18 results for Whigham, Peter A, Conference paper, 2000

  • Fixation of neutral alleles in spatially structured populations via genetic Drift: Describing the spatial structure of faster-than-panmictic configurations

    Whigham, Peter A; Dick, Grant (2005-11)

    Conference paper
    University of Otago

    This paper considers spatially-structured populations described as a network, and examines the properties of these networks in terms of their affect on fixation of neutral alleles due solely to genetic drift. Individuals are modelled as two allele, one locus haploid, diploid and tetraploid structures. The time to fixation for a variety of network configurations is discovered through simulation. The concept of hyperfixation is introduced, which refers to when time to fixation for a network of n nodes occurs more rapidly than the corresponding panmictic n node structure. A hyperfixation index, h, is developed that attempts to characterise a spatial arrangement such that when h < 1 hyperfixation will occur. Issues regarding fixation with ploidy independence, and possible improvements to the described hyperfixation index are discussed.

    View record details
  • The use of boundary conditions for inductive models

    Whigham, Peter A (2004-11)

    Conference paper
    University of Otago

    There is a large amount of interest in creating models from data using a variety of machine learning methods. Most of these approaches require a good distribution of observed values to produce reliable models. The use of background knowledge to augment the observed values has also been explored as a method to supplement the original feature set of training data. This paper argues that there is an additional set of data that can be created for many types of problems, based on the concept of boundary conditions. This boundary data incorporates an understanding of the modeled system behaviour under certain extreme values and therefore reduces the degrees of freedom within the inferred model. This paper argues that by using this information when training an inductive model a more robust generalization of the data can be achieved under some circumstances.

    View record details
  • A preliminary investigation of the stability of Geographically-Weighted Regression

    Whigham, Peter A; Hay, Geoff (2007-12-06)

    Conference paper
    University of Otago

    This paper describes preliminary work analysing the stability of parameter coefficient estimates for Geographically-Weighted Regression (GWR). Based on a large dataset (35721 points) various random samplings of this data were performed and models built using GWR. An analysis of the coefficient values for the independent variables showed that these values could varying significantly both between runs and between sampling sizes. This suggests that the results from GWR must be carefully considered in terms of the form of data, assumed coefficient surface being modelled, and the confidence of the resulting parameter estimates.

    View record details
  • A vector-agent paradigm for dynamic urban modelling

    Hammam, Yasser; Moore, Antoni; Whigham, Peter A; Freeman, Claire (2003-12)

    Conference paper
    University of Otago

    In the past, researchers and model developers were restricted by their theoretical knowledge about the city and how it might be simulated as well as constrained by technological limitations. Nevertheless, the simulation environment is now appropriate for the infusion of new ideas into urban modelling. Urban simulation is a relatively unique modelling problem. The urban systems commonly represented in urban models considering different factors (economic, social, environmental …. etc.) and are notoriously difficult to simulate. It is proposed that agents are more flexible than cellular automata at modelling the city, and in turn that vector-based agents are more suitable than cell-based agents at doing the same. Therefore, this paper is intended to review the limitations raised by using cell-based models with an agent simulation system, and introduce a new paradigm for integrating vector-based spatial model with agent system.

    View record details
  • Discovering population structures with extreme fixation rates via evolutionary search

    Dick, Grant; Whigham, Peter A (2005-11)

    Conference paper
    University of Otago

    Genetic drift is a well known and important force in directing the evolution of a population. The nature of genetic drift in panmictic populations is well understood, and new research is shedding light on the behaviour of genetic drift. This paper explores the concept of using evolutionary algorithms to search for population structures that exhibit the minimum and maximum conditions for loss of variation via genetic drift. Two spatial structures repeatedly emerge as candidates: a star topology that reduces fixation time to a logarithm of population size, and a “line with islands” topology that can delay fixation via genetic drift to a greater extent than any previously known population structure.

