Levy--Khintchine representation of Bernstein functions #
This file proves the existence part of the converse Levy--Khintchine representation: every
Bernstein function is the sum of a nonnegative killing term, a nonnegative linear drift, and the
jump exponent of a Bernstein Levy measure. Uniqueness of the three parameters is proved in
TauCeti.Analysis.CompletelyMonotone.Bernstein.LevyKhintchine.Uniqueness.
The proof represents the completely monotone derivative by a measure sigma. The atom of
sigma at zero is the drift coefficient, while weighting sigma by x⁻¹ away from zero gives
the Levy measure. Continuity of the Bernstein function at zero is exactly what makes the
truncated coordinate integrable against this weighted measure.
Main declaration #
TauCeti.IsBernsteinFunction.exists_eqOn_bernsteinLevyKhintchineExponent: existence of a Levy--Khintchine triplet for a Bernstein function.TauCeti.isBernsteinFunction_iff_exists_bernsteinLevyKhintchineExponent: the resulting characterization of Bernstein functions.
References #
- R. Schilling, R. Song, Z. Vondracek, Bernstein Functions: Theory and Applications (de Gruyter, 2nd ed. 2012), Theorem 3.2.
- Roadmap:
TauCetiRoadmap/OneParameterSemigroups/README.md, Part B (Levy--Khintchine representation of Bernstein functions).
Every Bernstein function admits a Levy--Khintchine representation on the nonnegative half-line. The witnesses are a killing coefficient, a drift coefficient, and a Levy measure; this theorem asserts existence only.
A function is Bernstein exactly when it has a Levy--Khintchine representation on the nonnegative half-line.