Documentation

TauCeti.Analysis.CompletelyMonotone.Stieltjes.CompleteBernstein

Complete Bernstein functions and Stieltjes functions #

A complete Bernstein function has a representation

g(t) = a + b t + ∫ x, t / (t + x) ∂μ,

where a, b ≥ 0, the positive measure μ has no atom at zero, and ∫ (1 + x)⁻¹ ∂μ < ∞. The normalization μ {0} = 0 stops an atom at zero from making the integral term jump — such an atom contributes 1 at every positive parameter and 0 at zero — so the integral term vanishes at zero and the representation is the intended continuous extension. This file packages that standard representation and proves the first fundamental correspondence: f is Stieltjes exactly when t ↦ t f(t) on (0, ∞) has a complete Bernstein extension to [0, ∞).

The predicate defined here is the integral representation, that is, the representing clause of the cited theorem, and it carries exactly the data μ, a, b of TauCeti.RepresentsStieltjes. The correspondence below is therefore the change of normalization f ↦ t f(t) performed on that shared data. The equivalence of the representation with the analytic characterizations of a complete Bernstein function — extension to a Pick function on ℂ ∖ (-∞, 0], or operator monotonicity — is not formalized here.

The extension is necessary when the Stieltjes representation has a nonzero a / t term: its value at zero must be a, whereas the literal product 0 * f(0) is zero. Values of a Stieltjes function outside (0, ∞) remain irrelevant, and values of a complete Bernstein function outside [0, ∞) remain irrelevant.

Main declarations #

References #

A measure μ and coefficients a, b ≥ 0 represent a complete Bernstein function when μ satisfies the standard Stieltjes integrability condition and the function agrees on [0, ∞) with a + b t + ∫ x, t / (t + x) ∂μ. Requiring μ {0} = 0 stops an atom at zero from making the integral term jump between zero and the positive parameters.

Equations
Instances For

    Characterization of complete-Bernstein representing data without unfolding the predicate.

    A complete-Bernstein representing measure has no atom at zero.

    A complete-Bernstein representing measure satisfies the Stieltjes weight condition.

    Evaluation of a complete-Bernstein representation at a nonnegative parameter.

    A represented complete Bernstein function takes the value a at zero: the boundary value records the coefficient of the a / t term of the associated Stieltjes function.

    A complete-Bernstein representation depends only on the represented function's values on [0, ∞).

    Dividing a complete Bernstein function by its parameter on (0, ∞) gives the Stieltjes function with the same representing data.

    theorem TauCeti.RepresentsCompleteBernstein.add {μ ν : MeasureTheory.Measure NNReal} {a b c d : NNReal} {f g : ℝ → ℝ} (hf : RepresentsCompleteBernstein μ a b f) (hg : RepresentsCompleteBernstein ν c d g) :
    RepresentsCompleteBernstein (μ + ν) (a + c) (b + d) (f + g)

    The sum of two complete-Bernstein representations is represented by the sum of their coefficients and measures.

    A nonnegative scalar multiple of a complete-Bernstein representation is represented by scaling its coefficients and measure.

    A function with a complete-Bernstein representation is a Bernstein function.

    A real function is a complete Bernstein function if it has the standard Stieltjes-type representation on [0, ∞).

    Equations
    Instances For

      Characterization of a complete Bernstein function by its representing data.

      Every complete Bernstein function is a Bernstein function.

      The complete Bernstein property depends only on values on [0, ∞).

      Complete Bernstein functions are closed under addition.

      Complete Bernstein functions are closed under multiplication by a nonnegative scalar.

      Dividing a complete Bernstein function by its parameter gives a Stieltjes function.

      The Stieltjes--Bernstein transform of normalized, integrable measure data has a complete Bernstein representation with the given measure and coefficients.

      The Stieltjes--Bernstein transform of normalized, integrable measure data is a complete Bernstein function.

      Stieltjes--complete-Bernstein correspondence. A function f is Stieltjes exactly when its product t ↦ t * f t on (0, ∞) extends to a complete Bernstein function on [0, ∞). The extension's value at zero records the coefficient of the possible t⁻¹ singularity. Both sides are the same representing data μ, a, b, so the content is the change of normalization f ↦ t f(t) and the continuous extension of the product across zero.

      theorem TauCeti.isCompleteBernsteinFunction_affine {c d : ℝ} (hc : 0 ≤ c) (hd : 0 ≤ d) :
      IsCompleteBernsteinFunction fun (t : ℝ) => c + d * t

      Nonnegative affine functions are complete Bernstein functions.

      Nonnegative constant functions are complete Bernstein functions.

      The identity function is a complete Bernstein function.

      theorem TauCeti.isCompleteBernsteinFunction_div_add {x : ℝ} (hx : 0 < x) :
      IsCompleteBernsteinFunction fun (t : ℝ) => t / (t + x)

      For x > 0, the basic function t ↦ t / (t + x) is complete Bernstein, represented by the unit point mass at x.