4

arXiv:2412.17850v1 [math.NT] 19 Dec 2024

with less than five prime factors

Olivier Rahavandrainy
Univ Brest, UMR CNRS 6205
Laboratoire de Mathématiques de Bretagne Atlantique

December 25, 2024

Mathematics Subject Classification (2010): 11T55, 11T06.

Abstract We identify all non-splitting bi-unitary perfect polynomials over

the field F4 , which admit at most four irreducible divisors. There is an
infinite number of such divisors.

1 Introduction
In this paper, we work over the finite field F4 of 4 elements:

F4 = {0, 1, α, α + 1} where α2 + α + 1 = 0.

As usual, N (resp. N∗ ) denotes the set of nonnegative integers (resp. of

positive integers).
Throughout the paper, every polynomial is a monic one.
Let S ∈ F4 [x] be a nonzero polynomial. A divisor D of S is called
unitary if gcd(D, S/D) = 1. We designate by gcdu (S, T ) the greatest com-
mon unitary divisor of S and T . A divisor D of S is called bi-unitary if
gcdu (D, S/D) = 1. We denote by σ(S) (resp. σ ∗ (S), σ ∗∗ (S)) the sum of all
divisors (resp. unitary divisors, bi-unitary divisors) of S. The functions σ,
σ ∗ and σ ∗∗ are all multiplicative. We say that S is perfect (resp. unitary
perfect, bi-unitary perfect) if σ(S) = S (resp. σ ∗ (S) = S, σ ∗∗ (S) = S).
Finally, we say that A is indecomposable bi-unitary perfect if A has no
proper divisor which is bi-unitary perfect.
Several studies are done about perfect, unitary and bi-unitary perfect
polynomials (see [1], [2], [3], [4], [7], [8], [9] and references therein).

1
 In this paper, we are interested in non-splitting polynomials over F4
which are bi-unitary perfect (b.u.p.) and divisble by r irreducible factors,
where r ≤ 4.
The splitting case is already treated in ([11], Proposition 3.1 and Theo-
rem 3.2). However, we better precise these results in Theorem 1.1.
We consider the two following sets:

Ω1 := {P ∈ F4 [x] : P and P + 1 are both irreducible}

Ω2 := {P ∈ F4 [x] : P, P + 1, P 2 + P + 1 and P 3 + P 2 + 1 are all irreducible}.

We see that Ω2 ⊂ Ω1 , Ω2 contains the four (monic) monomials of F4 [x] and

it is an infinite set ([6], Lemma 2). For example, for any k ∈ N, Pk :=
k k
x2·5 + x5 + α ∈ Ω2 .
We get the following two results related to the fact that A splits or not.
Theorem 1.1. Let A = xa (x + 1)b (x + α)c (x + α + 1)d ∈ F4 [x], where
a, b, c, d ∈ N are not all odd. Then, A is b.u.p if and only if one of the
following conditions holds:
i) a = b = c = d = 2,
ii) a = b = 2 and c = d = 2n − 1, for some n ∈ N,
iii) a = b = 2n − 1 and c = d = 2, for some n ∈ N,
iv) a, b, c, d are given by Table (1).

a 4 4 4 4 4 4 5 5 5 5 6 6 6 6
b 3 3 4 4 4 4 3 3 4 4 6 6 6 6
(1)
c 3 4 3 4 5 6 4 6 4 5 3 4 5 6
d 4 3 5 4 3 6 4 6 5 4 5 4 3 6

Theorem 1.2. Let A = P a Qb Rc S d ∈ F4 [x], where A does not split and

a, b, c, d are not all odd. Then, A is b.u.p if and only if one of the following
conditions holds:
i) a = b = c = d = 2, P, R ∈ Ω1 , Q = P + 1 and S = R + 1,
ii) a = b = 2, c = d = 2n − 1, for some n ∈ N, P, R ∈ Ω1 , Q = P + 1 and
S = R + 1,
iii) a = b = 2n − 1, for some n ∈ N, c = d = 2, P, R ∈ Ω1 , Q = P + 1 and
S = R + 1,
iv) P ∈ Ω2 , Q = P + 1, R, S ∈ {P 3 + P + 1, P 3 + P 2 + 1} and (a, b, c, d) ∈
{(7, 13, 2, 2), (13, 7, 2, 2), (14, 14, 2, 2)}.
Note that if a, b, c and d are all odd, then σ ∗∗ (A) = σ(A). So, A is b.u.p.
if and only if A is perfect. We also see that there exists no b.u.p. polynomial
A with ω(A) = 3.

