J
OURNAL DE
T
HÉORIE DES
N
OMBRES DE
B
ORDEAUX
S
TANISLAV
J
AKUBEC
Note on the jacobi sum J(χ, χ)
Journal de Théorie des Nombres de Bordeaux, tome
7, n
o2 (1995),
p. 461-471
<http://www.numdam.org/item?id=JTNB_1995__7_2_461_0>
© Université Bordeaux 1, 1995, tous droits réservés.
L’accès aux archives de la revue « Journal de Théorie des Nombres de Bordeaux » (http://jtnb.cedram.org/) implique l’accord avec les condi-tions générales d’utilisation (http://www.numdam.org/conditions). Toute uti-lisation commerciale ou impression systématique est constitutive d’une infraction pénale. Toute copie ou impression de ce fichier doit conte-nir la présente mention de copyright.
Article numérisé dans le cadre du programme Numérisation de documents anciens mathématiques
Note
onthe
Jacobi
sumJ(~,~).
par STANISLAV JAKUBEC
Notation
X - the Dirichlet character modulo
- Jacobi sum
- Gaussian sum
Introduction
be the Jacobi sum,
Q«(z).
It is well known that~ =
p, and oneeasily
proves thatThe main aim of this paper is to solve the
problem:
When isX)
upto association and
conjugation uniquely
determinedby
the solution of theequation
We
give
acomplete
solution in cases 1= 11, 19.
On the basis of this
result,
thefollowing
question
is answered:When is the
prime
2 an 11-th resp. a 19-th power modulo p if p is notrepresentable by
thequadratic
formX2 +
l ly2 ,
resp.x2
+19y2.
1991 Mathematics Subject Classification. Primary 11R18.. Manuscrit requ le 19 janvier 1993
We shall now
present
a survey of results obtainedby
solving
theproblem
when the
prime
2 is an 1-th power modulo p.Jacobi has
given
necessary and sufficient conditions forprimes q
37 to be cubes moduloprimes
p = 1(mod 3).
Forexample,
he proves thefollowing
PROPOSITION 1. ~ is a cube modulo
p if
andonty
0(mod 2),
whereEmma Lehmer
[2]
finds thefollowing
result:PROPOSITION 2. Let p = 1
(mod 5)
be aprime.
Then 2 is afifth
powermodulo p
if
andonly
if
x = 0(mod 2),
where(x,
u, v,w)
is oneof
theex-actly
four
solutions(x,
u, v,w), (x,
-u, -v,w),
(x,
v, -u,-w), (x,
-v, u,-w)
of
thediophantine
system
(Dickson):
P.A. Leonard and K.S. Williams
[4]
prove thefollowing
PROPOSITION 3. Let p = 1
(mod 7)
be aprime.
Then 2 is a -seventh powermodulo p
if
andonly
if x, =-
0(mod
2),
where(xl, x2, ... , z6 )
is oneof
theexactly
six non-trivial solutionsof
thediophantine
system
of
equations
PROPOSITION 4. Let p - 1
(mod 11)
be aprime.
Then 2 is an eleventhpower modulo p
if
andonly if
a certain conditioninvolving
solutionsof
a verycomplicated
diophantine
system
holds(the
exact statement may beseen in
[31).
J.C.
Parnami,
M.K.Agrawal
and A.R.Rajwade
[5]
have thefollowing
PROPOSITION 5. Let p = 1
(mod l).
Then 2 is an l-th power rriodulo pif
and
only if
where
(al,
a2, ... ,ai-I)
is oneof
theexactly
l -1 solutionsof
thediophan-tine
system
of
equations
where
A(n) is
the leastnon-negative
residueof
n modulol,
and gk is theautomorphism (I
->(f,
Now let XX = p, and let
X)
be associated with the numberX,
i.e.J(x,
x)
=eX,
where E is a unit of the field Thenimplies
e"£ =1 and hence c =(2013~)~.
So we haveLet 2 be a
primitive
root modulo l. Consider a residue class fieldZ((1) /(2)
of thedegree
f
= 1 - 1 overZ/2Z.
Let g
be agenerator
ofthe
multiplicative
group(Z«(I)/(2»*
of the fieldZ(~l)/(2)
such that thereLEMMA 1.
Every
unit Eof
thefield
Q(~1)
is power in the group(Z«(z)j(2»*,
m =1;1.
Proof.
Let e =g’~
(mod
2).
It is necessary to prove that, in.
Considerthe unit hence 6"1 must be a root of
1,
thereforeell
= 1.Further,
It follows that
2Ln(2"‘ - 1) -
0(mod
2l-1 -
1),
and therefore n - 0(mod 2ml+l). 0
LEMMA 2. Let 2 be a
primitive
root modulo l. For a natural number a,0 a L - 1 the
following identity
holds(the
decomposition
intobinary
system),
where r,, =- l = 2 -ind(n)
+ind(a)
(mod
l-l),
0 rnl -1,
andind(x)
is the indexof
the element x in the group(Z/lZ)*
under the base2,
i.e.(mod l).
Proof.
The lemma can bereadily proved
when the rational numbera
is"0
expressed
in thebinary
system, ’
= E
an2 -n.
n=l
LEMMA 3. The
factorisation of
the Jacobi sumJ(X,
X)
intoprime
divisors1-1
11
is aprime
divisor of
n=1 n
1
the
field
Q(l), pip,
and is anautomorphism
=Cn
n n
Proof. According
to[1],
I-I
The factorisation
11
is obtainedusing
theConsider a divisor
where ji
=0; 1,
and defineLEMMA 4. The
factorisation .
is aconjugation of
thefactori-Proof.
