Sci. Bull. Fac. Educ., Nagasaki Univ., No. 47, pp. 1 9 (1992)
Explicit formulas of L-functions of some hyperelliptic curves To the Memory of Professor Koichi Yamamoto
Tadashi WASHIO and Tetsuo KODAMA*
Department of Mathematics, Faculty of Education,
Nagasaki University, Nagasaki 852, Japan
(Received Feb. 29, 1992)
Abstract
Let F=GF(p) be a prime field of characteristic p>2. Let g be a positive integer.
Denote by P(x) a polynomial over F of the form P(x)=x2gF1+1 where (2g-f-1, p)
=1 or of the form P(x)=x(x2g+1) where (2g,p)=1 and by K= Q= (x, a hyperelliptic
function field defined by y2=P(x) over F. Assume that, in the case P(x)-=x2g+1+
1, there exists nEN satisfying —1 (mod 4g+2) and that, in the case P(x) = x(x2g+1), there exists nEN satisfying —1 or 1 +2g (mod 4g).
Then, it is shown that the L-function L(u) of K is given by
L(u).-= 1 + (pu2).,/2} b,
J=1
for suitable positive integers a, b, and even nj.
The proof is done by calculations of the number of points on the curves in suitable constant field extentions of K and by the MObius's inversion formula instead of using
the techniques of Weil.
I . INTRODUCTION
Let F=GF(ci) be a finite field of characteristic p>2. Let K be an algebraic function field of genus g>0 over F. We will denote by L(u) its L-function, i.e., the numerator
of its zeta-function Z(s). Then it is put in the form
L(u)=ao+aiu+a2u2+•••+a2gu2g — w
iu) (1-0)2u) ••• (1—w2gu) where a,EZ , a0=1, a2g_i=e-' a, (i=0,...g) and a),EC
The so-called Riemann hypothesis means that co, can be written as
Department of Mathematics, College of General Education, Kyushu University,Fukuoka 810, Japan
ωε=γ縄1/2, 1γ」=1(∫=1,2,...,2g)(see Eichler[1],Ch.V,§4and Hasse
[2],Ch.IV).
Let us assume thatκ=「(T,y)is a hyperelliptic function fdd defined byッ2=P(3じ)
where P(劣〉is a polynomial over「of the form
P(記);〆8+1十αwith(28十1,ρ)=1andα∈『×or P(コじ)=3じ(κ28十α)with(28,p)=1andα∈「×.
Then,in [7],we have proved that allγ1,γ2,.の.,γ28are roots of unity if and only ifthereexistsπ∈N satisfying pη…一1(mod4g十2)inthecaseP(3じ)ニr28+1
十αor satisfy重ng pπ……一10r1十28(mod48・)in the case P(r)=π(∫28十α).
We will study the expliciHormulas of the L(%) s for these cases which have a decomposition with integral polynomial factors and are useful to calculate the class
number ofκ.In this paper,we will determine such formulas in the essential case of¢ニp andα=1.These formulas are given by
L(鋤)一H{1+(P秘2)π∫/21δノ ノニ1
for suitable positive integers α, ゐノ and even ηノ.
In [9],・A.Weil has already considered curves of the form ye=γπ「十δ,where γ,δ∈GF(p処〉satisfyingp/εブ,andshown
L(%)一■五、,6(μ),whereLα,δ(%)一1+ζノ〆 α,δ
with a root of unityζ and a Jacobi sumブfor suitable integers α,6,d.Our proof is done by calculations of the number of points on the curves in suitable constant field extentions of K and by using the Mδbius s inversion formula instead of using
the techniques of WeiLTo do so,we will state several notaions and some lemmas concerned at them in
§2and prepare some lemmas for the Jacobi sums in §3.The desired results about
L(秘) s for P(T)=記29+1十1 and P(偲)=T(¢29十1) will be given in §§ 4and5
respectively. , 2. Notations and Some Lemmas
In this section we will give some notations and their properties which will be used
in §§4 and5..Let8be a positive integer and p a prime number.Putプ=28十1 0rプ=2&Then we assume that there exists,η∈N sati5fying P皿…一1(mod21)for プ=29十10r that there exists伽∈N satisfying Pη…≡一10r1十プ(mod21)forノ=28.Moreover denote byゐthe minimum of such,ガsl thus2んmeans the order of p modulo
2五We write,for everyπ∈N,
ε.=(Pπ一1,2か,δ.=(P』1,の.