2
 Since Ω1 and Ω2 are infinite sets, we see that there are infinitely many
indecomposable and odd b.u.p. polynomials over F4 , even if there are only
three 4-tuples available exponents.

2 Preliminaries
Some of the following results are obvious or (well) known, so we omit their
proofs. See also [10].

Lemma 2.1. Let T be an irreducible polynomial over F4 and k, l ∈ N∗ .

Then, gcdu (T k , T l ) = 1 (resp. T k ) if k 6= l (resp. k = l).
In particular, gcdu (T k , T 2n−k ) = 1 for k 6= n, gcdu (T k , T 2n+1−k ) = 1 for
any 0 ≤ k ≤ 2n + 1.

Lemma 2.2. Let T ∈ F4 [x] be irreducible. Then

i) σ ∗∗ (T 2n ) = (1 + T )σ(T n )σ(T n−1 ), σ ∗∗ (T 2n+1 ) = σ(T 2n+1 ).
ii) For any c ∈ N, 1 + T divides σ ∗∗ (T c ) but T does not.

Corollary 2.3. Let A = P h Qk Rl S t be such that h, k, l and t are all odd.

Then, A is b.u.p. if and only if it is perfect.

Lemma 2.4. If A = A1 A2 is b.u.p. over F4 and if gcd(A1 , A2 ) = 1, then

A1 is b.u.p. if and only if A2 is b.u.p.

Lemma 2.5. If A is b.u.p. over F4 , then the polynomial A(x + λ) is also

b.u.p. over F4 , for any λ ∈ {1, α, α + 1}.

Lemma 2.6. i) σ ∗∗ (x2k ) splits over F4 if and only if 2k ∈ {2, 4, 6}.

ii) σ ∗∗ (x2k+1 ) splits over F4 if and only if 2k + 1 = N · 2n − 1 where
N ∈ {1, 3}.

Remark 2.7. We get from Lemma 2.6 and for an irreducible polynomial T :
 ∗∗ 2
 σ (T ) = (T + 1)2 (i)
 σ ∗∗ (T 4 ) = (T + 1)2 (T + α)(T + α + 1)

(ii)


σ ∗∗ (T 6 ) = (T + 1)4 (T + α)(T + α + 1) (iii) (2)
2n −1 2n −1
σ (T
∗∗ ) = (T + 1) (iv)



 ∗∗ 3·2n −1
 n n n
σ (T ) = (T + 1)2 −1 (T + α)2 (T + α + 1)2 (v)

We sometimes use the above equalities for a suitable T .

3
3 Proof of Theorem 1.1
This theorem is already stated in [11]. We do not rewrite its proof.
Lemma 3.1 below completes Theorem 3.4 in [4], where one family of
splitting perfect polynomials over F4 was missing.
See [5] and [6] for the non-splitting case.

Lemma 3.1. The polynomial xh (x + 1)k (x + α)l (x + α + 1)t is perfect over

F4 if and only if one of the following conditions is satisfied:
i) h = k = 2n − 1, l = t = 2m − 1, for some n, m ∈ N,
ii) h = k = l = t = N · 2n − 1, for some n ∈ N and N ∈ {1, 3},
iii) h = l = 3 · 2r − 1, k = t = 2 · 2r − 1, for some r ∈ N,
iv) h = k = 3 · 2r − 1, l = 6 · 2r − 1, t = 4 · 2r − 1 for some r ∈ N.

