Regenerative Composition Structures

Alexander Gnedin,Jim Pitman

DOI: https://doi.org/10.48550/arXiv.math/0307307

2004-07-26

Abstract:A new class of random composition structures (the ordered analog of Kingman's partition structures) is defined by a regenerative description of component sizes. Each regenerative composition structure is represented by a process of random sampling of points from an exponential distribution on the positive halfline, and separating the points into clusters by an independent regenerative random set. Examples are composition structures derived from residual allocation models, including one associated with the Ewens sampling formula, and composition structures derived from the zero set of a Brownian motion or Bessel process. We provide characterisation results and formulas relating the distribution of the regenerative composition to the L{é}vy parameters of a subordinator whose range is the corresponding regenerative set. In particular, the only reversible regenerative composition structures are those associated with the interval partition of $[0,1]$ generated by excursions of a standard Bessel bridge of dimension $2 - 2 \alpha$ for some $\alpha \in [0,1]$.

Probability,Combinatorics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to define and study a new class of random composition structures, namely the component sizes defined by regenerative description. Specifically, the author explores how to represent each regenerative composition structure by randomly sampling points from the exponential distribution on the positive half - axis and separating these points into clusters by independent regenerative random sets.

### Main problems and objectives

1. **Define regenerative composition structures**:
- The author introduces a new class of composition structures, called regenerative composition structures. Such structures can be represented by randomly sampling points from the exponential distribution and grouping these points according to independent regenerative random sets.

2. **Characterize the properties of regenerative composition structures**:
- Study the relationship between regenerative composition structures and subordinators (increasing processes with stationary and independent increments). In particular, the author shows that each regenerative composition structure can be described by the Lévy parameter of a subordinator.
- Provide the formula between the probability distribution of the regenerative composition structure and the Lévy parameter of its corresponding subordinator.

3. **Reversibility problem**:
- Determine which regenerative composition structures are reversible. The author proves that the only reversible regenerative composition structures are those associated with the interval partitions generated by the standard Bessel bridge.

4. **Illustrate with examples**:
- Discuss several specific examples of regenerative composition structures, including the composition structures derived from the residual allocation model, the composition structures related to the Ewens sampling formula, and the composition structures derived from the zero sets of the Brownian motion or the Bessel process.

### Mathematical formulas

- **Probability distribution of regenerative composition structures**:
\[
q(n:m)=\frac{\Phi(n:m)}{\Phi(n)}
\]
where \(\Phi(n:m)\) is the binomial moment defined by the Lévy measure \(\nu\) of the subordinator and the drift coefficient \(d\).

- **Laplace exponent of the subordinator**:
\[
\Phi(\rho)=\int_{0}^{\infty}(1 - e^{-\rho z})\nu(dz)+\rho d
\]

- **Recursive relation of regenerative composition structures**:
\[
q(n:m)=\frac{m + 1}{n + 1}q(n + 1:m + 1)+\frac{n + 1 - m}{n + 1}q(n + 1:m)+\frac{1}{n + 1}q(n + 1:1)q(n:m)
\]

Through these formulas and theoretical frameworks, the author provides a systematic method to understand and analyze the properties and applications of regenerative composition structures.

### Summary

The main contribution of this paper lies in introducing and studying in detail the new concept of regenerative composition structures, revealing the profound connection between them and subordinators, and providing specific mathematical tools to describe and analyze such structures. This provides a new perspective and method for understanding the combinatorial phenomena in complex random processes.