Explicit formulas of L‑functions of some hyperelliptic curves 3
It is clear that if n m(mod 2k) or n= =e and = . Moreover
(m, 2k) then e
the equality e =2 is equivalent to ord2 f, where ord2z means the (p"‑1) > ord2
number of times 2 dividing z.
Now we set
D= { ; ne J e =26 >2 }
and o! = #D(the cardinal number of the set D).
Furthermore, for each djeD, we denote by nj the minimum of n's such that e = 2 =2dj. By renumbering, we may assume that
nl< n2<... < n and we put
N= {nl, n2,..., n.} and D= {dl, d2,..., d.}, where dj= j (1 j a ).
Clearly njl2k, djlf, n.=2k and d.=f. These definitions lead to the following lemma.
LEMMA 2.1. (i) If njeN then nj is the minimum of n's satisfying p"=1 (mod 2d )
( ii) Let di, djeD; then dil dj < > ni I nj.(iii) Let djeD and n e i; then d 1 < > n I n
Since we can easily prove that njl 2k and 2njJ 2k for njeN, we see the followirg
lemma.LEMMA 2.2. ord2nj=0rd 2k for all n EN especlally 2 1 n
LEMMA 2 3 (1) If p I (mod 2 then p"/2 1 (mod 2d ) and e j/2=2 for all
nj e N.
(ii) If f=2g aud pk 1+f (mod 2h for all njeN.
then p"j/2 EI +dj (mod 2dj) and j/2 =dJ
PROOF. Because of 21ljX 2k, we can put 2k=2nju+2v (u, v e : , 0<v <nj).
From njf 2k we get njl 2v and so v=nj/2; consequently pksp"j/2(mod 2dj).
Case (i) Obviously p"j/2 ̲1 (mod 2dj). Also, since nj/2 1 k and ek=2, we have
e j/2 = 2 .
Case (ii) Evidently p"j/2 1 +f (mod 2dj). Since djl f and nj is the minimurn of n's
such that p" 1 (mod 2dj) we get p"j/2E:EI +dj(mod 2dj). Thus we see' djl j/2 16 j and hence dj= j/2
We will now deflne the notatlon N(n) for n E I by N(n) = {njeN; njln}.
Then the following lemma follows from Lemma 2.1.
LEMMA 2.4. If N(n)
6 and N(n) = N(nj).
and nj is the l arge s t
We can regard the set N=(nl, n2,"', n.} as
to the divisibility relation and so we can define
i. e. ,
( i ) P (x,x) = I (xeN), (ii) P (x,y) = O (x,yeN, xXy),
(iii) p (x,z) = O (x,yeN, x l
'eN=* I ',' I y
Then, by making use of the Mobius's inversion lemma.
number
contained
a partially the Mobius
y, x y).
f ormula,
in N(n) then
ordered set with
function p (x,y)we obtain
thed= j
respect
on N,
following
LEMMA 2.
" ' Xa) X2, '
5. Let dl ' of a system
da
d 2 , " "
of linear
nie N( n j ) ' n
‑ e a given real
equations
ixi = dj‑ (j=1,2
numbers.