Proof. Put A = xh (x + 1)k (x + α)l (x + α + 1)t . The sufficiency is obtained

by direct computations.
For the necessity, we recall the following facts in ([4], Lemma 2.7):
Lemma 3.2. Let a ∈ {h, k, l, t}. Then, a = 3 · 2w − 1 if a ≡ 2 mod 3 and
a = 2w − 1 if a 6≡ 2 mod 3, with w ∈ N.
In particular, a = 2 if (a is even and a ≡ 2 mod 3).
We take account of Lemmas 2.2, 2.3, 2.6 and 2.7 in [4], for the congruency
modulo 3 of each exponent. We get 16 possible cases according to a ≡ 2
mod 3 or not.
Moreover, by Lemma 2.2 in [4], the 3 maps x 7→ x + 1, x 7→ x + α and
x 7→ x + α + 1 preserve “perfection”.
Therefore, h and k (resp. h and l, h and t) play symmetric roles. It
remains 4 cases:
i) h, k 6≡ 2 mod 3
ii) h, l ≡ 2 mod 3 and k, t 6≡ 2 mod 3
iii) h, k, l, t ≡ 2 mod 3
iv) h, k, l ≡ 2 mod 3 and t 6≡ 2 mod 3.

The first three of them are already treated in the proof of ([4], Theorem
3.4). We got the families i), ii) and iii) in Lemma 3.1.
Now, for the case iv), we may write:

h = 3 · 2r − 1, k = 3 · 2s − 1, l = 3 · 2u − 1 et t = 2v − 1, where r, s, u, v ∈ N.

4
Compute σ(A) = σ(xh ) · σ((x + 1)k ) · σ((x + α)l ) · σ((x + α + 1)t ).
r r r
σ(xh ) = (x + 1)2 −1 · (x + α)2 · (x + α + 1)2
s s s
σ((x + 1)k ) = x2 −1 · (x + α)2 · (x + α + 1)2
u u u
σ((x + α)l ) = (x + α + 1)2 −1 · x2 · (x + 1)2
v v
σ((x + α + 1)t ) = σ((x + α + 1)2 −1 ) = (x + α)2 −1 .

Since σ(A) = A, by comparing exponents in A and those of in σ(A), we get:

2u + 2s − 1 = h = 3 · 2r − 1
2u + 2r − 1 = k = 3 · 2s − 1
2r + 2s + 2v − 1 = l = 3 · 2u − 1
2u + 2s + 2u − 1 = t = 2v − 1

It follows that s = r, u = r + 1 et v = r + 2. Thus, h = k = 3 · 2r − 1,

l = 3 · 2r+1 − 1 and t = 2r+2 − 1. We obtain the family iv).

4 Proof of Theorem 1.2

The sufficiency is obtained by direct computations. Propositions 4.1, 4.4
and 4.15 give the necessity.
As usual, ω(S) denotes the number of distinct irreducible factors of a
polynomial S.

4.1 Case ω(A) = 2

Put A = P h Qk with deg(P ) ≤ deg(Q).

Proposition 4.1. If A is b.u.p., then Q = P + 1 and either (h = k = 2) or

(h = k = 2r − 1, for some r ∈ N).

Proof. We get:

σ ∗∗ (P h )σ ∗∗ (Qk ) = σ ∗∗ (A) = A = P h Qk and ω(σ ∗∗ (P h )) = 1 = ω(σ ∗∗ (Qk )).

If h and k are both odd, then A is perfect so that Q = P + 1 and h = k =

2r − 1 for some r ∈ N.
Now, we may suppose that h is even. If h = 2, then σ ∗∗ (P h ) = (1 + P )2 .
Since P does not divide σ ∗∗ (P h ), one has Qk = (1 + P )2 and thus Q = P + 1
and k = 2. If h ≥ 4, then ω(σ ∗∗ (P h )) ≥ 2, which is impossible.

5
4.2 Case ω(A) = 3
Put A = P h Qk Rl with deg(P ) ≤ deg(Q) ≤ deg(R). Suppose that

σ ∗∗ (P h )σ ∗∗ (Qk )σ ∗∗ (Rl ) = σ ∗∗ (A) = A = P h Qk Rl .

Lemma 4.2. The polynomial P + 1 is irreducible and Q = P + 1.

Proof. The polynomial 1 + P is divisible by Q or by R, since it divides

σ ∗∗ (P h ) (Lemma 2.2). We may suppose that Q | (1 + P ). So, deg(Q) =
deg(P ) and Q = P + 1.

