of
f is Δ = 4
a3 − 27
b2. Denoting the discriminant of
K by
D, the
index i(θ) of θ is then defined by Δ =
i(θ)2
D. In the case of a non-cyclic cubic field
K this index formula can be combined with the conductor formula
D =
f2
d to obtain a decomposition of the polynomial discriminant Δ =
i(θ)2
f2
d into the square of the product
i(θ)
f and the discriminant
d of the quadratic field
k associated with the cubic field
K, where
d is
squarefree up to a possible factor 22 or 23.
Georgy Voronoy gave a method for separating
i(θ) and
f in the square part of Δ. The study of the number of cubic fields whose discriminant is less than a given bound is a current area of research. Let
N+(
X) (respectively
N−(
X)) denote the number of totally real (respectively complex) cubic fields whose discriminant is bounded by
X in absolute value. In the early 1970s,
Harold Davenport and
Hans Heilbronn determined the first term of the asymptotic behaviour of
N±(
X) (i.e. as
X goes to infinity). By means of an analysis of the
residue of the
Shintani zeta function, combined with a study of the tables of cubic fields compiled by Karim Belabas and some
heuristics, David P. Roberts
conjectured a more precise asymptotic formula: :N^\pm(X)\sim\frac{A_\pm}{12\zeta(3)}X+\frac{4\zeta(\frac{1}{3})B_\pm}{5\Gamma(\frac{2}{3})^3\zeta(\frac{5}{3})}X^{\frac{5}{6}} where
A± = 1 or 3,
B± = 1 or \sqrt{3}, according to the totally real or complex case, ζ(
s) is the
Riemann zeta function, and Γ(
s) is the
Gamma function. Proofs of this formula have been published by using methods based on Bhargava's earlier work, as well as by based on the Shintani zeta function. ==Unit group==