Institute of Natural and Mathematical Sciences

Permanent URI for this community

https://mro.massey.ac.nz/handle/10179/4224

Browse

Now showing 1 - 20 of 130

64-bit architechtures and compute clusters for high performance simulations
(Massey University, 2006) Hawick, K.A.; James, H.A.; Scogings, C.J.
Simulation of large complex systems remains one of the most demanding of high performance computer systems both in terms of raw compute performance and efficient memory management. Recent availability of 64-bit architectures has opened up the possibilities of commodity computers accessing more than the 4 Gigabyte memory limit previously enforced by 32-bit addressing. We report on some performance measurements we have made on two 64-bit architectures and their consequences for some high performance simulations. We discuss performance of our codes for simulations of artificial life models; computational physics models of point particles on lattices; and with interacting clusters of particles. We have summarised pertinent features of these codes into benchmark kernels which we discuss in the context of wellknown benchmark kernels of the 32-bit era. We report on how these these findings were useful in the context of designing 64-bit compute clusters for high-performance simulations.
Accelerated face detector training using the PSL framework
(Massey University, 2009) Susnjak, T.; Barczak, A.L.C.; Hawick, K.A.
We train a face detection system using the PSL framework [1] which combines the AdaBoost learning algorithm and Haar-like features. We demonstrate the ability of this framework to overcome some of the challenges inherent in training classifiers that are structured in cascades of boosted ensembles (CoBE). The PSL classifiers are compared to the Viola-Jones type cas- caded classifiers. We establish the ability of the PSL framework to produce classifiers in a complex domain in significantly reduced time frame. They also comprise of fewer boosted en- sembles albeit at a price of increased false detection rates on our test dataset. We also report on results from a more diverse number of experiments carried out on the PSL framework in order to shed more insight into the effects of variations in its adjustable training parameters.
Agglomerate properties
(Massey University, 2003) Rynhart, P.R.; Jones, J.R.; McKibbin, R.
Modelling of wet granulation requires the rate of agglomerate coalescence to be estimated. Coalescence is dependent on the frequency of collisions that occur, and the fraction of collisions which result in coalescence. The collision rate is a function of granulator kinetics and powder properties, while the coalescence success rate is dependent on factors including the Stokes number and particle geometry. This work investigates an aspect of the geometry by examining the distribution of liquid on the surface of agglomerates in the capillary state. Agglomerates are created by adding particles, one at a time, about a central tetrahedral arrangement of four primary particles. For a given agglomerate, the wetted fraction of surface area, defined as the wetness, is evaluated using an approximate fluid surface. Packing density and binder saturation parameters are incorporated into the model. Given a number of primary particles and the volume of binder in a particle, the agglomerate wetness is able to be estimated using computational geometry.
Algebraic conversions
(Massey University, 2001) Mills, Bruce
An examination of the pure algebraic properties of computational type conversion leads to a new generalizations of the concept of a homomorphism for which the term conversion seems appropriate. While an homomorphism is a mapping that respects the value of all terms, a conversion is a mapping that respects the value of all sufficiently small terms. Such a mapping has practical value, as well as theoretical interest that stems from conversions forming a category. This paper gives a precise definition of the concept and demonstrates an application to formal computer science based on work completed by the author in his PhD thesis.
Analysis of a non-minimum phase acoustic beamformer
(Massey University, 2004) Moir, T.J.
The two input Griffiths-Jim acoustic beamformer is analysed in the frequency domain using a Wiener type formulation. Unlike previous solutions the approach here is to look at the problem of non-minimum phase acoustic transfer functions which are encountered in many real filtering problems. The polynomial transfer function approach gives an elegant way of obtaining the frequency response of the beamformer and gives new insight to the problem in general.
Analysis of logistic growth models
(Massey University, 2001) Tsoularis, A.
A variety of growth curves have been developed to model both unpredated, intraspecific population dynamics and more general biological growth. Most successful predictive models are shown to be based on extended forms of the classical Verhulst logistic growth equation. We further review and compare several such models and calculate and investigate properties of interest for these. We also identify and detail several previously unreported associated limitations and restrictions. A generalized form of the logistic growth curve is introduced which is shown incorporate these models as special cases. The reported limitations of the generic growth model are shown to be addressed by this new model and similarities between this and the extended growth curves are identified. Several of its properties are also presented. We furthermore show that additional growth characteristics are accommodated by this new model, enabling previously unsupported, untypical population dynamics to be modelled by judicious choice of model parameter values alone.
Analysis of reinforcement learning strategies for predation in a mimic-model prey environment
(Massey University, 2005) Tsoularis, A.; Wallace, J.
