From Gauss to Eisenstein
by t0rajir0u, Apr 13, 2007, 4:28 AM
On this blog I've discussed numbers of the form
, which are intimately related to a mathematical construct known as the Gaussian integers,
. Here we will explore another binary quadratic form.
Problem 1: (Engel, Chapter 6, #117) The product of two positive integers of the form
has the same form.
Solution: We should recognize this expression from the sum-of-cubes expression. Consequently, let
, a primitive third root of unity; then

This is analogous to the complex factorization
. You may remember that I discussed the Gaussian integers in a previous post. Well, these are the Eisenstein integers, which also have unique prime factorization.
Anyway,
$\begin{eqnarray*}(a^{2}+ab+b^{2})(c^{2}+cd+d^{2}) &=& (a-b \omega)(a-b \omega^{2})(c-d \omega)(c-d \omega^{2}) \\ &=& (ac+bd \omega^{2}-(bc+ad) \omega)(ac+bd \omega-(bc+ad) \omega^{2}) \\ &=& ( (ac-bd)-(bc+ad+bd) \omega)( (ac-bd)-(bc+ad+bd) \omega^{2}) \\ &=& (ac-bd)^{2}+(ac-bd)(bc+ad+bd)+(bc+ad+bd)^{2}\\ \end{eqnarray*}$
(I have used the identity
.) Knowing the analogue for the Gaussian integers makes working through the algebra that much easier. QED.
Remark: Every Eisenstein integer can be expressed in the form
owing to the identity I used. The multiplicative norm (as we have shown here) is
; indeed,
is the complex conjugate of
. We can prove the above result using an argument based on the fact that conjugation is multiplicative here too.
Problem 2: (Engel, Chapter 6, #127) Find infinitely many integer solutions to

Solution: Engel gives a simple solution, so I will give a complicated one. We know from Problem 1 that the LHS is equal to

Clearly one easy subset of solutions occurs when
. Indeed, this gives the identity

Interestingly enough, there are two choices for
. The second choice is a somewhat uncommon factorization; it can be derived by difference of squares.
Problem 3: (Engel, Chapter 6, #136) Show that if
is prime, where
, then
is a power of
.
Solution: The factorization derived in Problem 2 suggests the possibility that

Is it true? Well, a polynomial divides another iff the roots of the divisor are all roots of the dividend; those roots are the roots of unity
. Plugging in, we desire that

In other words, if
has any prime factors
which are not
then

Hence
has no such factors. QED.
Practice Problem: (Engel, Chapter 6, #118) Given
, show that if
has one rational solution
then it has infinitely many.


Problem 1: (Engel, Chapter 6, #117) The product of two positive integers of the form

Solution: We should recognize this expression from the sum-of-cubes expression. Consequently, let


This is analogous to the complex factorization

Anyway,
$\begin{eqnarray*}(a^{2}+ab+b^{2})(c^{2}+cd+d^{2}) &=& (a-b \omega)(a-b \omega^{2})(c-d \omega)(c-d \omega^{2}) \\ &=& (ac+bd \omega^{2}-(bc+ad) \omega)(ac+bd \omega-(bc+ad) \omega^{2}) \\ &=& ( (ac-bd)-(bc+ad+bd) \omega)( (ac-bd)-(bc+ad+bd) \omega^{2}) \\ &=& (ac-bd)^{2}+(ac-bd)(bc+ad+bd)+(bc+ad+bd)^{2}\\ \end{eqnarray*}$
(I have used the identity

Remark: Every Eisenstein integer can be expressed in the form




Problem 2: (Engel, Chapter 6, #127) Find infinitely many integer solutions to

Solution: Engel gives a simple solution, so I will give a complicated one. We know from Problem 1 that the LHS is equal to

Clearly one easy subset of solutions occurs when


Interestingly enough, there are two choices for

Problem 3: (Engel, Chapter 6, #136) Show that if




Solution: The factorization derived in Problem 2 suggests the possibility that

Is it true? Well, a polynomial divides another iff the roots of the divisor are all roots of the dividend; those roots are the roots of unity


In other words, if




Hence

Practice Problem: (Engel, Chapter 6, #118) Given


