Absbqfs
Absbqfs
Absbqfs
SAMUEL A. HAMBLETON
A short summary of binary quadrati
forms is provided. This
in
ludes
lasses of forms under various equivalen
e relations, the relationship
between the existen
e of a primitive representation of an integer n by a binary
quadrati
form Q and an ideal fa
torization of (n),
onstru
tion of the automorph matrix, Ardnt's
omposition algorithm, Gauss' bilinear transformation,
Bhargava
ubes, and proof of the isomorphism between the ideal
lass group
and the
lass group of forms.
Abstra
t.
1.
Introdu tion
There are several ex
ellent introdu
tory resour
es on binary quadrati
forms
[3, 14, 19, 20. This arti
le is
on
erned with presenting a few important fa
ts about
forms and their relation to ideals within eight pages, and is almost pre
isely the
ba
kground required to study the topi
s
overed in [11. Ideals permit
onsideration
of one of the most important problems of arithmeti
, unique fa
torization. A binary
quadrati
form
ontains more information than an ideal of a quadrati
number eld
K = Q( ), an integral basis. With forms, questions
on
erning the ideal
lass
group may be interpreted geometri
ally, for example, representations of an integer
n by a form Q. The group of units of K
an be thought of as representations of
1, by the prin
ipal form of dis
riminant , see [12, 14. Di
kson's third volume
of `History of the theory of numbers' [6 is entirely
on
erned with forms. The main
on
epts one should know about binary quadrati
forms will be illustrated in this
arti
le. Important fa
ts about ideals will be in
luded in footnotes.
S. HAMBLETON
of fundamental dis
riminant = B 2 4AC . Let F denote the set of all binary
quadrati
forms of this des
ription of dis
riminant . For every Q F, there exists
a unique integer satisfying B = 2 + b. The symbols , and will always
refer to the unique integers
Bb
b+
b
=
,
=
,
=
,
2
2
2
= A2 ,
(mod A1 ).
The proof of this remark is a simple exer
ise. Of
ourse if 0 B1 , B2 < A1 , then
1 = 2 . There are innitely many possible
oe
ients A of a binary quadrati
form of dis
riminant so does not partition F into nitely many equivalen
e
+
lasses. We denote the sets of equivalen
e
lasses under , , and by Cl(),
+
Cl (), and F respe
tively. These are known to be groups under
omposition of
forms and former two are isomorphi
to the ideal
lass groups Cl(K) and Cl+ (K)
respe
tively. Due to the fa
t that VL2 (Z) is a subgroup of SL2 (Z), a subgroup of
GL2 (Z), we have2 hF > h+
K hK .
If Qa Qb then a = b, with Qa , Qb as in Eqn. (3).
The
lass numbers hK = h+
K unless > 0 and the norm of the fundamental unit is equal to 1,
in whi
h
ase h+
=
2h
.
See
[4, 8.
K
K
1
2
(3)
Qa (t, u) =
and
A
=
(A,
+
)
are
in
the
same ideal
t
lass and A
orresponds to Q, the forms atta
hed to A and a are respe
tively
NK/Q (Ax + ( + )y)
= Q(x, y),
A
NK/Q (nx + ((n + Cu)/t + + )y )
B 2 2
Qa =
= nx2 + Bx y +
y .
n
4n
+
t
The forms QA Qa under substitution with u u+1 SL2 (Z). Conversely let
QA
2
(n) = (n, +)(n, + ). The binary quadrati
form nt2 +(2+b)tu+ (2+b)
u
4n
orresponds to the ideal (n, + ) and represents n primitively with (1, 0). Sin
e
this form belongs to the same
lass of Cl() as Q, it follows that Q represents n
primitively.
2.
(4)
1 = 2 ,
S. HAMBLETON
and let Q F be as in Eqn. (1). Expanding the right hand side of Eqn. (4),
2
(x + y)(At2 + ( + )u2 ),
Axt2 + (x + My)u2 + Ayt2 + (x + y + by)u2 ,
=
=
( 2 = b + M),
A(x y)t2 ( 2 + b M)yu2 + ( + ) Ayt2 + (x + y + by)u2 ,
A (x y)t2 Cyu2 + ( + ) Ayt2 + (x + y + by)u2 .
Comparing the left and right hand sides of Eqn. (4) reveals an automorph matrix
for Q = (A, B, C),
Cy
t1
t2
= xy
(5)
u1
u2 .
Ay x+(+b)y
Taking norms,4 on the left and right hand side of Eqn. (4),
NK/Q (1 )
NK/Q (2 )
= A Q(t1 , u1 ),
= A Q (x y)t2 Cyu2 , Ayt2 + (x + ( + b)y)u2 .