Lemma 4.3. One has ω(σ ∗∗ (P h )) ≤ 2, ω(σ ∗∗ (Qk )) ≤ 2, ω(σ ∗∗ (Rl )) ≤ 2.

Moreover, if h is even (resp. odd), then h = 2 (resp. h = 2r − 1, r ∈ N∗ ).

Proof. Since P does not divide σ ∗∗ (P h ), at most Q and R divide it. Hence,
ω(σ ∗∗ (P h )) ≤ 2. Similarly, we get ω(σ ∗∗ (Qk )) ≤ 2 and ω(σ ∗∗ (Rl )) ≤ 2.
- If h = 2n is even, then 2 ≥ ω(σ ∗∗ (P 2n )) = ω((1 + P )σ(P n )σ(P n−1 )). So,
n = 1.
- If h is odd. Put h = 2r u − 1, with u odd. One has:
r u−1 r −1
2 ≥ ω(σ ∗∗ (P 2 )) = ω((1 + P )2 σ(P u−1 )).

So, u = 1 because ω(σ(P u−1 )) ≥ 2 if u ≥ 3.

Proposition 4.4. If h, k and l are not all odd, then A is not b.u.p.

Proof. By Lemma 4.2, one has Q = P + 1 and so A = P h (P + 1)k Rl .

- If h, k are all even, then h = k = 2. Therefore,

(1 + P )2 (1 + Q)2 σ ∗∗ (Rl ) = σ ∗∗ (A) = A = P 2 Q2 Rl .

Hence, σ ∗∗ (Rl ) = Rl . It is impossible.

- If h is even, k odd and l even, then h = l = 2, k = 2r − 1. Therefore,
r −1 r −1 r −1
Q2 P 2 (1+R)2 = (1+P )2 (1+Q)2 (1+R)2 = σ ∗∗ (A) = A = P 2 Q2 R2 .

Hence, R divides P Q. It is impossible.

- If h is even, k and l odd, then h = 2, k = 2r − 1, l = 2s − 1. One has:
r −1 s −1 r −1 s −1 r −1 s −1
Q2 P 2 (1+R)2 = (1+P )2 (1+Q)2 (1+R)2 = σ ∗∗ (A) = A = P 2 Q2 R2 .

Hence, R divides P Q. It is impossible.

6
4.3 Case ω(A) = 4
Put A = P h Qk Rl S t with deg(P ) ≤ deg(Q) ≤ deg(R) ≤ deg(S).
We suppose that A is b.u.p. and indecomposable (i.e., neither P h Qk nor
Rl S t are b.u.p).
Lemma 4.5. One has: Q = P + 1, 1 + R = P u1 Qv1 , 1 + S = P u2 Qv2 Rz
where u1 , v1 ≥ 1 and u2 , v2 , z ≥ 0.
Moreover, if deg(R) = deg(S) then u2 , v2 ≥ 1 and z = 0.
Proof. The polynomial 1 + P divides σ ∗∗ (A) = A, so Q divides 1 + P and
thus, Q = 1 + P because deg(P ) ≤ deg(Q).
Now, 1 + R divides σ ∗∗ (A) = A, so 1 + R = P u1 Qv1 S u3 and u3 = 0 because
deg(R) ≤ deg(S). Since R = P u1 Qv1 + 1 is irreducible, we conclude that
u1 , v1 ≥ 1 and gcd(u1 , v1 ) = 1. By the same reason, 1 + S = P u2 Qv2 Rz
where u2 , v2 , z ≥ 0 and z may be positive.