    View record details
  • Irregular vector-agent based simulation for land-use modelling

    Hammam, Yasser; Moore, Antoni; Whigham, Peter A; Freeman, Claire (2004-11)

    Conference paper
    University of Otago

    Urban structures exhibit complex patterns made of heterogeneous and irregular objects. Few works in the computational urban modelling literature have considered and examined the real geometric boundary of the city’s objects. However, most of these works are driven by Cellular Automata (CA) as a spatial modelling vehicle. This model has had success, but also has its limitations regarding the study of urban dynamics in computer simulation. Extensive modification of CA or use of a different modelling paradigm should be considered. We argue here that, representational realism must be achieved in urban complexity. This paper is an attempt to fill this gap to address the rigid structure of CA: we present a novel technique called the “vector-agent based simulation”, which uses discrete irregular objects as an autonomous spatial entity beneath an agent modelling structure. Through computer simulation, this new technique has been applied to von Thunen’s theory of agricultural land use as a hypothetical environment for model verification. The findings demonstrate that our proposal can be a new paradigm for urban simulation

    View record details
  • Sport and time geography: a good match?

    Moore, Antoni; Whigham, Peter A; Holt, Alec; Aldridge, Colin H; Hodge, Ken (2003-12)

    Conference paper
    University of Otago

    This paper proposes using the rich visual “language” of Hägerstrand’s time geography to represent time-space relationships in sport, in particular within the spatial and temporal constraints of a game of rugby. Despite being applied outside of its traditional social context, it is argued that time geography’s ability to model movements and relationships at the individual level makes it (and its modelling constructs such as prisms and lifelines) a powerful visualisation tool able to provide valuable insights into goal-oriented team sport. The visual tools of time geography are shown in the context of a video information system, SCRUM (Spatio-Chronological Rugby Union Model).

    View record details
  • Generic vector-agents

    Hammam, Yasser; Moore, Antoni; Whigham, Peter A; Freeman, Claire (2005-11)

    Conference paper
    University of Otago

    The work reported here has been motivated by the need for a generic spatial model to overcome the limitations of Cellular Automata (CA) regarding the rigid square-cell structure and limited neighbourhood configurations. A novel approach for spatial modelling technique is developed: the “vector-agent” in which the individual entity is represented by their real geometric boundaries (which can change over time) beneath an agent modelling structure. We show in this paper how the theory behind CA and agents can be combined to produce a generic and dynamic agent based on the vector data structure. This new paradigm has extended capabilities over the Geographic Automata (Torrens and Benenson, 2005) in terms of CA disunity and the abstraction of non-fixed-objects. Through computer simulation, different techniques and algorithms have been derived achieving a high degree of representational realism for a variety of phenomena

    View record details
  • The storage and reconstruction of polygon boundaries using circles

    Moore, Antoni; Mason, Chris; Whigham, Peter A; Thompson-Fawcett, Michelle (2003-12)

    Conference paper
    University of Otago

    This paper investigates the accuracy of using circles to store polygon boundaries. Can a series of xy points be accurately replaced by a smaller array of variably-sized circles used in a non space filling sense to approximate to the edge of the polygon? Two measures were used to ascertain this, areal error and visual error. A vector dataset representing the island of Rarotonga was converted from boundary coordinates to circle storage using a Voronoi-based medial axis approach. Three parameters were altered to derive as many as 150 circle-based realisations – minimum allowed circle radius, degree of overlap allowed between circles and Douglas-Peucker algorithm threshold (the latter algorithm of line reduction was used to extract “important” points from the polygon and add them to the circle list as “zero-radius circles”). This presentation presents and analyses the areal error and visual error results for all realisations.

    View record details
  • Genetic tradeoff as a model for parapatric speciation

    Jakoby, Oliver; Whigham, Peter A (2005-11)

    Conference paper
    University of Otago

    This paper uses an individual-based modelling approach to explore the effects of genetic tradeoff across a spatial gradient. Tradeoff is represented by the ability of an individual to live at higher environmental values - individuals that can live at high values can also live at low values, but the cost for this generalised ability is an increased time to breeding maturity. This paper shows that for very low tradeoff gradients significant zones still occur across the environment, with a pronounced set of heterozygotic bands. It is proposed that these bands represent hybrid zones, and that the model demonstrates that weak tradeoff across an environmental gradient can produce isolated genetic individuals. This model therefore shows one mechanism leading to parapatric speciation.

