Searcharxiv⌕ Search

arXiv subjects

Ross A. Maller

Publications and source records attributed to Ross A. Maller.

At least 19 recordsLinked to original sources
arXiv2024

A Gibbs Sampling Scheme for a Generalised Poisson-Kingman Class

A Bayesian nonparametric method of James, Lijoi \& Prunster (2009) used to predict future values of observations from normalized random measures with independent increments is modified to a class of models based on negative binomial processes for which the increments are not independent, but are independent conditional on an underlying gamma variable. Like in James et al., the new algorithm is formulated in terms of two variables, one a function of the past observations, and the other an updating by means of a new observation. We outline an application of the procedure to population genetics, for the construction of realisations of genealogical trees and coalescents from samples of alleles.

stat.ME↗
arXiv2020

Extremes of Censored and Uncensored Lifetimes in Survival Data

The i.i.d. censoring model for survival analysis assumes two independent sequences of i.i.d. positive random variables, $(T_i^*)_{1\le i\le n}$ and $(U_i)_{1\le i\le n}$. The data consists of observations on the random sequence $\big(T_i=\min(T_i^*,U_i)$ together with accompanying censor indicators. Values of $T_i$ with $T_i^*\le U_i$ are said to be uncensored, those with $T_i^*> U_i$ are censored. We assume that the distributions of the $T_i^*$ and $U_i$ are in the domain of attraction of the Gumbel distribution and obtain the asymptotic distributions, as sample size $n\to\infty$, of the maximum values of the censored and uncensored lifetimes in the data, and of statistics related to them. These enable us to examine questions concerning the possible existence of cured individuals in the population.

math.ST↗
arXiv2019

Models for Genetic Diversity Generated by Negative Binomial Point Processes

We develop a model based on a generalised Poisson-Dirichlet distribution for the analysis of genetic diversity, and illustrate its use on microsatellite data for the genus Dasyurus (the quoll, a marsupial carnivore listed as near-threatened in Australia). Our class of distributions, termed $PD_α^{(r)}$, is constructed from a negative binomial point process, generalizing the usual one-parameter $PD_α$ model, which is constructed from a Poisson point process. Both models have at their heart a Stable$(α)$ process, but in $PD_α^{(r)}$, an extra parameter $r>0$ adds flexibility, analogous to the way the negative binomial distribution allows for "overdispersion" in the analysis of count data. A key result obtained is a generalised version of Ewens' sampling formula for $PD_α^{(r)}$. We outline the theoretical basis for the model, and, for the quolls data, estimate the parameters $α$ and r by least squares, showing how the extra parameter r aids in the interpretability of the data by comparison with the standard $PD_α$ model. The methods potentially have implications for the management and conservation of threatened populations.

stat.AP↗
arXiv2018

Convergence to stable limits for ratios of trimmed Levy processes and their jumps

We derive characteristic function identities for conditional distributions of an r-trimmed Levy process given its r largest jumps up to a designated time t. Assuming the underlying Levy process is in the domain of attraction of a stable process as t goes to 0, these identities are applied to show joint convergence of the trimmed process divided by its large jumps to corresponding quantities constructed from a stable limiting process. This generalises related results in the 1-dimensional subordinator case developed in Kevei & Mason (2014) and produces new discrete distributions on the infinite simplex in the limit.

math.PR↗
arXiv2018

Negative Binomial Construction of Random Discrete Distributions on the Infinite Simplex