4.3.1 Case deg(P ) = 1

We may suppose that P = x. Lemma 4.2 implies that Q = x + 1. Moreover,
deg(S) = 1 if deg(R) = 1. So, deg(S) ≥ deg(R) > 1.
We write: A = xh (x + 1)k Rl S t . The exponents h and k play symmetric
roles.
Lemma 4.6 ([5], Lemma 2.6). If 1 + x + · · · + x2w = U V , then deg(U ) =
deg(V ) and U (0) = 1 = V (0).
Moreover, if R and S are both of the form xu1 (x + 1)v1 + 1, then 2w = 6
and U, V ∈ {x3 + x + 1, x3 + x2 + 1}.
Lemma 4.7. If h is even, then h ∈ {2, 14}. Moreover, R, S ∈ {x3 + x +
1, x3 + x2 + 1} if h = 14.
Proof. • If h ∈ {4, 6}, then x + α and x + α + 1 both divide σ ∗∗ (xh ) and
thus, they divide σ ∗∗ (A) = A. So, A splits, which is impossible.
• If h = 2n ≥ 8, then σ ∗∗ (xh ) = (1 + x)σ(xn )σ(xn−1 ).
- If n = 2w ≥ 4, then σ(xn ) = RS because it divides σ ∗∗ (A) = A and
neither x nor x + 1 divide σ(xn ). So, by Lemma 4.6, deg(R) = deg(S) and
R(0) = 1 = S(0). From Lemma 4.5, we may put 1 + R = xu1 (x + 1)v1 ,
1 + S = xu2 (x + 1)v2 , where u1 , u2 , v1 , v2 ≥ 1. Therefore, 2w = 6 and
h = 12. But, the monomials x + 1, x + α and x + α + 1 all divide σ ∗∗ (x12 ).
It contradicts the fact that A does not split.
- If n = 2w + 1 is odd, then σ(xn−1 ) = RS and as above, n − 1 = 2w = 6.
So, h = 14, σ ∗∗ (x14 ) = (x+ 1)8 RS where R, S ∈ {x3 + x+ 1, x3 + x2 + 1}.

7
Lemma 4.8. If h is odd, then h = 2r u − 1 where r ∈ N∗ and u ∈ {1, 7}.

Proof. Put h = 2r u − 1 with u odd. One has:

r −1 r
σ ∗∗ (xh ) = σ(xh ) = (1 + x)2 [σ(xu−1 )]2 .

If u ≥ 3, then σ(xu−1 ) = RS. So, as we have just seen above, u − 1 = 6 and

R, S ∈ {x3 + x + 1, x3 + x2 + 1}.

Lemma 4.9. If l is even (resp. odd), then l = 2 (resp. l = 2s − 1, with

s ≥ 1).

Proof. • If l is even and l ≥ 4, then put l = 2n, n ≥ 2. As above, σ(Rn )

and σ(Rn−1 ) divide A.
- If n is even, then we must have σ(Rn ) = S z because P, Q divide 1 + R,
R does not divide σ(Rn ) and gcd(1 + R, σ(Rn )) = 1. Hence z = 1 and
S = σ(Rn ) is irreducible. It is impossible.
- If n is odd, then σ(Rn−1 ) = S which is impossible, as above.
r r
• If l = 2r u−1 is odd, with u odd, then σ ∗∗ (Rl ) = σ(Rl ) = (1+R)2 −1 [σ(Ru−1 )]2 .
If u ≥ 3, then σ(Ru−1 ) = S, which is impossible.

4.3.2 Case deg(P ) > 1

Several proofs are similar to those in Section 4.3.1. As above, Lemma 4.2
implies that Q = P + 1. We write: A = P h (P + 1)k Rl S t .

Lemma 4.10. If 1 + P + · · · + P 2w = RS, then deg(R) = deg(S), 2w = 6,

P ∈ Ω2 and R, S ∈ {P 3 + P + 1, P 3 + P 2 + 1}.

Proof. Suppose that 1+P +· · ·+P 2w = RS. One has 1+x+· · ·+x2w = U V
where U (P ) = R and V (P ) = S. By Lemma 4.6, one has: U (0) = 1 =
V (0), deg(U ) = deg(V ). So, deg(R) = deg(S).
Moreover, U and V must be of the form xu (x + 1)v + 1. Indeed, if 1 + U =
xu1 (x + 1)v1 Lz , with z ≥ 0, then 1 + R = P u (P + 1)v L(P )z , L(P ) = S y ,
y ≥ 1, deg(S) = deg(R) = u deg(P ) + zy deg(S), zy = 0. Thus, z = 0 and
1 + U = xu1 (x + 1)v1 . Analogously, 1 + V = xu2 (x + 1)v2 . Therefore, by
Lemma 4.6, 2w = 6 and R, S ∈ {P 3 + P + 1, P 3 + P 2 + 1}.