    View record details
  • Keep your eye on the ball: Local versus global statistics in sport visualisation

    Whigham, Peter A (2005-11)

    Conference paper
    University of Otago

    This paper describes several simple local space-time statistical measures that can be applied to data that describes ball position over space and time. The main aim is to show that global statistics of rugby union do not describe the most important aspects of team domination, and that other measures may be more meaningful in terms of understanding the overall patterns of play. This paper introduces several measures for describing these local statistics, and shows for one example that they reflect the dominance of one team even though this is not reflected in the half-time score. The main conclusion is that the most important local statistics maintain an explicit spatial representation, and that therefore global measures of behaviour, that have removed the spatial patterns of play, are not appropriate descriptors for rugby union.

    View record details
  • How does space alter the formulation of evolutionary models?

    Whigham, Peter A; Dick, Grant (2006-11)

    Conference paper
    University of Otago

    This short paper discusses how space is introduced within evolutionary models and reviews several approaches from the literature. In particular, the Moran process is used as one type of model that has been described in several ways when space is introduced. The evolutionary operations of parental selection, offspring placement, synchronous and asynchronous updating, survival and death timings are shown to be factors that may be interpreted in a number of ways once space is introduced into a model. One conclusion from this discussion is that the introduction of space requires a consideration of the time-step in models, directly through event driven processes or indirectly through the random selection of space prior to a local process.

    View record details
  • Point allocation inside polygons and GWR: an experimental analysis with survey data

    Francisco, Eduardo de Rezende; Whigham, Peter A; Moore, Antoni (2007-12-06)

    Conference paper
    University of Otago

    The aim of this paper is to analyse different alternative implementations for a problem defined as "point allocation inside polygons" for Geographically Weighted Regression (GWR). The problem involves situations where the precise location of each observation is not known - just its district, municipality or region, i.e. a polygon geographical location. However, associated data were available that could potentially allows point placement of observations. These analyses were applied in a Income predicting model based on electricity consumption from a survey for a power distribution company in Sao Paulo, Brazil. Completely spatially random allocation and allocations based on spatial distributions of population (universe) and of the independent variable (electricity consumption) were utilized. Results showing the coefficients of determination (R2) suggest that a more realistic measure of the relationship between these two constructs could be evaluated.

    View record details
  • Spatial analysis of home range, core area and habitat selection of red deer (Cervus elaphus) on an extensively managed high-country station

    Netzer, Michael; Whigham, Peter A; Dickinson, Katharine; Lord, Janice (2007-12-06)

    Conference paper
    University of Otago

    The aim of this research was to determine the behavior and habitat selection of red deer (Cervus elaphus) hinds during the calving period, on an extensively managed rangeland in the high-country of New Zealand’s South Island. The research was developed using ArcView and eCognition software, GPS collars on five red deer hinds, an aerial photograph and topographic data. The results showed that three out of the five study deer escaped the original study paddock into adjacent extensive paddocks. The deer did not use the entire study paddocks they were in, but instead formed distinct home ranges (total area of movement) and core areas (areas of intensive use, 45% of deer locations). These home ranges were highly variable in size from 225 ha. To 36 ha.. Core areas ranged from 102 ha. To 36 ha.. The variation in size of home ranges and core areas was directly related to the size of the paddock, and indirectly to social constraints within the paddock. The results suggest that deer within these extensive paddocks are under strong densitydependent effects during the calving season that may be forcing some hinds (probably less-dominant hinds) to the marginal edges of the pasture, where they may eventually escape. Habitat selection indicated, that while red deer often selected the naturalized pasture grass areas (high in metabolized energy), tussock grassland, when present in the home range, was also highly selected for. The selection of tussock grasslands was probably a result of tussock habitat providing both good cover/security during the calving season, and the presence of reasonably good forage in the inter-tussock area.