,, ., )
a
Then the solution (xl,
is given by
xj ‑
1 n j
T
njeN(nj)
p(n,, nj) di‑ j=1 , 2,. ., ) a
3. Jacobi Sums
In this section we will give some lemmas for the Jacobi sums. Let [ be a finite field. Then a Jacobi sum in O with respect to multiplicative characters o and ,,
of is given byJ(c,v)= o(u)v(v)
+.=1
with the summation extended over all pairs (u, v) of elements of [ satisfying u+v=1.
The general theory of the Jacobi sums may be found in B.C. Berndt and R.J. Evans [1]. H. Davenport and H. Hasse[2], K. Ireland and M. RosenL5] and R. Lidl and H.
Niederreiter[8]. In [7], we have known the following r sults. ([7],p.192 and p.196).
LEMMA 3. I .
Suppose that character of
LEMMA 3.2.
Suppose that
Let p d>1 is
order d
Let p d>1 is
be an
o f
be
an
a prime number and n a positive integer.
odd integer satisfying p ‑1(mod 2d). If A
If t¥i then
J(Aj, Aj) =p" for j=1,2,..., d‑1.
a
prime number and n a positive
even integer satisfying p " E‑1integer.
or I +d
Let O .¥ = GF ( p2 ).
is a multiplicative
Let =GF (p2 ).
(mod 2d). If A is
Explicit formulas of L‑functions of some hyperelliptic curves 5 a multiplicative character of o.rder 2d and 7 is the ( uadratic character of [ then
A2j 1(̲1) J(A2j+1, )= " .. d‑1. p forj=0,1, .,
4 L Functlon for P(x)=xf+ l
Let rrl =GF (p) be a prime field of characteristic p>2 and K= IF(x;y) a hyperelliptic function field defined by y2=xf+1 where f=2g+1 and (f, p)=1. We will denote by
L (u) its L‑function and put
L (u) =(1 ‑(vlu) (1 ‑w2u) "' (1 ‑a'2gu).
Let L (u)=ao( ) +al( )u +a2( )u2+... +a2g( )u2g be the L‑function of the constant
field extension of K of degree n. Then it is well‑known thatL (u) (1 a' u) (1 co u) (1 w2"u) (see L3], [4]).
Now we assume that there exists me i such that p ;‑1 (mod 2h and denote by k the minimum of such m's.
LEMMA 4.1. Let notations be same as in S 2 and let n er } (i ) If N (n) = then al( ) =0.
(ii) If N (n) then al( ) =(‑1) "/ ‑1(d‑1) p"/2 where m is the largest number contained in N(n) and d=6 .
PROOF Put q=p" and =GF(q). Then, as is well known, al( ) is given by
d ‑ 1
al( )= V (vf+1)= ' Aj(‑4)J(Aj,Aj)
'e , j=1
where 7 is the quadratic character of O and A is a multiplicative character of [
of order d= . Clearly, if N(n)= then d= I and so al( )=0.Suppose that N(n) . Since Lemma 2.4 Ieads to d= = we obtain
J(Aj,Aj) = (‑1) / I J((1J (T')"/
where (1 is a multiplicative character of GF(p ) of order d such that o is lifted
to A ; A=6. Norm.Then, from Lemmas 2.2 and 2.3, we see 2 1 m and p /2= ‑1 (mod 2d). So it follows from Lemma 3.1 and cj(‑4)=1 that
(rj(‑4) J (cj,Cj)=p /2 Therefore we get
A'( 4)J(AJ Aj) = (‑1)"/ ‑1{oj(‑4)J(cJ crJ)}"/
= ‑1)"/ ‑1 p"/2
and hence
al( ) = (‑1)"/ ‑1 (d‑1) p"I2
6 Tadashi WASHIO and
Tetsuo
KODAMATHEOREM 1. Let Jf"=GF(p) be a prime field of characteristic p>2 andK= fJ'(x; y)
a hyperelliptic function field defined by y2=xf+1 where f=2g+1 and (f, p) = I . Assume that there exists mef l such that p E‑1 (mod 2 . Let notationsN= { nl, n2,...,
n. } aud D= { dl, d2,"', d. }
be same as in S 2. Then the L‑function of K is given by L(u) = II { 1+(pu ) 7 } j, 2 /2 b
j=1
where bj= I p (n,, n ) (d 1) (J 1 2,...,a )
l
n j
"ieN( j)
PRooF. From Lemma 3.5 and the definition of bi, we see
' bi ni=d j ‑ I (J I ,2. . . , a).
