We consider d-dimensional stochastic processes which take values in (R+)d These processes generalize Galton-Watson branching processes, but the main assumption of branching processes, independence between particles, is dropped. Instead, we assume for some

Here τ:(R+)d→R +, |x| = σ1d |x(i)|, A {x ∈(R+)d: |x| 1} and T: A→A. Under various assumptions on the maps τ and T it is shown that with probability one there exists a ρ > 1, a fixed point p ∈ A of T and a random variable w such that limn→∞Zn|ρnwp. This result is a generalization of the main limit theorem for supercritical branching processes; note, however, that in the present situation both ρ and ρ are random as well. The results are applied to a population genetical model for zygotic selection without mutation at one locus.

The asymptotic behaviour of those k-type (k ≥ 1) Galton-Watson processes (both with and without immigration) with ρ close to unity is considered. The first principal result (Theorem 3) relates to the vector denoting the numbers of particles of each type at the nth generation in processes without immigration. It is shown that, when normed in a certain way and conditioned on non-extinction, this vector has approximately, for large n, a negative exponential distribution which is degenerate on a line. Theorem 4 is an analogous result for processes with immigration. In this case, no conditioning is required, and the limiting distribution is again degenerate on a line, although now it relates to the gamma rather than the negative exponential.

Consider maxima Mn of a sequence of random variables defined on a finite Markov chain. Necessary and sufficient conditions for the existence of normalizing constants Bn such that are given. The problem can be reduced to studying maxima of i.i.d. random variables drawn from a finite product of distributions πi=1mHi(x). The effect of each factor Hi(x) on the behavior of maxima from πi=1mHi is analyzed. Under a mild regularity condition, Bn can be chosen to be the maximum of the m quantiles of order (1 - n-1) of the H's.

Estimates are given for the rates of convergence in functional central limit theorems for quantities of interest in the GI/G/1 queue and a general multiple channel system. The traffic intensity is fixed ≧ 1. The method employed involves expressing the underlying stochastic processes in terms of Brownian motion using the Skorokhod representation theorem.

For the single server queueing system, whose distributions of service and inter-arrival times have rational Laplace-Stieltjes transforms, limit theorems are derived for the supremum of the virtual waiting time during k successive busy cycles for k→∞. Similarly, for the supremum of the actual waiting times of all customers arriving in k successive busy cycles. Only the cases with the load of the system less than one and equal to one are considered. The limit distributions are extreme value distributions. The results are obtained by first deriving a number of asymptotic expressions for the quantities which govern the analytic description of the system Km/Kn/1. Using these asymptotic relations limit theorems for entrance times can also be obtained, a few examples are given.

Various types of time and event sampling of a stationary and orderly bivariate point process are considered. Fundamental relations between inter-event intervals and the event counting process are derived. Relations between first forward recurrence times and their moments for different types of sampling are obtained.

Estimates are given for the rates of convergence in functional central limit theorems for the queue length process in a sequence of general multiple channel queues. The situation is considered where the traffic intensity in the nth. queue, ρn, tends to ρ ≧ 1 as n approaches infinity. This extends previous work by the author, [6], in which the traffic intensity was fixed ≧ 1.

Consider a Poisson point process with an intensity parameter forming a Markov chain with continuous time and finite state space. A system of ordinary differential equations is derived for the conditional distribution of the Markov chain given observations of the point process. An estimate of the current intensity, optimal in the least-squares sense, is computed from this distribution. Applications to reliability and replacement theory are given. A special case with two states, corresponding to a process in control and out of control, is discussed at length. Adjustment rules, based on the conditional probability of the out of control state, are studied. Regarded as a function of time, this probability forms a Markov process with the unit interval as state space. For the distribution of this process, integro-differential equations are derived. They are used to compute the average long run cost of adjustment rules.

A fundamental identity, due to Miller (1961a), (1962a, b) and Kemperman (1961), is generalized to semi-Markov processes. Thus the identity applies to processes defined on a Markov chain with discrete state space and random walks with Markov dependent steps (Section 2). Wald's identity is discussed briefly in Section 3. Section 4 is a study of the maxima of partial sums, and Section 5 of maxima in a semi-Markov process.

It is here shown that for a deterministic simple epidemic in which the spatial distribution of the contacts made by an infectious individual is negative exponential, propagation of the epidemic as a travelling wave is only possible for velocities at or above a certain minimal velocity; and that there exists exactly one waveform for each such velocity. The method employed is to reduce the equation of propagation to a three-dimensional autonomous differential equation, and attack this mainly using topological considerations.

This result is similar to those obtained by Kolmogoroff et al. (1937) for the advance of an advantageous gene, and by Kendall (1965) for the deterministic “general” epidemic, but these were achieved using the considerable simplification of approximating the spatial dependence by a diffusion term in the equation of propagation. Taken together with Mollison (1972), in which it was shown that when the spatial distribution of contacts (V) is more spread than negative exponential (∫eksdV(s) divergent for all k > 0), no finite bound can be set on the velocity of propagation: the present work suggests that the diffusion-term approximations may be reasonable if, as well as only if, V is less spread than negative exponential.

In the paper a single server semi-Markov queue is considered. The analysis is based on a fundamental matrix identity due to H. D. Miller. A natural method for the solution of semi-Markov queues is indicated; use is also made of a duality relation, which is discussed in the paper.

We consider d-dimensional stochastic processes which take values in (R+)d. These processes generalize Galton-Watson branching processes, but the main assumption of branching processes, independence between particles, is dropped. Instead, we assume for some

Here τ: (R+)d → R+, |x| = Σ1d |x(i)| A = {x ∈ (R+)d: |x| = 1} and T: A → A. Under various assumptions on the maps τ and T it is shown that with probability one there exists a ρ > 1, a fixed point p ∈ A of T and a random variable w such that limn→∞Znρ−n = wp. This result is a generalization of the main limit theorem for super-critical branching processes; note, however, that in the present situation both p and ρ are random as well. The results are applied to a population genetical model for zygotic selection without mutation at one locus.

Low density traffic refers to the study of macroscopic properties of a traffic stream when vehicles travel independently of one another. It is usually assumed that each vehicle travels at a constant velocity, the velocity varying from vehicle to vehicle. We allow very general vehicular motions and study various aspects of the traffic streams. For example, it is shown that if π [a,b] is the expected time for a vehicle to travel from a to b under the stochastic process governing the motion of vehicles, then a non-homogeneous Poisson spatial process with mean measure π is invariant.

Newman (1970) introduced an interesting new class of point processes which he called Gauss-Poisson. They are characterized, in the most general case, by two measures. We determine necessary and sufficient conditions on these measures for the resulting point process to be well defined, and proceed to a systematic study of its properties. These include stationarity, ergodicity, and infinite divisibility. We mention connections with other classes of point processes and some statistical results. Our basic approach is through the probability generating functional of the process.

We give moment approximations to the density function of the wavelength, i. e., the time between “a randomly chosen” local maximum with height u and the following minimum in a stationary Gaussian process with a given covariance function. For certain processes we give similar approximations to the distribution of the amplitude, i. e., the vertical distance between the maximum and the minimum. Numerical examples and diagrams illustrate the results.

The paper gives a survey of the theory of univariate characteristic functions. These functions were originally introduced as tools in the study of limit theorems but it was later realized that they had an independent mathematical interest. Those parts of the theory which can be found in textbooks are treated only briefly; the main emphasis is placed on more recent developments and areas where active research is still in progress.