arXiv subjects

Ross Maller

Publications and source records attributed to Ross Maller.

11 recordsLinked to original sources

arXiv2023

Limiting Behaviour of Poisson-Dirichlet and Generalised Poisson-Dirichlet Distributions

We derive large-sample and other limiting distributions of the ``frequency of frequencies'' vector, ${\bf M_n}$, together with the number of species, $K_n$, in a Poisson-Dirichlet or generalised Poisson-Dirichlet gene or species sampling model. Models analysed include those constructed from gamma and $\alpha$-stable subordinators by Kingman, the two-parameter extension by Pitman and Yor, and another two-parameter version constructed by omitting large jumps from an $\alpha$-stable subordinator. In the Poisson-Dirichlet case ${\bf M_n}$ and $K_n$ turn out to be asymptotically independent, and notable, especially for statistical applications, is that in other cases the conditional limiting distribution of ${\bf M_n}$, given $K_n$, is normal, after certain centering and norming.

math.ST↗

arXiv2022

Exact and Asymptotic Tests for Sufficient Followup in Censored Survival Data

The existence of immune or cured individuals in a population and whether there is sufficient followup in a sample of censored observations on their lifetimes to be confident of their presence are questions of major importance in medical survival analysis. So far only a few candidates have been put forward as possible test statistics for the existence of sufficient followup in a sample. Here we investigate one such statistic and give a detailed analysis, obtaining an exact finite sample as well as asymptotic distributions for it, and use these to calculate the power of the test as a function of the followup in the sample.

math.ST↗

arXiv2021

Splitting the Sample at the Largest Uncensored Observation

We calculate finite sample and asymptotic distributions for the largest censored and uncensored survival times, and some related statistics, from a sample of survival data generated according to an iid censoring model. These statistics are important for assessing whether there is sufficient followup in the sample to be confident of the presence of immune or cured individuals in the population. A key structural result obtained is that, conditional on the value of the largest uncensored survival time, and knowing the number of censored observations exceeding this time, the sample partitions into two independent subsamples, each subsample having the distribution of an iid sample of censored survival times, of reduced size, from truncated random variables. This result provides valuable insight into the construction of censored survival data, and facilitates the calculation of explicit finite sample formulae. We illustrate for distributions of statistics useful for testing for sufficient followup in a sample, and apply extreme value methods to derive asymptotic distributions for some of those.

math.ST↗

arXiv2018

Small Time Convergence of Subordinators with Regularly or Slowly Varying Canonical Measure

We consider subordinators $X_α=(X_α(t))_{t\ge 0}$ in the domain of attraction at 0 of a stable subordinator $(S_α(t))_{t\ge 0}$ (where $α\in(0,1)$); thus, with the property that $\overlineΠ_α$, the tail function of the canonical measure of $X_α$, is regularly varying of index $-α\in (-1,0)$ as $x\downarrow 0$. We also analyse the boundary case, $α=0$, when $\overlineΠ_α$ is slowly varying at 0. When $α\in(0,1)$, we show that $(t \overlineΠ_α(X_α(t)))^{-1}$ converges in distribution, as $t\downarrow 0$, to the random variable $(S_α(1))^α$. This latter random variable, as a function of $α$, converges in distribution as $α\downarrow 0$ to the inverse of an exponential random variable. We prove these convergences, also generalised to functional versions (convergence in $\mathbb{D}[0,1]$), and to trimmed versions, whereby a fixed number of its largest jumps up to a specified time are subtracted from a process. The $α=0$ case produces convergence to an extremal process constructed from ordered jumps of a Cauchy subordinator. Our results generalise random walk and stable process results of Darling, Cressie, Kasahara, Kotani and Watanabe.

math.PR↗

arXiv2018

Trimmed Lévy Processes and their Extremal Components

We analyse a trimmed stochastic process of the form ${}^{(r)}X_t= X_t - \sum_{i=1}^r Δ_t^{(i)}$, where $(X_t)_{t \geq 0}$ is a driftless subordinator on $\mathbb{R}$ with its jumps on $[0,t]$ ordered as $ Δ_t^{(1)}\ge Δ_t^{(2)} \cdots$. When $r\to\infty$, both ${}^{(r)}X_t \to 0$ and $Δ_t^{(r)} \to 0$ a.s. for each $t>0$, and it is interesting to study the weak limiting behaviour of $\bigl({}^{(r)}X_t, Δ_t^{(r)}\bigr)$ in this case. We term this "large-trimming" behaviour. Concentrating on the case $t=1$, we study joint convergence of $\bigl({}^{(r)}X_1, Δ_1^{(r)}\bigr)$ under linear normalization, assuming extreme value-related conditions on the Lévy measure of $X$ which guarantee that $Δ_1^{(r)}$ has a limit distribution with linear normalization. Allowing ${}^{(r)}X_1$ to have random centering and scaling in a natural way, we show that $\bigl({}^{(r)}X_1, Δ_1^{(r)}\bigr)$ has a bivariate normal limiting distribution, as $r\to\infty$; but replacing the random normalizations with natural deterministic ones produces non-normal limits which we can specify.