The Poisson-Kingman distributions, $\mathrm{PK}(ρ)$, on the infinite simplex, can be constructed from a Poisson point process having intensity density $ρ$ or by taking the ranked jumps up till a specified time of a subordinator with Lévy density $ρ$, as proportions of the subordinator. As a natural extension, we replace the Poisson point process with a negative binomial point process having parameter $r>0$ and Lévy density $ρ$, thereby defining a new class $\mathrm{PK}^{(r)}(ρ)$ of distributions on the infinite simplex. The new class contains the two-parameter generalisation $\mathrm{PD}(α, θ)$ of Pitman and Yor (1997) when $θ>0$. It also contains a class of distributions derived from the trimmed stable subordinator. We derive properties of the new distributions, with particular reference to the two most well-known $\mathrm{PK}$ distributions: the Poisson-Dirichlet distribution $\mathrm{PK}(ρ_θ)$ generated by a Gamma process with Lévy density $ρ_θ(x) = θe^{-x}/x$, $x>0$, $θ> 0$, and the random discrete distribution, $\mathrm{PD}(α,0)$, derived from an $α$-stable subordinator.

math.PR↗
arXiv2017

Functional Laws for Trimmed Levy Processes

Two different ways of trimming the sample path of a stochastic process in D[0, 1]: global ("trim as you go") trimming and record time ("lookback") trimming are analysed to find conditions for the corresponding operators to be continuous with respect to the (strong) J1-topology. A key condition is that there should be no ties among the largest ordered jumps of the limit process. As an application of the theory, via the continuous mapping theorem we prove limit theorems for trimmed Levy processes, using the functional convergence of the underlying process to a stable process. The results are applied to a reinsurance ruin time problem.

math.PR↗
arXiv2017

Generalised Poisson-Dirichlet Distributions and the Negative Binomial Point Process

When $S=(S_t)_{t\ge 0}$ is an $α$-stable subordinator, the sequence of ordered jumps of $S$, up till time $1$, omitting the $r$ largest of them, and taken as proportions of their sum $^{(r)}S_t$, defines a 2-parameter distribution on the infinite dimensional simplex, $\nabla_{\infty}$, which we call the $\mathrm{PD}_α^{(r)}$ distribution. When $r=0$ it reduces to the $\mathrm{PD}_α$ distribution introduced by Kingman in 1975. We observe a serendipitous connection between $\mathrm{PD}_α^{(r)}$ and the negative binomial point process of Gregoire (1984), which we exploit to analyse in detail a size-biased version of $\mathrm{PD}_α^{(r)}$. As a consequence we derive a stick-breaking representation for the process and a useful form for its distribution. This program produces a large new class of distributions available for a variety of modelling purposes.

math.PR↗
arXiv2016

Distributional representations and dominance of a Lévy process over its maximal jump processes

Distributional identities for a Lévy process $X_t$, its quadratic variation process $V_t$ and its maximal jump processes, are derived, and used to make "small time" (as $t\downarrow0$) asymptotic comparisons between them. The representations are constructed using properties of the underlying Poisson point process of the jumps of $X$. Apart from providing insight into the connections between $X$, $V$, and their maximal jump processes, they enable investigation of a great variety of limiting behaviours. As an application, we study "self-normalised" versions of $X_t$, that is, $X_t$ after division by $\sup_{0<s\le t}ΔX_s$, or by $\sup_{0<s\le t}| ΔX_s|$. Thus, we obtain necessary and sufficient conditions for $X_t/\sup_{0<s\le t}ΔX_s$ and $X_t/\sup_{0<s\le t}| ΔX_s|$ to converge in probability to 1, or to $\infty$, as $t\downarrow0$, so that $X$ is either comparable to, or dominates, its largest jump. The former situation tends to occur when the singularity at 0 of the Lévy measure of $X$ is fairly mild (its tail is slowly varying at 0), while the latter situation is related to the relative stability or attraction to normality of $X$ at 0 (a steeper singularity at 0). An important component in the analyses is the way the largest positive and negative jumps interact with each other. Analogous "large time" (as $t\to \infty$) versions of the results can also be obtained.

