Department of Mathematics and Computing Seminars

The ability to run multiple versions of the same code with different input parameters is becoming increasingly important in many disciplines (e.g. parameter searching and sensitivity testing etc.). USQ's HPC computation environment is ideal for this purpose. This seminar will use two case studies from the environmental sciences to illustrate the tools available on HPC which allow this. The first case study illustrates how the same Fortran/C code can be run on multiple cores with different input parameters using UNIX scripts and the Portable Batch Submission (PBS) system. The second case study will illustrate how same Matlab function can be run on several cores with different input parameters.

Understanding sampling variability is an important goal. A useful approach is to consider replication. What information do inferential techniques give about replication? Confidence intervals (CIs) do well: A 95% CI is, approximately, an 83% prediction interval for the mean of a replication experiment. A p value, however, gives almost no information about a replication result. CIs beat p values, although for very small N a CI may be misleading. I will present novel graphics and simulations, and emphasise cognition - how people read graphics and draw conclusions. There is more in my book: Understanding The New Statistics: Effect Sizes, Confidence Intervals, and Meta-Analysis, Routledge, 2012. www.thenewstatistics.com

This seminar will introduce the basic Windows software UNIX and Matlab commands required for accessing USQ's HPC environment (e.g. WinSCP, Putty). The basic philosophy and structures used in parallel computation will be discussed in relationship to Matlab code design and implementation. Practical examples will be provided on the usage of Matlab's Parallel Toolbox and the Torque submission system within the HPC environment.

We focus on dynamics of phase of oscillations in a variety of chemical, physical and biological systems. An important mechanism of such dynamics is due to nonlinear self-excitation. We plan to numerically solve the phase equation (which is a partial differential equation) in 1D and 2D; investigate the dependence of solutions on the controlling parameters, derive different variants of the nonlinearly excited phase equation and explore regular and chaotic regimes of the phase dynamics.

Following the success of establishing USQ Cisco Networking Academy and embedding the Cisco Certified Networking Associate (CCNA) certificate training courses into USQ Networking and Security Major of the BIT Program in 2007, we have been working hard to set up a remote access facility for external students to access, configure and manipulate real networking equipment for students' hands-on practicals. The facility will also help on-campus students who are not able to come to Toowoomba campus for the lab work for various reasons. The remote access will provide all students maximal flexibility for their learning and laboratory needs.

This networking remote access project is a practical application of USQ Open and Fleximode policy. It is the "pioneering" work in Australia in networking education for completely remote access and control of network equipment, and place USQ at the forefront of flexible education.

In this seminar, we will introduce the project, present the networking remote access system through the demonstrations of several real-time lab sessions.

Coherent structures are responsible for non homogeneous transport in atmosphere and ocean driving vertical mixing and pollutants concentration.

High resolution numerical models and associated measurement tools have been recently developed including HF and VHF radar technique including interferometry. Front capturing techniques and Lagrangian trajectories are commonly performed to better localize vortices and barriers.

Appropriate scales have to be identified from phycical processes in order to better underdstand and predict unexpected complex structures like filaments and reverse flows acting as retention zones for biogeochemical production. Examples are given both in atmospheric boundary layer and coastal flows

Non-cdn is a CDN designed around the principle that every request should be authenticated. This is different to other CDN's where the default rule is to allow access to the system by default and in some situations provide limited access control. A perfect example of this is the Facebook CDN where if you have the direct URL to an item you can always retrieve it regardless of if you would be permitted to access the file via the Facebook web page. Non-CDN aims to provide the ability to authenticate and authorise every request, log access and provide the ability to detect potentially abusive behaviour from clients. The system's design is inspired by potential needs for securely distributing course content at USQ in ways that can safely provide the benefits of a CDN.

Meta-analysis combines results from several independent studies. However, meta-analysis with more than two outcomes has not been carried out as frequently as with binary outcomes. Data from studies with several outcome categories are analysed by various methods including proportional odds model which requires a `proportionality assumption'. Other methods estimate the effect size by collapsing the 2xL contingency table of randomised controlled trials (RCTs) into 2x2 table. Such merging causes loss of information and unrealistic reduction of spread by artificially creating two homogeneous categories. To overcome these drawbacks, we propose generalised odds ratio (GOR) as an effect measure for ordinal categorical outcomes. A meta-analysis method is developed using GOR for ordinal categorical outcomes under independent multinomial distribution. The proposed method is illustrated using data from five RCTs of anti-cholinesterase drug tacrine in patients with Alzheimer's disease and the results compared with that of other existing methods. Also a quasi-empirical Bayes method (QEBM) is developed for heterogeneous ordinal outcomes. This method is useful in identifying the extreme studies and improving the meta-analysis in the presence of heterogeneity. Three different examples of various degree of heterogeneity are also presented.

A dynamical evolution equation governing concentration of contaminants, averaged across the flow, in channels and boundary layers can serve as an effective tool for prediction of the spreading of the contaminants in environmental and industrial situations. An accurate principle for constructing such an equation, based on centre manifold theory, has been proposed by Roberts and co-authors. However, the dynamical structure of turbulence has not been fully taken into account in the models.
The current project aims to: (1) Formulate an analytical framework of the averaged transport of contaminants in turbulent boundary layers near smooth and rough substrates, (2) Derive advection-, diffusion-, dispersion- and higher-order coefficients for the contaminant transport in channels, (3) Develop and analyse averaged transport models for the flows through canopies, (4) Numerically justify the derived models by comparing solutions with solutions of the original (non-averaged) transport equations.