The existence of extension fields
For example, a polynomial in the reals such as x2 + 1 = 0 has two roots, both in the complex field.
A result in diophantine approximation
Kronecker's theorem may also refer to a result in diophantine approximations applying to several real numbers xi, for 1 ≤ i ≤ n, that generalises Dirichlet's approximation theorem to multiple variables.
The classical Kronecker's approximation theorem is formulated as follows; Given real numbers and , for any small there exist integers and such that
if and only if for any with
the number is also an integer.
Kronecker's approximation theorem had been firstly proved by L. Kronecker in the end of the 19-th century. It has been now revealed to relate to the idea of n-torus and Mahler measure since the later half of the 20-th century. In terms of physical systems, it has the consequence that planets in circular orbits moving uniformly around a star will, over time, assume all alignments, unless there is an exact dependency between their orbital periods.
The relation to n-torus
- T = RN/ZN,
- T′ = T,
which is that the numbers xi together with 1 should be linearly independent over the rational numbers, is also sufficient. Here it is easy to see that if some linear combination of the xi and 1 with non-zero rational number coefficients is zero, then the coefficients may be taken as integers, and a character χ of the group T other than the trivial character takes the value 1 on P. By Pontryagin duality we have T′ contained in the kernel of χ, and therefore not equal to T.
In fact a thorough use of Pontryagin duality here shows that the whole Kronecker theorem describes the closure of <P> as the intersection of the kernels of the χ with
- χ(P) = 1.
This gives an (antitone) Galois connection between monogenic closed subgroups of T (those with a single generator, in the topological sense), and sets of characters with kernel containing a given point. Not all closed subgroups occur as monogenic; for example a subgroup that has a torus of dimension ≥ 1 as connected component of the identity element, and that is not connected, cannot be such a subgroup.
The theorem leaves open the question of how well (uniformly) the multiples mP of P fill up the closure. In the one-dimensional case, the distribution is uniform by the equidistribution theorem.