math.PR↗
arXiv2015

Thin and Thick Strip Passage Times for Lévy Flights and Lévy Processes

We review some of the theory relevant to passage times of one-dimensional Lévy processes out of bounded regions, highlighting results that are useful in physical phenomena modelled by heavy-tailed Lévy flights. The process is hypothesised to describe the motion of a particle on the line, starting at $0$, and exiting either a fixed interval $[-r, r]$, $r > 0$, or a time-dependent, expanding, set of intervals of the form $[-r t^κ, r t^κ]$, $r > 0$, $κ> 0$. Asymptotic behaviour of the exit time may be as $r \downarrow 0$ or as $r \to \infty$, but particular emphasis is placed herein on "small time" approximations, corresponding to exits from or transmissions through thin strips. Applications occur for example in the transmission of photons through moderately doped thin or thick wafers by means of "photon recycling", and in atmospheric radiation modelling.

math.PR↗
arXiv2013

Finite Time Ruin Probabilities for Tempered Stable Insurance Risk Processes

We study the probability of ruin before time $t$ for the family of tempered stable Lévy insurance risk processes, which includes the spectrally positive inverse Gaussian processes. Numerical approximations of the ruin time distribution are derived via the Laplace transform of the asymptotic ruin time distribution, for which we have an explicit expression. These are benchmarked against simulations based on importance sampling using stable processes. Theoretical consequences of the asymptotic formulae are found to indicate some potential drawbacks to the use of the inverse Gaussian process as a risk reserve process. We offer as alternatives natural generalizations which fall within the tempered stable family of processes.

math.PR↗
arXiv2012

Path decomposition of ruinous behavior for a general Lévy insurance risk process

We analyze the general Lévy insurance risk process for Lévy measures in the convolution equivalence class $\mathcal{S}^{(α)}$, $α>0$, via a new kind of path decomposition. This yields a very general functional limit theorem as the initial reserve level $u\to \infty$, and a host of new results for functionals of interest in insurance risk. Particular emphasis is placed on the time to ruin, which is shown to have a proper limiting distribution, as $u\to \infty$, conditional on ruin occurring under our assumptions. Existing asymptotic results under the $\mathcal{S}^{(α)}$ assumption are synthesized and extended, and proofs are much simplified, by comparison with previous methods specific to the convolution equivalence analyses. Additionally, limiting expressions for penalty functions of the type introduced into actuarial mathematics by Gerber and Shiu are derived as straightforward applications of our main results.

math.PR↗
arXiv2011

Small and Large Time Stability of the Time taken for a Lévy Process to Cross Curved Boundaries

This paper is concerned with the small time behaviour of a Lévy process $X$. In particular, we investigate the {\it stabilities} of the times, $\Tstarb(r)$ and $\Tbarb(r)$, at which $X$, started with $X_0=0$, first leaves the space-time regions $\{(t,y)\in\R^2: y\le rt^b, t\ge 0\}$ (one-sided exit), or $\{(t,y)\in\R^2: |y|\le rt^b, t\ge 0\}$ (two-sided exit), $0\le b<1$, as $r\dto 0$. Thus essentially we determine whether or not these passage times behave like deterministic functions in the sense of different modes of convergence; specifically convergence in probability, almost surely and in $L^p$. In many instances these are seen to be equivalent to relative stability of the process $X$ itself. The analogous large time problem is also discussed.

math.PR↗
arXiv2011

The time at which a Lévy process creeps

We show that if a Lévy process creeps then, as a function of $u$, the renewal function $V(t,u)$ of the bivariate ascending ladder process $(L^{-1},H)$ is absolutely continuous on $[0,\infty)$ and left differentiable on $(0,\infty)$, and the left derivative at $u$ is proportional to the (improper) distribution function of the time at which the process creeps over level $u$, where the constant of proportionality is $\rmd_H^{-1}$, the reciprocal of the (positive) drift of $H$. This yields the (missing) term due to creeping in the recent quintuple law of Doney and Kyprianou (2006). As an application, we derive a Laplace transform identity which generalises the second factorization identity. We also relate Doney and Kyprianou's extension of Vigon's équation amicale inversée to creeping. Some results concerning the ladder process of $X$, including the second factorization identity, continue to hold for a general bivariate subordinator, and are given in this generality.

