Computationally Efficient Methods for MDL-Optimal Density Estimation and Data Clustering

The Minimum Description Length (MDL) principle is a general, well-founded theoretical formalization of statistical modeling. The most important notion of MDL is the stochastic complexity, which can be interpreted as the shortest description length of a given sample of data relative to a model class. The exact definition of the stochastic complexity has gone through several evolutionary steps. The latest instantation is based on the so-called Normalized Maximum Likelihood (NML) distribution which has been shown to possess several important theoretical properties. However, the applications of this modern version of the MDL have been quite rare because of computational complexity problems, i.e., for discrete data, the definition of NML involves an exponential sum, and in the case of continuous data, a multi-dimensional integral usually infeasible to evaluate or even approximate accurately. In this doctoral dissertation, we present mathematical techniques for computing NML efficiently for some model families involving discrete data. We also show how these techniques can be used to apply MDL in two practical applications: histogram density estimation and clustering of multi-dimensional data.

Magnetohydrodynamic flow between torsionally oscillating eccentric disks

The unsteady MHD flow of an incompressible, viscous electrically conducting fluid contained between two torsionally oscillating eccentric disks has been investigated. The state of uniform rotation of the central region visualised in the steady flow is seen to be absent in the case of oscillatory flow.

Gene flow from Corymbia hybrids in northern New South Wales

Some of the most productive taxa for forestry are interspecific F1 hybrids grown as exotics in the tropics and subtropics. Attributes of resilience, adaptability and vigour which engender the hybrids for wood production, may also exacerbate the risk they present from gene flow to native species gene pools or to local ecologies as weeds. To determine the biological and genetic factors that influence the extent of hybridisation, we examine the distribution and genealogy of wildlings surrounding plantings of locally-exotic Corymbia torelliana (Section Cadageria) near native C. henryi (Section Maculatae) in northern New South Wales. Our study showed pre-mating and pre- and post-zygotic barriers were incomplete, with in situ generation and natural establishment of both F1 hybrids (n = 3) and advanced generation hybrids under the disturbed conditions bordering native forest. As hybrids were located on alluvial flats exposed to frost, they also likely have an extended ecological range relative to native C. henryi. Despite the likely generation of large viable seed crops on F1 trees at the site over many years, establishment success and survival of advanced generation hybrids may be low, as only 5 immature and no mature advanced generation hybrids were identified. Propagation and genetic analysis of a seed crop from one F1 wildling showed early survival and vigour of seedlings in cultivation was high, and that at least for some F1 in some seasons, backcrossing to the recurrent native C. henryi parent is favoured (60%), whereas selfing (10%) and crossing with other F1 (30%) was less frequent. Transport of seed by stingless bees probably accounted for long distance dispersal from C. torelliana, but this mechanism does not appear to supplement gravity-dispersal of seed from the F1. Coupled with other evidence from studies of bee behaviour, controlled pollination in Corymbia sp., and long-term fitness in second generation eucalypt hybrids, we anticipate gene flow via pollen rather than seed will be the greater challenge for managing the risk of introgression of C. torelliana ancestry into native species from the planted F1 hybrid. If large sources of F1 pollen become available to compete with native pollen, gene flow will probably be frequent and hybrids may establish in disturbed conditions and in habitats beyond the ecological range of their native parent. Further study is needed to determine the degree to which outbreeding depression and poor survival inhibits on-going gene flow.

Unsteady laminar boundary layers in a compressible stagnation flow

The unsteady laminar compressible boundary-layer flow in the immediate vicinity of a two-dimensional stagnation point due to an incident stream whose velocity varies arbitrarily with time is considered. The governing partial differential equations, involving both time and the independent similarity variable, are transformed into new co-ordinates with finite ranges by means of a transformation which maps an infinite interval into a finite one. The resulting equations are solved by converting them into a matrix equation through the application of implicit finite-difference formulae. Computations have been carried out for two particular unsteady free-stream velocity distributions: (1) a constantly accelerating stream and (2) a fluctuating stream. The results show that in the former case both the skin-friction and the heat-transfer parameter increase steadily with time after a certain instant, while in the latter they oscillate thus responding to the fluctuations in the free-stream velocity.

Intrinsic equations of steady rotating, incompressible viscous fluid flow

The equations governing the flow of a steady rotating incompressible viscous fluid are expressed in intrinsic form along the vortex lines and their normals. Using these equations the effects of rotation on the geometric properties of viscous fluid flows are studied. A particular flow in which the vortex lines are right circular helices is discussed.

Flow field and traverse times for fan forced injection of fumigant via circular or annular inlet into stored grain

Fan forced injection of phosphine gas fumigant into stored grain is a common method to treat infestation by insects. For low injection velocities the transport of fumigant can be modelled as Darcy flow in a porous medium where the gas pressure satisfies Laplace's equation. Using this approach, a closed form series solution is derived for the pressure, velocity and streamlines in a cylindrically stored grain bed with either a circular or annular inlet, from which traverse times are numerically computed. A leading order closed form expression for the traverse time is also obtained and found to be reasonable for inlet configurations close to the central axis of the grain storage. Results are interpreted for the case of a representative 6m high farm wheat store, where the time to advect the phosphine to almost the entire grain bed is found to be approximately one hour.

A method for the experimental evaluation of the acoustic characteristics of an engine exhaust system in the presence of mean flow

The paper deals with an exact analysis of standing waves in an impedance tube with mean flow. A method is offered for the experimental evaluation of the various wave parameters. Navier–Stokes equations have been solved for evaluating the volume velocity taking into account mean flow, viscosity, etc. The engine exhaust system has been characterized as an acoustic source with an acoustic pressure and internal impedance. A method is suggested for the evaluation of these hypothetical parameters using the exhaust pipe as an impedance tube.Subject Classification: [43]85.20; [43]20.40.