| Popis: |
This thesis deals with dynamical systems, numerical software for the continuation study of dynamical systems, and some important neurobiological applications. First there are two introductory chapters, in which a background is given in dynamical systems and neuroscience. We elucidate what the problems are with some existing classifications of neural models, and suggest an improved version. We introduce the Phase Response Curve (PRC), which is a curve that describes the effect of an input on a periodic orbit. We derive an efficient method to compute this PRC. The extended functionalities of MatCont, a software package for the study of dynamical systems and their bifurcations, are explained: the user can compute the PRC of a limit cycle and its derivative, he can detect and continue homoclinic bifurcations, initiate these curves from different bifurcations and detect many codim 2 bifurcations on these curves. The speed of the software was improved by introducing C-code among the matlab-routines. We have for the first time made a complete bifurcation diagram of the Morris-Lecar neural model. We show that PRCs can be used to determine the synchronizing and/or phase-locking abilities of neural networks, and how the connection delay plays a role in this, and demonstrate some phenomena to do with PRCs and bifurcations. In collaboration with biologists at the University of Bristol, we have built detailed models of the neurons in the spinal cord of the hatchling Xenopus laevis. The biological background and the equations and parameters for the models of individual neurons and synapses are listed elaborately. These models are used to construct biologically realistic networks of neurons. The first network was used to simulate the swimming behaviour of the tadpole and to show that to disregard some important differences in the models for different neurons, will result in breakdown of the good network output. Then we have used the individual models to study a hypothesis regarding synaptogenesis, which states that the specificity in connection between neurons could be purely based on the anatomical organization of the neurons, instead of the ability of growing synapses to make a distinction between the different neurons. |