math.PR↗
arXiv2011

Stability of the Exit Time for Lévy Processes

This paper is concerned with the behaviour of a Lévy process when it crosses over a positive level, $u$, starting from 0, both as $u$ becomes large and as $u$ becomes small. Our main focus is on the time, $τ_u$, it takes the process to transit above the level, and in particular, on the {\it stability} of this passage time; thus, essentially, whether or not $τ_u$ behaves linearly as $u\dto 0$ or $u\to\infty$. We also consider conditional stability of $τ_u$ when the process drifts to $-\infty$, a.s. This provides information relevant to quantities associated with the ruin of an insurance risk process, which we analyse under a Cramér condition.

math.PR↗
arXiv2008

GARCH modelling in continuous time for irregularly spaced time series data

The discrete-time GARCH methodology which has had such a profound influence on the modelling of heteroscedasticity in time series is intuitively well motivated in capturing many `stylized facts' concerning financial series, and is now almost routinely used in a wide range of situations, often including some where the data are not observed at equally spaced intervals of time. However, such data is more appropriately analyzed with a continuous-time model which preserves the essential features of the successful GARCH paradigm. One possible such extension is the diffusion limit of Nelson, but this is problematic in that the discrete-time GARCH model and its continuous-time diffusion limit are not statistically equivalent. As an alternative, Klüppelberg et al. recently introduced a continuous-time version of the GARCH (the `COGARCH' process) which is constructed directly from a background driving Lévy process. The present paper shows how to fit this model to irregularly spaced time series data using discrete-time GARCH methodology, by approximating the COGARCH with an embedded sequence of discrete-time GARCH series which converges to the continuous-time model in a strong sense (in probability, in the Skorokhod metric), as the discrete approximating grid grows finer. This property is also especially useful in certain other applications, such as options pricing. The way is then open to using, for the COGARCH, similar statistical techniques to those already worked out for GARCH models and to illustrate this, an empirical investigation using stock index data is carried out.

q-fin.ST↗
arXiv2008

Passage of Lévy Processes across Power Law Boundaries at Small Times

We wish to characterise when a Lévy process $X_t$ crosses boundaries like $t^κ$, $κ>0$, in a one or two-sided sense, for small times $t$; thus, we enquire when $\limsup_{t\downarrow 0}|X_t|/t^κ$, $\limsup_{t\downarrow 0}X_t/t^κ$ and/or $\liminf_{t\downarrow 0}X_t/t^κ$ are almost surely (a.s.) finite or infinite. Necessary and sufficient conditions are given for these possibilities for all values of $κ>0$. Often (for many values of $κ$), when the limsups are finite a.s., they are in fact zero, as we show, but the limsups may in some circumstances take finite, nonzero, values, a.s. In general, the process crosses one or two-sided boundaries in quite different ways, but surprisingly this is not so for the case $κ=1/2$. An integral test is given to distinguish the possibilities in that case. Some results relating to other norming sequences for $X$, and when $X$ is centered at a nonstochastic function, are also given.

math.PR↗
arXiv2006

On Continuity Properties of the Law of Integrals of Lévy Processes

Let $(ξ,η)$ be a bivariate Lévy process such that the integral $\int\_0^\infty e^{-ξ\_{t-}} dη\_t$ converges almost surely. We characterise, in terms of their \LL measures, those Lévy processes for which (the distribution of) this integral has atoms. We then turn attention to almost surely convergent integrals of the form $I:=\int\_0^\infty g(ξ\_t) dt$, where $g$ is a deterministic function. We give sufficient conditions ensuring that $I$ has no atoms, and under further conditions derive that $I$ has a Lebesgue density. The results are also extended to certain integrals of the form $\int\_0^\infty g(ξ\_t) dY\_t$, where $Y$ is an almost surely strictly increasing stochastic process, independent of $ξ$.

math.PR↗