Let A be aconjugation
ofJ(X,
X).
Then A =X-)
for some s.If s = k
then we can write2.
Conversely,
letBy
Lemma2,
Since the
expansion
of a number in thebinary
system
isuniquely
deter-mined,
Lemma 4 isproved.
DFor
pip,
denoteby hi,
the least natural number such that aprincipal
divisor =
(a).
THEOREM 1. Let 2 be a
primitive
root and tet a =gm
(mod
2),
If
then X
is,
up to association andconjugation, equal
to the Jacobi sumJ(x, X).
Proof. According
to Lemma1,
the choice of agenerator
a of theprincipal
divisor =
(a)
is not substantial.Suppose
the factorisation of X intoprime
divisors of the fieldQ«(l)
isIt is necessary to prove that this factorisation is a
conjugate
of thefactori-Clearly
hence
where e is a unit of the field
Q(~1).
But a = 1
implies
e =(-(1 ) ,
and thereforeFrom X = 1
(mod
2)
we obtainhence
Consequently
It is easy to prove that the condition XX = p
gives
From the congruences
(1)
and(2),
using
theassumption
(M, 2 ~+i ) -
1and the fact that
(2"’~ -
1, 2m
+1)
=1,
we obtain the congruenceshence
From XX = p, it follows
1,
and thisimplies
a l - 1. Due toLemma
4,
Theorem 1 isproved.
RRemark. If
(M,
+)
= d >1,
then instead of the congruence(3)
weget
the congruence
It can be
proved
that this congruence hasalways
the solutionwhich is not
corresponding
to theconjugates
of the Jacobi sumJ(x, x~ ~
The
question
of whether for p with d > 1 the Jacobi sumJ(x,
x)
can beuniquely
determined transforms in thequestion,
forwhich ji
the divisorCOROLLARY 1. Let 2 be a
primitive
root modulo1,
and let the class numberof
thefield
Q«(l)
beequal
to 1. Then the Jacobi sumJ(X, X)
isuniquely
determined, up
to association andconjugation,
from
the solutionof
theequation
XX =p, X
=1(mod
2) X
EZ(~1)
if
andonly if
(M,
=1.Proo f.
This follows from thepreceding Remark,
because in such a case,every divisor is
principal.
EXAMPLE 1. 1 =
5,
=22t1
= 1. Itfollows
thatfor
all p one hasd =
1,
hence the Jacobi sumJ(X, X)
isuniquely determined,
up toasso-ciation and
conjugation,
by
the solutionof
theequation
XX = p, X ! 1(mod
2),
X EIt is easy to see, that if we want to answer the
question,
for whichprimes
p is the Jacobi sum
J(X,
X)
uniquely determined,
up to association andconjugation,
by
the solution of theequation
XX = p X -1(mod 2),
X EZ(CI),
then we must know for whichprimes
p the number a, whereN(a) =
p, is a third power in the group(Z(~l)/(2))*.
Thisquestion
is solved in thefollowing
lemma.LEMMA 5. Let 2 be a
primitive
root modulo 1 =- 3(mod 4), plp,
andphp
=a.
Then a is a third power in the groups
(Z(~l)/(2))*
if
andonty
=X2
+ly2,
where x, y are notsimultaneously
divisiblebg
p.Proof. By
Lemma1,
the choice of agenerator
a of theprincipal
divisorphp
is not substantial. Consider the
product
Let
So we have a’ = b’ = 1
(mod
2).
Hence
therefore
Let
conversely
This
implies a’ =
b’
= 1(mod 2),
henceLet Then since
,B
is invariant onwhere (7)
=1,
we set thatBut if
(7)
_-1,
then7n{p)
does notdivide!3
(in
theopposite
case wewould
get
hence acontradiction).
It
implies
therefore
and
From this we
finally
obtainTHEOREM 2. Let 1 =
11; 19,
and let p = 1(mod l),
4p
=A2
+ lB2.
TheJacobi sum
J(X, X)
isuniquely
determined,
up toconjugation
andassocia-tion,
by
the solutionof
if
andonly
(mod 2).
Proof.
Theproof
follows from Lemma 5 andCorollary
1. DBy Proposition
5 and Theorem2,
we come to thefollowing
THEOREM 3. Let I =
11; 19,
and letp - 1
(mod 1),
4p
= A2
+1B2,
A - B - 1
(mod 2).
Then 2 is an l-th power modulop if
andonly if
where
(al,
a2, ... ,al-1)
is oneof
theexactly
1 - 1 solutionsof
thediophan-tine
system
of
equations
Remark. As we can see, Theorem 2 enables us to remove condition
(iii)
ofProposition
5.REFERENCES
[1]
H. HASSE, Vorlesungen uber Zahlentheorie, Berlin 1950.[2]
E. LEHMER, The quintic character of 2 and 3, Duke math. J. 18(1951),
11-18.[3]
P. A. LEONARD, B. C. MORTIMER and K. S. WILLIAMS, The eleventh power[4]
P. A. LEONARD and K. S. WILLIAMS, The septic character of 2,3,5 and 7, Pacific J. Math. 52(1974),
143-147.[5]
J. C. PARNAMI, M. K. AGRAWAL and A. R. RAJWADE, Criterion for 2 to be l-th power, Acta Arithmetica 43(1984),
361-364.Stanislav JAKUBEC Mathematical Insti.tute of Slovak Academy of Sciences Stefanikova 49
814 73 Bratislava Slovakia