Computing Inverses of Stieltjes Transforms of Probability Measures

James Chen,Sheehan Olver

2024-10-22

Abstract:The Stieltjes (or sometimes called the Cauchy) transform is a fundamental object associated with probability measures, corresponding to the generating function of the moments. In certain applications such as free probability it is essential to compute the inverses of the Stieltjes transform, which might be multivalued. This paper establishes conditions bounding the number of inverses based on properties of the measure which can be combined with contour integral-based root finding algorithms to rigorously compute all inverses.

Numerical Analysis

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is the computation of the inverse problem of the Stieltjes transform (or Cauchy transform) of probability measures. Specifically, the authors are concerned with how to effectively compute the inverse of the Stieltjes transform in certain applications, such as in free probability theory. Since the Stieltjes transform may be multi - valued, it is necessary to determine the number of its inverses and propose a method to rigorously compute all of these inverses.

### Core problems of the paper

1. **Inverse problem of the Stieltjes transform**:
- The Stieltjes transform is a fundamental object related to probability measures, corresponding to the moment - generating function.
- In free probability theory, it is necessary to compute the inverse of the Stieltjes transform in order to define the free convolution of two probability measures, which describes the distribution of the sum of two freely independent, self - adjoint non - commutative random variables.
- The authors aim to provide a strict upper bound to determine the number of inverses of the Stieltjes transform and combine a root - finding algorithm based on contour integration to rigorously compute all of the inverses.

2. **Treatment of measures with multi - interval support**:
- For measures with multi - interval support, the Stieltjes transform is usually not single - valued, which increases the complexity of computing the inverse.
- The authors propose a method to recover all inverses of the Stieltjes transform of measures with multi - interval support under a wide range of conditions.

### Key points of the solution

- **Conditions and upper bounds**: The authors establish conditions based on the properties of measures to limit the number of inverses of the Stieltjes transform.
- **Numerical methods**: Combining contour integration and polynomial root - finding algorithms, a specific method for computing the inverses is proposed.
- **Extension to generalized measures**: Through approximation methods, the results are generalized to more general measures, including density functions with jump discontinuities and unbounded behavior at the boundaries.

### Markdown representation of formulas

The formulas involved in the paper are represented in the standard Markdown format, for example:

- Definition of the Stieltjes transform:
\[
G_\mu(z) = \int_{\Gamma} \frac{1}{z - x} d\mu(x) \quad \text{(1.1)}
\]
- Definition of the Hilbert transform:
\[
H_\mu(x) = -\int_{\Gamma} \frac{1}{x - t} d\mu(t) \quad \text{(1.2)}
\]
- Relationship of the R - transform:
\[
G_\mu\left(R_\mu(z) + \frac{1}{z}\right) = z
\]
\[
R_{\mu_a \boxplus \mu_b}(z) = R_{\mu_a}(z) + R_{\mu_b}(z) \quad \text{(1.3)}
\]

Through these methods, the paper provides a theoretical basis and practical computational means for the inverse problem of the Stieltjes transform, especially for applications in free probability theory.