Brauer's theorem on forms

Statement of Brauer's theorem

Let K be a field such that for every integer r > 0 there exists an integer ψ(r) such that for n ≥ ψ(r) every equation

( ∗ ) a 1 x 1 r + ⋯ + a n x n r = 0 , a i ∈ K , i = 1 , … , n

has a non-trivial (i.e. not all x_i are equal to 0) solution in K. Then, given homogeneous polynomials f₁,...,f_k of degrees r₁,...,r_k respectively with coefficients in K, for every set of positive integers r₁,...,r_k and every non-negative integer l, there exists a number ω(r₁,...,r_k,l) such that for n ≥ ω(r₁,...,r_k,l) there exists an l-dimensional affine subspace M of Kⁿ (regarded as a vector space over K) satisfying

f 1 ( x 1 , … , x n ) = ⋯ = f k ( x 1 , … , x n ) = 0 , ∀ ( x 1 , … , x n ) ∈ M .

An application to the field of p-adic numbers

Letting K be the field of p-adic numbers in the theorem, the equation (*) is satisfied, since Q p ∗ / ( Q p ∗ ) b , b a natural number, is finite. Choosing k = 1, one obtains the following corollary:

A homogeneous equation f(x₁,...,x_n) = 0 of degree r in the field of p-adic numbers has a non-trivial solution if n is sufficiently large.

One can show that if n is sufficiently large according to the above corollary, then n is greater than r². Indeed, Emil Artin conjectured that every homogeneous polynomial of degree r over Q_p in more than r² variables represents 0. This is obviously true for r = 1, and it is well known that the conjecture is true for r = 2 (see, for example, J.-P. Serre, A Course in Arithmetic, Chapter IV, Theorem 6). See quasi-algebraic closure for further context.

In 1950 Demyanov verified the conjecture for r = 3 and p ≠ 3, and in 1952 D. J. Lewis independently proved the case r = 3 for all primes p. But in 1966 Guy Terjanian constructed a homogeneous polynomial of degree 4 over Q₂ in 18 variables that has no non-trivial zero. On the other hand, the Ax–Kochen theorem shows that for any fixed degree Artin's conjecture is true for all but finitely many Q_p.