The Arithmetic of Quaternion Algebra
The Arithmetic of Quaternion Algebra
The Arithmetic of Quaternion Algebra
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
28 CHAPTER 2. QUATERNION ALGEBRA OVER A LOCAL FIELD<br />
2.2 Study <strong>of</strong> M(2, K)<br />
Let V be a vector space <strong>of</strong> dimension 2 over K. Suppose a basis (e1, e2) <strong>of</strong> V/K<br />
to be fixed such that V = e1K +e2K. This basis allowed�us to identify � M(2, K)<br />
a b<br />
with the ring <strong>of</strong> endomorphisms End(V ) <strong>of</strong> V . If h = ∈ M(2, K), it<br />
c d<br />
associates an endomorphism :v ↦→ v.h, which is defined by the product <strong>of</strong> the<br />
row matrix (x, y) by h, if v = e1x + e2y. Recall that a complete lattice in V is a<br />
R-module containing a basis <strong>of</strong> V/K. If L, M are two complete lattices in V , we<br />
denote the ring <strong>of</strong> R-endomorphisms from L to M by End(L, M), or End(L) if<br />
L = M.<br />
Lemma 2.2.1. (1) <strong>The</strong> maximal orders <strong>of</strong> End(V ) are the rings End(L), where<br />
L runs through the complete lattices <strong>of</strong> V .<br />
(2) <strong>The</strong> normal ideals <strong>of</strong> End(v) are the ideals End(L, M), where L, M runs<br />
through the complete lattices <strong>of</strong> V .<br />
Pro<strong>of</strong>. (1) Let O be an order <strong>of</strong> End(V ) and M a complete lattice <strong>of</strong> V . Set L =<br />
{m ∈ M|m ∈ O}. It is a R-module contained in V . <strong>The</strong>re exists a ∈ R such that<br />
aEnd(M) ⊂ M}. It follows aM ⊂ L ⊂ M, hence L is a complete lattice. It is<br />
clear that O ⊂ End(L). (2) Let I be an ideal to the left <strong>of</strong> End(L). We identify<br />
I to a R-module f(I) <strong>of</strong> V 2 by the mapping h ↦→ f(h) = (e1h, e2h). Let xi,jbe<br />
the endomorphism permuting e1 and e2 if i �= j, but fixing ei if i = j, and taking<br />
the other element <strong>of</strong> the basis to zero. Choosing all the possibility for (i, j), and<br />
computing f(xi,j, we see that f(I) contains (e1.h, 0), (0, e2.h) (e2h, e1.h).<br />
<strong>The</strong>refore f(I) = M + M by putting M = L.I. We then see easily that M is a<br />
complete lattice. It follows finally I = End(L, M).<br />
We recall here some classical results about the elementary theory <strong>of</strong> divisors.<br />
Lemma 2.2.2. Let L ⊂ M be two complete lattices <strong>of</strong> V .<br />
(1) <strong>The</strong>re exists a R-basis (f1, f2) <strong>of</strong> M and a R-basis (f1π a , f2π b ) <strong>of</strong> L whee<br />
a, b are integers uniquely determined.<br />
(2) If (f1, f2) is a R-basis <strong>of</strong> L, there exists a unique basis <strong>of</strong> M/R <strong>of</strong> the form<br />
(f1π n , f1r + f2π m ), where n, m are integers, and r belongs to a given set Um <strong>of</strong><br />
the representation <strong>of</strong> R/(π m R) in R.<br />
Pro<strong>of</strong>. (1) is classical. We� prove �(2).<br />
<strong>The</strong> basisf1a + f2b, f1c + f2d) <strong>of</strong> M are<br />
a b<br />
such that the matrix A = satisfies L.A = M. We can replace A by XA<br />
c d<br />
if X ∈ M(2, R) × � . We verify without difficulty that it can be modified again to<br />
n π r<br />
A =<br />
0 πm �<br />
where n, m are integers and r ∈ Um<br />
We are going to express these results in terms <strong>of</strong> matrix.<br />
<strong>The</strong>orem 2.2.3. (1) <strong>The</strong> maximal orders <strong>of</strong> M(2, K) are conjugate to M(2, R).<br />
(2) <strong>The</strong> two-sided ideal <strong>of</strong> M(2, R) forms a cyclic group generated by the prime<br />
ideal P = M(2, R)π.<br />
(3)<strong>The</strong> integral ideals to the left <strong>of</strong> M(2, R) are the distinct ideals<br />
�<br />
n π r<br />
M(2, R)<br />
0 πm �