In this paper we propose a mathematical learning model for a stochastic automaton simulating the behaviour of a predator operating in a random environment occupied by two types of prey: palatable mimics and unpalatable models. Specifically, a well known linear reinforcement learning algorithm is used to update the probabilities of the two actions, eat prey or ignore prey, at every random encounter. Each action elicits a probabilistic response from the environment that can be either favorable or unfavourable. We analyse both fixed and varying stochastic responses for the system. The basic approach of mimicry is defined and a short review of relevant previous approaches in the literature is given. Finally, the conditions for continuous predator performance improvement are explicitly formulated and precise definitions of predatory efficiency and mimicry efficiency are also provided.
Analysis of the last page
(Massey University, 2003) McDonald, Barry
A sample of death notices from the New Zealand Herald was used as the basis of a Data Analysis assignment. This note explores some interesting statistical aspects of these death notices, using common data analysis techniques, and illustrates how they can be used as a resource for teaching. In particular they provide a clear example of biased sampling, a concept that is usually hard to quantify.
Antieigenvalues and antisingularvalues of a matrix and applications to problems in statistics
(Massey University, 2005) Rao, Radhakrishna
Let A be p × p positive definite matrix. A p-vector x such that Ax = x is called an eigenvector with the associated with eigenvalue . Equivalent characterizations are: (i) cos = 1, where is the angle between x and Ax. (ii) (x0Ax)−1 = xA−1x. (iii) cos = 1, where is the angle between A1/2x and A−1/2x. We ask the question what is x such that cos as defined in (i) is a minimum or the angle of separation between x and Ax is a maximum. Such a vector is called an anti-eigenvector and cos an anti-eigenvalue of A. This is the basis of operator trigonometry developed by K. Gustafson and P.D.K.M. Rao (1997), Numerical Range: The Field of Values of Linear Operators and Matrices, Springer. We may define a measure of departure from condition (ii) as min[(x0Ax)(x0A−1x)]−1 which gives the same anti-eigenvalue. The same result holds if the maximum of the angle between A1/2x and A−1/2x as in condition (iii) is sought. We define a hierarchical series of anti-eigenvalues, and also consider optimization problems associated with measures of separation between an r(< p) dimensional subspace S and its transform AS. Similar problems are considered for a general matrix A and its singular values leading to anti-singular values. Other possible definitions of anti-eigen and anti-singular values, and applications to problems in statistics will be presented.
Appoximation-assisted [sic] estimation of eigenvectors under quadratic loss
(Massey University, 2005) Ahmed, S.E.
Improved estimation of eigen vector of covariance matrix is considered under uncertain prior information (UPI) regarding the parameter vector. Like statistical models underlying the statistical inferences to be made, the prior information will be susceptible to uncertainty and the practitioners may be reluctant to impose the additional information regarding parameters in the estimation process. A very large gain in precision may be achieved by judiciously exploiting the information about the parameters which in practice will be available in any realistic problem. Several estimators based on preliminary test and the Stein-type shrinkage rules are constructed. The expressions for the bias and risk of the proposed estimators are derived and compared with the usual estimators. We demonstrate that how the classical large sample theory of the conventional estimator can be extended to shrinkage and preliminary test estimators for the eigenvector of a covariance matrix. It is established that shrinkage estimators are asymptotically superior to the usual sample estimators. For illustration purposes, the method is applied to three datasets.
Automated schema matching techniques: an exploratory study
(Massey University, 2003) Sun, Xiao Long; Rose, Ellen
Manual schema matching is a problem for many database applications that use multiple data sources including data warehousing and e-commerce applications. Current research attempts to address this problem by developing algorithms to automate aspects of the schema-matching task. In this paper, an approach using an external dictionary facilitates automated discovery of the semantic meaning of database schema terms. An experimental study was conducted to evaluate the performance and accuracy of five schema-matching techniques with the proposed approach, called SemMA. The proposed approach and results are compared with two existing semi-automated schema-matching approaches and suggestions for future research are made.
Automatic C library wrapping Ctypes from the trenches
(Massey University, 2009) Kloss, Guy K.
At some point of time many Python developers at least in computational science will face the situation that they want to interface some natively compiled library from Python. For binding native code to Python by now a larger variety of tools and technologies are available. This paper focuses on wrapping shared C libraries, using Python's default Ctypes. Particularly tools to ease the process (by using code generation) and some best practises will be stressed. The paper will try to tell a step by step story of the wrapping and development process, that should be transferable to similar problems.
Automatic variance control and variance estimation loops
(Massey University, 2000) Moir, T.J.