"ieN( j)
For ner 1 we will define c by
O if N(n) = ,
( 1)"/ i‑1bini if N(n) .
c‑" "/2 ‑
"ieN( )
In the case N(n) , if we denote by nj the largest number contained in N(n),
then, from Lemma 2.4, we have N(n)=N(nj). Since Lemma 2.2 Ieads to ord2 ni=ord2 nj for nieN(nj) we get n/ni E n/nj (mod 2). Thus c = (‑1) "/ j‑1 p"/2 .bi ni
"ieN( j)
=
‑1) "/ j‑1 (dj‑1) p"/2.
Thus Lemma 4.1 shows that c =al( ) for ne 1 Hence It follows that log L(u)= ‑ (a'l"+ "'+ co2g") u"/n *
=1
=
al( )u"/n = c u"/n
=1 =1
=
j=1 ,=1
==
log {1 +(pu2)"j/2}
j=1
and so we have the desired formula.
The notations being as in Theorem 1,
the straightforward calculations.Numerical
example.
Let g=269 andthe
following
p= 61. Then
results
f = 539,
follow at
k = 105,
once f rom
a 5 N={6, 10,
Explicit formulas of L‑functions of some hyperelliptic curves 7
30, 42, 210}, D=}, D= { 7, Il, 77, 49, 539}, and so
L(u)=(1 +p3u6) (1 +p5uro) (1 +pl5u30)2 (1 +p u ) (1 +pro5u210)2 21 42
COROLLARY l. If k=2'(r 0), then a =1, N { 2k } D { f } aud
L(u) = { I +(pu2)k} glk
COROLLARY 2. If f is a prime number satisfying f p and the ord r of p modulo 2f is even, then k is the half order of p modulo 2f, a=1, N= { 2k } , D= {f}, and
L(u) = { I +(pu2)k} g/k
COROLLARY 3. If f=2'(2 : odd prime, r>0) and p is a primitive root modulo 2f, then k= (2‑1) 2'‑1/2, a=r, N= {2‑1,..., (2‑1)2j‑1,..., 2k}, D= {2,..., j
..., f} and .
L(u) = 11 { I +(pu2)" /2}j=1
where nj= ( ‑1) 2j=1
5. L‑Function for P(x)=x(xf+ I )
Let =GF(p) be a prime field of characteristic p>2 andK= O'(J,;y) ahyperelliptic
furiction field defined by y2=x(xf 1) where f=2g and (f, p)=1, We will denoteby L (u) its L‑function.
Let L (u)=ao( )+al( )u +a ( )u2 +...+a2 (*)u2g be the L‑function of the constant field extension of K of degree n.
Throughout this section, we assume that there exists m e ¥ .1 such that p ;‑1 or
1 +f (mod 2h and denote by k the minimum of such m's.LEMMA 5.1. Let notations be same as in S 2 and let n e[¥¥1.
(i) If N(n) = '
hen al( ) = O.
(ii) If N(n) then al( ) = (‑1) "/ ‑1 d p"/2
where m is the largest number contained in N(u) and d .
PROoF. Put q=p" and =GF(q). Then, as is well known, al( ) is glven by the
Jacobsthal sum
al( ) = ' 7 (v(vf+1))
e *,
where 7 is the quadratic character of i . Thus if N(n) = then al( ) O and if N(n) then
al( ) = ' A2j+1(‑1)J(A2 +1 7 )
d‑1 j=0where A is a multiplicative character of l} of order 2d=2 .