Lemma 4.11. If h is even, then h ∈ {2, 14}.

Proof. • If h ∈ {4, 6}, then P + α and P + α + 1 both divide σ ∗∗ (P h ) and

thus, they divide σ ∗∗ (A) = A. So, P , P + 1, R = P + α and S = P + α + 1
are all irreducible over F4 , which is impossible.

8
• If h = 2n ≥ 8, then σ ∗∗ (P h ) = (1 + P )σ(P n )σ(P n−1 ).
- If n = 2w ≥ 4 is even, then σ(P n ) = RS, deg(R) = deg(S). We obtain
2w = 6 and h = 12.
But P +1, P +α and P +α+1 all divide σ ∗∗ (P 12 ). As above, it is impossible.
- If n = 2w + 1 is odd, then σ(P n−1 ) = RS and n − 1 = 2w = 6. So, h = 14
and R, S ∈ {P 3 + P + 1, P 3 + P 2 + 1}.

Lemma 4.12. If h is odd, then h = 2r u − 1 where r ∈ N∗ and u ∈ {1, 7}.

Proof. Put h = 2r u − 1 with u odd. One has:

r −1 r
σ ∗∗ (P h ) = σ(P h ) = (1 + P )2 [σ(P u−1 )]2 .

If u ≥ 3, then σ(P u−1 ) = RS and as we have just seen above, u − 1 = 6 and

R, S ∈ {P 3 + P + 1, P 3 + P 2 + 1}.

We also get the analoguos of Lemma 4.9.

Lemma 4.13. If l is even (resp. odd), then l = 2 (resp. l = 2s − 1, with

s ≥ 1).

4.3.3 End of the proof

We recapitulate below, for P ∈ Ω2 , Q = P + 1, R = P 3 + P + 1 and
S = P 3 + P 2 + 1, the expressions of σ ∗∗ (T z ), for T z ∈ {P h , Qk , Rl , S t }.
Keep in mind that h, k, l and t are not all odd.

h σ ∗∗ (P h ) k σ ∗∗ (Qk )
2 Q2 2 P2
14 Q8 RS 14 Q8 RS (3)
r s
2r − 1 Q2 −1 2s − 1 P 2 −1
r r r s s s
7 · 2r − 1 Q2 −1 R2 S 2 7 · 2s − 1 P 2 −1 R2 S 2

l σ ∗∗ (Rl ) t σ ∗∗ (S t )
2 P 2 Q4 2 P 4 Q2 (4)
e e f f
2e − 1 P 2 −1 · Q2·(2 −1) 2f − 1 P 2·(2 −1) · Q2 −1

We compare from Tables (3) and (4), the exponents of P, Q, R, S in σ ∗∗ (A)

and in A. Instead of considering several possible cases, we give an upper
bound to each exponent a ∈ {h, k, l, t}. We use Maple computations to
determine those which satisfy σ ∗∗ (A) = A. We obtain the following results.

9
Lemma 4.14. - If h and k are both even, then h, k ∈ {2, 14} and e, f ≤ 3.
So, l, t ∈ {1, 2, 3, 7}.
- If h is even and k odd, then h ∈ {2, 14} and s, e, f ≤ 3. So,
k ∈ {1, 3, 7, 13, 27, 55} and l, t ∈ {1, 2, 3, 7}.
- If h and k are both odd, then (h, k, l, t) ∈ {(3, 7, 2, 2), (7, 3, 2, 2)}.