Sin
e is a unit, NK/Q (1 ) = NK/Q (2 ). Under the substitution of Eqn. (5), the
binary quadrati
form Q is un
hanged.
3.
Arndt's Composition
We will now present group laws for Cl(), Cl+ (), and F a
ording to the
omposition algorithm of Arndt [1. For j = 1, 2, let Qj = (Aj , 2j + b, Cj ) F.
We will always use + , , and to refer to
+ = 1 + 2 + b,
= 1 2 ,
= 1 2 + M.
(7)
b3
A1 j + A2 k + + ,
A1
A2
2 j +
1 k +
,
e
e
e
4.
Bilinear transformation
Gauss writes in Art. 234 [9, `Now that we have explained the distribution of
forms into
lasses, genera, and orders and the general properties that result from
these distin
tions, we will go on to another very important subje
t, the
omposition
of forms. Thus far no one has
onsidered this point.'
For j = 1, 2, let Qj = (Aj , 2j + b, Cj ) F and Q3 = (A3 , 23 + b, Q0 (3 , 1)/A3 )
be a
omposed form given by Arndt's
omposition algorithm. There is a multipli
ative identity
(8)
where
e
e
e
(2 3 )t1 u2 +
(1 3 )t2 u1 +
( 3 + )u1 u2 ,
A2
A1
A1 A2
A1
A2
+
t1 u 2 +
t2 u 1 +
u1 u2 .
e
e
e
t3
= et1 t2 +
u3
This is known as bilinear transformation and is essentially the way in whi
h Gauss
omposed forms, see Art. 235, [9. For a re
ent perspe
tive, see [5, 14.
Given Q1 , Q2 F, there is no opportunity to obtain 3 by expanding Eqn. (8)
sin
e 3 vanishes. It fa
t the identity holds for all
(9)
3 Z satisfying
A1 A2
| Q0 (3 , 1).
e2
Gauss, Art. 235, [9, referred to the form Q3 of Eqn. (8) as a
omposite of Q1
and Q2 if the g
d of a
ertain six quantities is equal to 1. In our notation these
2 ,1)
1 ,1)
are A1 , A2 , + , , Q0 (
, Q0 (
. This will o
ur if e = g
d(A1 , A2 , + )
A2
A1
sin
e g
d(e, ) = 1 be
ause our forms are primitive. Arndt's
omposition formula
simply nds 3 satisfying Eqn. (9).
5.
=
?
c
?
g
- f
- b
?= [a, b, c, d, e, f, g, h]
- h
?
- d
S. HAMBLETON
N1 =
c g
dh
,
N2 = fb hd ,
N3 = ge hf .
To ea
h Bhargava
ube , the three asso
iated binary quadrati
forms are given
by
Qj (x, y) = det(Mj x + Nj y);
j {1, 2, 3},
whi
h is
Q1
Q2
Q3
be af, bg + de ah cf, dg ch ,
ce ag, cf + de ah bg, df bh ,
bc ad, bg + cf ah de, f g eh .
Remark 5.1. The binary quadrati
forms Q1 , Q2 , Q3 asso
iated with the Bhargava
ube have the same dis
riminant and satisfy [Q1 Q2 Q3 ] = [Q0 ], where is
form
omposition, Q0 is the prin
ipal form, and [Q] is the narrow
lass of the form
Q in Cl+ (). See [2, 13, 14.
Conversely, Lemmermeyer [14 has shown how we may
onstru
t a Bhargava
ube from the forms Q1 = (A1 , B1 , C1 ), Q2 = (A2 , B2 , C2 ) as follows. See [13, 14.
h A A +
i
2
1
= 0,
,
,
, e = g
d A1 , A2 , + , f, g, h ,
e e e
where f, g satisfy the linear Diophantine equation on the left
(10)
A1
A2
g
f = ,
e
e
h=
+ f eC2
.
A1
It may o
ur that h 6 Z, in whi
h
ase we must adjust the
ube to obtain an integer
Bhargava
ube as follows. Let
A A
1
2
e = g
d A1 , A2 , + ,
= g
d
,
,
e e
+ 1
+
(mod );
0 < ,
=1
.
e
e
Then an integer Bhargava
ube is
h A A +
A2 f + C2 + e + f eC2 i
1
2
= 0,
,
,
, e, f + C2 ,
,
,
e e e
A1
A1
where f, g satisfy Eqn. (10).
6.
One of the most important results
on
erned with binary quadrati
forms, due
to Dedekind [7, states that the
lass groups of forms and the
lass groups of ideals
are isomorphi
.