math.PR↗

arXiv2017

Ratios of Ordered Points of Point Processes with Regularly Varying Intensity Measures

We study limiting properties of ratios of ordered points of point processes whose intensity measures have regularly varying tails, giving a systematic treatment which points the way to "large-trimming" properties of extremal processes and a variety of applications. Our point process approach facilitates a connection with the negative binomial process of Gregoire (1984) and consequently to certain generalised versions of the Poisson-Dirichlet distribution.

math.PR↗

arXiv2016

Multivariate Subordination using Generalised Gamma Convolutions with Applications to V.G. Processes and Option Pricing

We unify and extend a number of approaches related to constructing multivariate Variance-Gamma (V.G.) models for option pricing. An overarching model is derived by subordinating multivariate Brownian motion to a subordinator from the Thorin (1977) class of generalised Gamma convolution subordinators. A class of models due to Grigelionis (2007), which contains the well-known Madan-Seneta V.G. model, is of this type, but our multivariate generalization is considerably wider, allowing in particular for processes with infinite variation and a variety of dependencies between the underlying processes. Multivariate classes developed by Pérez-Abreu and Stelzer (2012) and Semeraro (2008) and Guillaume (2013) are also submodels. The new models are shown to be invariant under Esscher transforms, and quite explicit expressions for canonical measures (and transition densities in some cases) are obtained, which permit applications such as option pricing using PIDEs or tree based methodologies. We illustrate with best-of and worst-of European and American options on two assets.

q-fin.MF↗

arXiv2016

Processes of rth Largest

For integers $n\geq r$, we treat the $r$th largest of a sample of size $n$ as an $\mathbb{R}^\infty$-valued stochastic process in $r$ which we denote $\mathbf{M}^{(r)}$. We show that the sequence regarded in this way satisfies the Markov property. We go on to study the asymptotic behaviour of $\mathbf{M}^{(r)}$ as $r\to\infty$, and, borrowing from classical extreme value theory, show that left-tail domain of attraction conditions on the underlying distribution of the sample guarantee weak limits for both the range of $\mathbf{M}^{(r)}$ and $\mathbf{M}^{(r)}$ itself, after norming and centering. In continuous time, an analogous process $\mathbf{Y}^{(r)}r$ based on a two-dimensional Poisson process on $\mathbb{R}_+\times \mathbb{R}$ is treated similarly, but we find that the continuous time problems have a distinctive additional feature: there are always infinitely many points below the $r$th highest point up to time $t$ for any $t>0$. This necessitates a different approach to the asymptotics in this case.

math.PR↗

arXiv2016

Passage time and fluctuation calculations for subexponential Lévy processes

We consider the passage time problem for Lévy processes, emphasising heavy tailed cases. Results are obtained under quite mild assumptions, namely, drift to $-\infty$ a.s. of the process, possibly at a linear rate (the finite mean case), but possibly much faster (the infinite mean case), together with subexponential growth on the positive side. Local and functional versions of limit distributions are derived for the passage time itself, as well as for the position of the process just prior to passage, and the overshoot of a high level. A significant connection is made with extreme value theory via regular variation or maximum domain of attraction conditions imposed on the positive tail of the canonical measure, which are shown to be necessary for the kind of convergence behaviour we are interested in.

math.PR↗

arXiv2011

On the Ruin Probability of the Generalised Ornstein-Uhlenbeck Process in the Cramér Case

For a bivariate \Levy process $(ξ_t,η_t)_{t\ge 0}$ and initial value $V_0$ define the Generalised Ornstein-Uhlenbeck (GOU) process \[ V_t:=e^{ξ_t}\Big(V_0+\int_0^t e^{-ξ_{s-}}\ud η_s\Big),\quad t\ge0,\] and the associated stochastic integral process \[Z_t:=\int_0^t e^{-ξ_{s-}}\ud η_s,\quad t\ge0.\] Let $T_z:=\inf\{t>0:V_t<0\mid V_0=z\}$ and $ψ(z):=P(T_z<\infty)$ for $z\ge 0$ be the ruin time and infinite horizon ruin probability of the GOU. Our results extend previous work of Nyrhinen (2001) and others to give asymptotic estimates for $ψ(z)$ and the distribution of $T_z$ as $z\to\infty$, under very general, easily checkable, assumptions, when $ξ$ satisfies a Cramér condition.

math.PR↗

arXiv2007

Curve crossing for random walks reflected at their maximum

Let $R_n=\max_{0\leq j\leq n}S_j-S_n$ be a random walk $S_n$ reflected in its maximum. Except in the trivial case when $P(X\ge0)=1$, $R_n$ will pass over a horizontal boundary of any height in a finite time, with probability 1. We extend this by giving necessary and sufficient conditions for finiteness of passage times of $R_n$ above certain curved (power law) boundaries, as well. The intuition that a degree of heaviness of the negative tail of the distribution of the increments of $S_n$ is necessary for passage of $R_n$ above a high level is correct in most, but not all, cases, as we show. Conditions are also given for the finiteness of the expected passage time of $R_n$ above linear and square root boundaries.

math.PR↗