Proof. - If h is even, then h ≤ 14. Each exponent of P in the tables equals

at most 14. So, s, e, f ≤ 3.
- If h = 2r − 1 and k = 2s − 1, then σ ∗∗ (P h Qk ) = P k Qh , h = k and P h Qk
is b.u.p. Hence, Rl S t is also b.u.p. and l = t = 2.
s s
- If h = 2r − 1 and k = 7 · 2s − 1, then only R2 and S 2 divide σ ∗∗ (A) = A.
So, s = 1, l = t = 2, k = 13. Thus, σ ∗∗ (Rl S t ) = P 6 Q6 . By comparing the
exponents of Q in the tables, we get 6 + 2r − 1 = k = 13. So, r = 3 and
h = 7.
Analogously, if h = 7 · 2r − 1 and k = 2s − 1, then h = 13, k = 7, l = t = 2.
r s r s
- If h = 7 · 2r − 1 and k = 2s − 1, then only R2 +2 and S 2 +2 divide
σ ∗∗ (A) = A. So, l = t = 2 and we get the contradiction: 2r + 2s = 2 with
r, s ≥ 1.

We also remark that the values of the exponents h, k, l and t do not

depend on the choice of P ∈ Ω2 . Therefore, for the computations with
Maple, we took two values of P : P = x and P = x2 + x+ α.

Proposition 4.15. If A = P h Qk Rl S t is b.u.p and indecomposable, where

h, k, l and t are not all odd, then

P ∈ Ω2 , Q = P + 1. R, S ∈ {P 3 + P + 1, P 3 + P 2 + 1},

and (h, k, l, t) ∈ {(7, 13, 2, 2), (13, 7, 2, 2), (14, 14, 2, 2)}.

4.3.4 Maple Computations

We search all A = P h Qk Rl S t such that h, k, l, t are not all odd, ω(A) ≥ 3
and σ ∗∗ (A) = A, by means of Lemmas 4.8, 4.9 and 4.14. We get the results
stated in Proposition 4.15.

The function σ ∗∗ is defined as Sigm2star

> Sigm2star1:=proc(S,a) if a=0 then 1;else if a mod 2 = 0

then n:=a/2:sig1:=sum(S^l,l=0..n):sig2:=sum(S^l,l=0..n-1):
Factor((1+S)*sig1*sig2) mod 2:
else Factor(sum(S^l,l=0..a)) mod 2:fi:fi:end:

10
> Sigm2star:=proc(S) P:=1:L:=Factors(S) mod 2:k:=nops(L[2]):
for j to k do S1:=L[2][j][1]:h1:=L[2][j][2]:
P:=P*Sigm2star1(S1,h1):od:P:end:

References
[1] J. T. B. Beard Jr, Unitary perfect polynomials over GF (q), Rend.
Accad. Lincei 62 (1977), 417–422.

[2] J. T. B. Beard Jr, Bi-Unitary Perfect polynomials over GF (q), An-

nali di Mat. Pura ed Appl. 149(1) (1987), 61–68.

[3] E. F. Canaday, The sum of the divisors of a polynomial, Duke Math.

J. 8 (1941), 721–737.

[4] L. H. Gallardo, O. Rahavandrainy, On perfect polynomials over

F4 , Port. Math. (N.S.) 62(1) (2005), 109–122.

[5] L. Gallardo, O. Rahavandrainy, Perfect polynomials over F4 with

less than five prime factors, Port. Math. (N.S.) 64(1) (2007), 21–38.

[6] L. H. Gallardo, O. Rahavandrainy, All perfect polynomials with

up to four prime factors over F4 , Math. Commun. 14(1) (2009), 47–65.

[7] L. H. Gallardo, O. Rahavandrainy, On unitary splitting perfect

polynomials over Fp2 , Math. Commun. 15(1) (2010), 159–176.

[8] L. H. Gallardo, O. Rahavandrainy, On splitting perfect polyno-

mials over Fpp , Int. Electron. J. Algebra 9 (2011), 85–102.

[9] L. H. Gallardo, O. Rahavandrainy, Unitary perfect polynomials

over F4 with less than five prime factors, Funct. et Approx. 45(1)
(2011), 67–78.

[10] L. H. Gallardo, O. Rahavandrainy, All bi-unitary perfect poly-

nomials over F2 only divisible by x, x + 1 and by Mersenne primes,
arXiv Math: 2204.13337 (2022).

[11] L. H. Gallardo, O. Rahavandrainy, On splitting bi-unitary per-

fect polynomials over Fp2 , arXiv Math: 2310.05540v3 (2023).

11