= (A1 A2 , A1 (2 + ), A2 (1 + ), + + ),
(e)(A3 , 3 + ) = (e)(A3 , b3 + ) = a3 ,
A1 (2 + ) =
A2 (1 + ) =
+ +
A3 e2 ,
Q ( , 1)
A1
0 2
k + e(b3 + ) ,
A3 e
A2
e
Q ( , 1)
A
0 1
2
A3 e
+ j + e(b3 + ) ,
A1
e
Q ( , 1)
Q0 (1 , 1)
+
0 2
A3 e
j+
k + e(b3 + )
.
A2
A1
e
e(b3 + ) = A1 (2 + )j + A2 (1 + )k + ( + + ),
The material in this arti
le has formed a small portion of a preliminary mathemati
s
hapter of a thesis written under the supervision of Vi
tor S
haras
hkin
at the University of Queensland. The author is grateful for this supervision and
funding from the university. The books of Franz Lemmermeyer [14, 15 have been
parti
ularly useful and mu
h of our notation is found there. The author would like
to thank Keith Matthews for multiple dis
ussions on binary quadrati
forms.
The equivalen
e relations for Cl(K) and Cl+ (K) are respe
tively a b if and only if there exists
+
K su
h that a = ()b, and a b if and only if there exists K su
h that a = ()b with
either < 0 or > 0 and , > 0. See [15. a, b are fra
tional ideals even though the proof
uses integers.
6
S. HAMBLETON
Referen es
1. F. Arndt, Ausung einer Aufgabe in der Composition der quadratis
hen Formen, J. Reine Angew. Math. 56, (1859) 6471.
2. M. Bhargava, Higher
omposition laws I : A new view on Gauss
omposition,
and quadrati
generalizations, Ann. of Math. 158 (2004) 217250.
3. D. Buell, Binary quadrati
forms Classi
al theory and modern
omputations,
Springer (1989)
4. H. Cohn, Advan
ed number theory, Dover, New York, (1962)
5. H. Cohn, A se
ond
ourse in number theory, John Wiley & Sons, (1962)
6. L. Di
kson, History of the theory of numbers, III, AMS Chelsea Publishing,
(1992)
7. L. Diri
hlet, R. Dedekind, Vorlesungen Uber Zahlentheorie, Supplement X,
(1871), (1879), (1894) available at:
http://www.ar
hive.org/details/vorlesungenberz04dedegoog
8. S. Fin
h, Class number theory, (2005), available at:
http://algo.inria.fr/
solve/
lss.pdf
9. C. Gauss, Disquisitiones Arithmeti
ae, Translated to English by A. Clarke, Yale
University Press (1966)
10. A. Granville, Le
ture notes in analyti
number theory (2007), available at:
http://www.dms.umontreal.
a/~andrew/
11. S. Hambleton, An isomorphism between the narrow ideal
lass group of squared
ideals of a quadrati
number eld and the kernel of a homomorphism between
ohomology groups for Pell
oni
s, available at:
http://arxiv.org/abs/1108.1610
12. F.
Lemmermeyer,
Coni
s A poor man's ellipti
urves,
arXiv:math/0311306v1, (2003), available at:
http://www.fen.bilkent.edu.tr/~franz/publ/
oni
s.pdf
13. F. Lemmermeyer, Binary quadrati
forms and
ounterexamples to Hasse's lo
alglobal prin
iple, (2010) available at:
http://www.rzuser.uni-heidelberg.de/~hb3/publ/pepin.pdf
14. F. Lemmermeyer, Binary quadrati
forms An elementary approa
h to the
arithmeti
of ellipti
and hyperellipti
urves, preprint (2010) available at:
http://www.rzuser.uni-heidelberg.de/~hb3/publ/bf.pdf
15. F. Lemmermeyer, Re
ipro
ity laws From Euler to Eisenstein, Springer-Verlag,
(2000)
16. H. Lenstra Jr., On the
al
ulation of regulators and
lass numbers of quadrati
elds, J. Armitage (ed.), Journes Arithmtiques 1980, London Math. So
.
Le
ture Note Ser. 56, Cambridge University Press, Cambridge, (1982) 123
150.
17. K. Matthews, A website for
omposing binary quadrati
forms,
http://www.numbertheory.org/php/
18. J. Milne, Algebrai
number theory, le
ture notes available at:
http://www.jmilne.org/math/
19. R. Mollin, Quadrati
s, Bo
a Raton, (1996)
20. G. Pall, Composition of binary quadrati
forms, Bull. Amer. Math. So
., 54,
(1948) 11711175.
21. I. Stewart, D. Tall, Algebrai
number theory and Fermat's last theorem, Third
Ed., A. K. Peters, Massa
husetts, (2002)
S hool of Mathemati s and Physi s, University of Queensland, St. Lu ia, Queensland, Australia 4072,
https://sites.google. om/site/samuelhambleton/