Author : Chandrasekhar Venkataraman
Publisher :
Page : pages
File Size : 23,78 MB
Release : 2011
Category :
ISBN :
In this thesis we investigate a model for biological pattern formation during growth development. The pattern formation phenomenon is described by a reaction-diffusion system on a time-dependent domain. We prove the global existence of solutions to reaction-diffusion systems on time-dependent domains. We extend global existence results for a class of reaction-diffusion systems on fixed domains to the same systems posed on spatially linear isotropically evolving domains. We demonstrate that the analysis is applicable to many systems that commonly arise in the theory of pattern formation. Our results give a mathematical justification to the widespread use of computer simulations of reaction-diffusion systems on evolving domains. We propose a finite element method to approximate the solutions to reaction-diffusion systems on time-dependent domains. We prove optimal convergence rates for the error in the method and we derive a computable error estimator that provides an upper bound for the error in the semidiscrete (space) scheme. We have implemented the method in the C programming language and we verify our theoretical results with benchmark computations. The method is a robust tool for the study of biological pattern formation, as it is applicable to domains with irregular geometries and nonuniform evolution. This versatility is illustrated with extensive computer simulations of reaction-diffusion systems on evolving domains. We observe varied pattern transitions induced by domain evolution, such as stripe to spot transitions, spotsplitting, spot-merging and spot-annihilation. We also illustrate the striking effects of spatially nonuniform domain evolution on the position, orientation and symmetry of patterns generated by reaction-diffusion systems. To improve the efficiency of the method, we have implemented a space-time adaptive algorithm where spatial adaptivity is driven by an error estimator and temporal adaptivity is driven by an error indicator. We illustrate with numerical simulations the dramatic improvements in accuracy and efficiency that are achieved via adaptivity. To demonstrate the applicability and generality of our methodology, we examine the process of parr mark pattern formation during the early development of the Amago trout. By assuming the existence of chemical concentrations residing on the surface of the Amago fish which react and diffuse during surface evolution, we model the pattern formation process with reactiondiffusion systems posed on evolving surfaces. An important generalisation of our study is the experimentally driven modelling of the fish's developing body surface. Our results add weight to the feasibility of reaction-diffusion system models of fish skin patterning, by illustrating that a reaction-diffusion system posed on an evolving surface generates transient patterns consistent with those experimentally observed on the developing Amago trout. Furthermore, we conclude that the surface evolution profile, the surface geometry and the curvature are key factors which play a pivotal role in pattern formation via reaction-diffusion systems.