University of Warwick
Neurons in the brain are highly complex structures which are bombarded by randomly arriving synaptic inputs. Indeed, most of the time, neurons in the cortex fire due to random fluctuations and are in the “subthreshold regime”. Understanding how neurons behave in this noisy environment is important to understanding computation in the brain. My poster gives an overview of how I have used Julia for my theoretical models, numerical simulations and data analysis.
Neurons are highly elongated with differences in synapses along their length, and I am interested in how the spatial properties of neurons affect their firing behaviour. Therefore, it is necessary to numerically solve stochastic differential equations in both space and time. Julia allowed me to create and run these simulations with little difficulty. Since I am interested in numerous random trials to calculate the average firing rate from a population of neurons, the relative ease of simple parallelisation in Julia has been very helpful to me. The profiling of simulations has also been very useful for finding inefficiencies and making my code run faster.
A wide range of different neuronal parameters are typically explored during simulations, and it is therefore useful to save these parameters of a simulation in an easy to read manner with the simulation data itself. To this end, the HDF5 derived JLD file format has been invaluable. It has also been useful in restructuring real-world data obtained from experiments, in which I used various techniques to infer the parameters.
I am a second year PhD student at the University of Warwick currently doing research in computational neuroscience. Previously, I have done some research in thin film transistors and resistive memory. I have been using Julia for about 3 years now, when I used it to find all possible solutions to a puzzle.