A closed loop servo approach is applied to the problem of controlling and estimating variance in nonstationary signals. The new circuit closely resembles but is not the same as, automatic gain control (AGC) which is common in radio and other circuits. The closed loop nature of the solution to this problem makes this approach highly accurate and can be used recursively in real time.
Automotive three-microphone voice activity detector and noise-canceller
(Massey University, 2005) Qr, Z; Moir, T.J.
This paper addresses issues in improving hands-free speech recognition performance in car environments. A three-microphone array has been used to form a beamformer with leastmean squares (LMS) to improve Signal to Noise Ratio (SNR). A three-microphone array has been paralleled to a Voice Activity Detection (VAD). The VAD uses time-delay estimation together with magnitude-squared coherence (MSC).
Bounds on expected coupling times in Markov chains
(Massey University, 2008) Hunter, Jeffrey J.
In the author’s paper “Coupling and Mixing Times in Markov Chains” (RLIMS, 11, 1- 22, 2007) it was shown that it is very difficult to find explicit expressions for the expected time to coupling in a general Markov chain. In this paper simple upper and lower bounds are given for the expected time to coupling in a discrete time finite Markov chain. Extensions to the bounds under additional restrictive conditions are also given with detailed comparisons provided for two and three state chains.
Building an inexpensive parallel computer
(Massey University, 2000) Grosz, Lutz; Barczak, Andre
Building cost-effective research platforms: utilising free | open-source software in research projects
(Massey University, 2003) Meyer, Tony
When prototyping or developing a system for use in research work, it is often necessary to create an entire system, even if only one part of the system is the focus of the research. Free | open-source software offers a solution to this problem, allowing the creation of cost-effective research platforms, utilising peer-reviewed, rapidly-developed code that is easily modified. One form of free | open-source software that is regularly used in research projects is engines from 3d games such as Unreal Tournament (Lewis & Jacobson, 2002). Although the core rendering engine is proprietary, the game engine is able to be freely utilised as a reasonably generic rendering engine and physics simulator, and most of the game code is modifiable. A synthetic characters development project, outlined in this paper, uses the Unreal Tournament game engine, via the Gamebots socket tool, as its primary output system. Several other free | open-source software packages are utilised, including a speech recognition system (Sphinx from Carnegie Mellon University), a speech generation system (Festival-Light from Carnegie Mellon University), and a vision system built with Lego™ Mindstorms™ (and the open-source NQC), cheap web-cameras and Intel®’s OpenCV library. These modules all communicate via standard network sockets and are able to operate independently. Each module required a different level of modification in order to form part of the synthetic actors system, from no modification at all (Unreal Tournament), to light modifications (Festival-Light), to a new system based on open-source code (the NQC-based tripod code); one package – the OpenCV library – is simply linked to by completely new code.
Building knowledge-based economies: research projects in knowledge management and knowledge transfer
(Massey University, 2000) Gibson, Virginia R.
Small and medium-sized enterprises (SMEs) are viewed as the growth engines of the new knowledgebased economy. This new economic growth model differs from the old in significant ways, many of which are related to the knowledge base that will be required by the SMEs. Based upon prior research a set of factors important to the success of SMEs in a knowledge-based economy is described. Focusing on those factors related to the knowledge base, the paper concludes with a set of research questions and brief descriptions of three research projects on knowledge management and knowledge transfer.
C/C++ implementation of functions of the class LT0
(Massey University, 2000) Calude, Elena; Kay, Peter; Luo, Weiwei
This report describes an on-going implementation, in C/C++, of the functions and schemes of the formal system LT0, presented in the paper Caporaso, Pani and Covino [1]. The final aim is to be able to effectively construct a "small manageable" Exponential Diophantine Equation which represents (in the sense of Chaitin [2]) an algorithmical random binary sequence.
Calculating animal performance from limited liveweight measurements of the population
(Massey University, 2003) Wake, G.C.; Pleasants, A.B.
We consider the problem of estimating the distribution of carcass weights in a flock of animals from estimates made on a truncated sample. This arises when a farmer chooses the heaviest lambs for slaughter and then measurements are made by the meat processor. This enables a farmer to answer two questions: what proportion of the animals remaining exceed a nominated carcass weight, and/or what carcass weight is exceeded by a nominated proportion of the population? Estimates of these statistics and their uncertainties are derived and are exact if the animal weights are normally distributed. These calculations can be the basis of decisions about future feeding and drafting strategies, important for farmers producing animals on contracts for future delivery. An example is given based on 1000 lambs using a cut-off weight of 15.5 kg with mean of this upper group of 16 kg. Using a realistic estimate of a standard deviation (of the weighing scales) of 0.3 kg, this gives an estimated mean of 14.6± 0.04 kg, with a standard deviation of 0.94±0.044 kg, and that 75% of the lambs in the population exceed 13.9±0.07 kg. The proportion of lambs that exceed 14.5 kg is then between 51.3% and 55.7%.

Browse

Browsing Institute of Natural and Mathematical Sciences by Title

Results Per Page

Sort Options