The equation already suggests grouping the four real variables into two Gaussian integers. Put
and
. Then
Thus Lagrange’s theorem says that every positive integer is represented by the binary Hermitian form on
. This reformulation changes the point of view. Rather than seeking four integers directly, we study positive definite Hermitian forms over the Gaussian integers and ask whether an apparently new unimodular form can really differ from the standard norm form.
A general positive definite integral binary Hermitian form is
where ,
, and positivity means
together with
Equivalently, is represented by the Hermitian matrix
.
A Gaussian-integral change of variables by with unit determinant replaces this matrix by
. When one writes the transformation in real and imaginary parts, it becomes an integral unimodular change of four real variables compatible with the complex structure.
Reduction of forms
Every positive definite integral binary Hermitian form of determinant is equivalent over
to the standard form
This is the Hermitian analogue of Gauss reduction. Let
have determinant
, and choose a nonzero vector
on which
is as small as possible. The vector
must be primitive: if
with a nonunit
, then
. Since
is Euclidean, a primitive vector extends to a Gaussian-integral basis. After changing variables, we may therefore suppose that the minimum occurs at
, so that
is the least positive value of
on nonzero Gaussian-integral vectors.
Now use a shear. Replacing by
, with
, changes the mixed coefficient by a multiple of
. Choose
so that the real and imaginary parts of the resulting coefficient
lie between
and
. Then
. The new coefficient
of
is the value of the transformed form at
, hence
by minimality of
. Determinant is unchanged, so
As is a positive integer, this forces
. Then
, so
, and the determinant identity gives
. The reduced form is exactly
. Thus there is only one positive definite unimodular binary Hermitian form over
.
Congruence
We now apply the lemma to an odd prime . First choose
with
Such a pair always exists. The set of quadratic residues modulo , including zero, has
elements, and so does its translate by
. Since the two sets together have
elements inside a set of only
residue classes, they intersect. Let
. Then
.
Define the Gaussian lattice
It is a rank-two -module, with basis
and
: every element has the form
for
. The reason for choosing this lattice is that its vectors automatically have norm divisible by
. Indeed, if
, then
We may therefore divide the norm by and obtain the integral Hermitian form
Expanding gives
Its Hermitian matrix is
and its determinant is Thus
is a positive definite integral binary Hermitian form of determinant
. The reduction lemma says that it is equivalent to
; in particular it represents
. Hence there are
such that
Returning to the corresponding vector , the definition of
yields
Writing
and
, we obtain
Thus every odd prime is a sum of four squares; and . Finally, Euler’s four-square identity shows that products of sums of four squares are again sums of four squares. Every positive integer is therefore a sum of four integer squares.
The elementary descent proof begins with a representable multiple and repeatedly reduces its multiplier. The present proof packages that descent into reduction theory. The congruence
builds a Gaussian lattice on which the usual norm is automatically divisible by
. After dividing by
, one obtains a unimodular integral Hermitian form. The reduction lemma says that there is only one such form, namely
. A vector of norm
for the rescaled form is exactly a vector of norm
in the original congruence lattice.
This is the Hermitian analogue of the class-number-one proof of the two-square theorem. For itself, a prime
yields a Gaussian ideal above
, and principality yields an element of norm
. Here the passage from one Gaussian variable to two turns the binary norm into the four-square form. The congruence
always has a solution, so the Hermitian setting has enough room to represent every prime.