In the case N(u) , Lemma 2.4 gives us d= = and so we have
J(A2j‑1,7)= (‑1) "/ ‑1 J(02jtl v) "/
where v is the quadratic character and o is a multiplicative character of GF(p ) of order 2d such that o is lifted to A ; A = cF' Norm.
Then, from Lemmas 2.2 and 2.3, we see 2lm and p /2= I or 1+d (mod 2d) So Lemma 3.2 Ieads to
62j+1(‑1) J((7 2j 1 v ) p 12 Therefore we obtain
A2j+1(‑1) J(A2j+1,7 ) = (̲1)"/ ‑1 {c2j 1(‑1) J (02j 1 v) } "/
= ‑1)"/ ‑1 p"/2
and so we get the desired assertion
al( ) = ( 1) "/ =1 d p"/2
The proof of the following theorem is same as the proof of Theorem I with the exception of replacing Lemma 4.1 by Lemma 5.1.
THEOREM 2. Let =GF(p) be a prime field of characteristic p>2 and K= If(x,y) a hyperelliptic function field defined by y2=x(xf+1) where f=2g and (f, p) =1 . Assume that there exists meT J such p :‑1 or 1+f (mod 2h. Let the notations
, ', n.} aud D={ dl, d2,..., d.}
N= {nl, n2 ..
be same as in S 2. Then the L‑function of K is given by L(u) = 11 {1 +(pu2)" /2} b
j=1
where bj = i ' p(ni, nj) di (j=1 2 a )
nj ieN( j)
Let the notations be as in Theorem 2. Then we can also get easily the follo ving results.
COROLLARY I If k=2"(r O); then a=1, N= { 2k }, D= { f} , and L(u) = { 1+ (pu2)k}g/k.
COROLLARY 2 If 2g=p2" 1, (r>0), then k=2" and
L(u) = { I + (pu2) k} glk.
COROLLARY 3. If g=P ( odd prlme) and k>1 then a 2 N { 2 2k} D
{ 2, f }, and
L(u) = (1+pu ) { 1+ (pu2)k } (g 1)lk
1
2
3
4
5
6
7
8 9
Explicit formulas of L‑functions of some hyperelliptic curves 9 CoRoLLARY4. If 2g=pg'‑1, ( : odd prime, r>0), then k=2・, a r+1 N {2
2pi,..., 2k }, D= { p‑1,..., pP'‑1,...f}, and
L(u)= h { I +(pu2)2'}b where b (p ‑ pPi 1) /22 i. i=0
Remark. The fact that the numbers bj (j=1,2,..., a) which appear in Theorems 1 and 2 are positive integers follows from the fact that L(u) is a polynomial and from
induction on a.References
B.C. Berndt and R.J. Evans, Sums of Gauss, Jacobi, and Jacobsthal, J. Number
Theory 11 (1979), 349‑398.
H. Davenport and H. Hasse, Die Nullstellen der kougruenzzeta‑fwaktionen in gewissen zyklischen Fallen, J. Reine Angew. Math. 172 (1934), 151‑182.
M. Eichler, "hltroduction to the theory of algebraic numbers and functions," Academic Press, New York‑London, 1966.
H. Hasse "The Rlemann hypothesls In algebronc functwa flelds over a fmlte constauts field," Pennsylvania, 1968.
K Ireland and M Rosen "A classlcal rntroductlon to modern number theory,"
Springer, New York‑Heidelberg‑Berlin, 1982.
T. Kodama and T. Washio, On class numbers of hyperelliptic function fields with Hasse‑Witt‑invariant zero. Arch. Math. 49 (1987), 208‑213.
T. Kodama and T. Washio, A family of hyperelliptic function fields with Hasse‑
Witt‑invariant zero, J. Number Theory 36 (1990), 187‑200.
R Lldl and H. Niederreiter, "Finlte flelds " Addlson Wesley Readlng MA 1983