    View record details
  • A spatially-explicit model of genetic tradeoff

    Whigham, Peter A; Green, David (2004-12)

    Conference paper
    University of Otago

    Spatially-explicit, individual based models are used to explore the consequences of introducing two simple genetic tradeoff mechanisms to the gradient response of monoecious diploid individuals. A simple linear gradient is used to produce a structured environment. One tradeoff relates the plasticity of an individual to its overall fitness, where individuals that can tolerate a wide range of environmental values are less fit than individuals with a narrow tolerance range. The second tradeoff assumes that individuals that can live at greater environmental values are placed under stress and therefore reach maturity later in their life cycle. Here all individuals have the same fitness, but the penalty for tradeoff is less time for breeding after reaching maturity. In both cases the resulting population distributions produced clear genetic differentiation, showing that simple tradeoff mechanisms are one route towards genetic isolation and, ultimately, speciation.

    View record details
  • Spatial variation in the association between neighbourhood deprivation and access to alcohol outlets

    Hay, Geoff; Whigham, Peter A; Kypri, Kypros; Langley, John (2007-12-06)

    Conference paper
    University of Otago

    Background: People who live in poor areas suffer higher mortality than those living in wealthier areas. People who live in poor areas in New Zealand have better access to alcohol and this association appears to vary spatially. We sought to investigate this spatial non-stationarity using Geographically Weighted Regression. Methods: The location of bars was geocoded for all of New Zealand and closest facility analysis was used to calculate distance to the nearest bar from each meshblock. A neighbourhood level census-based index of socioeconomic deprivation, and urban/rural status data were added as inputs to a Geographic Weighted Regression model to investigate spatial variation in the association between access to alcohol outlets and deprivation. Results: Spatial non-stationarity was discovered in deprivation and urban/rural status parameters with some large rural areas of New Zealand exhibiting significant departures from the global model of the association between distance to the nearest bar and neighbourhood deprivation. Conclusions: Lack of association discovered for rural areas may be the result of spatial heterogeneity. Research into the association between deprivation and access to alcohol should consider rural areas individually for environmental inequity rather than relying on global models showing no association.

    View record details
  • A Voronoi-based distributed genetic algorithm

    Whigham, Peter A; Dick, Grant (2003-12)

    Conference paper
    University of Otago

    The use of space for supporting evolution has been previously studied in the context of distributed Genetic Algorithms (DGA), where two standard approaches, island and grid based, are employed to define the population structure and connectivity relationships between individuals. The grid-based approach uses a fixed, regular grid to define the neighbourhood relations between individuals, resulting in Moore or Von Neumann relationships between population individuals. This short paper begins to address the question of the influence of non-fixed spatial relationships between individuals in a distributed genetic algorithm, where the sub-population of each population member is defined by the 1st order Voronoi neighourhood of that individual. Initial results suggest that the irregular nature of the distribution produces an improved performance for the DGA, and that the Voronoi model of neighbours is appropriate for dynamic environments.

    View record details
  • Fitness landscapes and gene location

    Whigham, Peter A (2004-11)

    Conference paper
    University of Otago

    The paper outlines initial results into a study of the preference of gene location for diploid organisms under the presence of an environmental gradient. The work explores the properties of a spatially-explicit model of monoecious diploid individuals that evolve preferential coding of gene location. The individuals evolve over a space with a linear gradient, where the response to an individual’s phenotype determines the age before breeding. Each individual has three chromosomes that determine the position and value of two genes. These genes combine to determine the resulting breeding response. The concept of an NK fitness landscape is used to represent two different scenarios of gene linkage. The results indicate that when gene linkage is related by a rugged fitness landscape, the genes cluster on the same chromosome, whereas when the fitness landscape is smooth the genes are more likely to be on separate chromosomes. The work has implications for understanding some of the possible mechanisms that lead to gene clusters.

    View record details