Parallel algorithms for environmental modeling¶
Algorithms for environmental modeling are at the heart of any raster-based environment model. The environmental modeller combines these core model building blocks to build a unique model. There are many different environmental modelling algorithms, some of which are also found in geographic information systems (GIS).
Until around 2005, CPU cores found in computers doubled in clock speed about every two years. Environmental modellers who wanted to use more complex modelling rules and/or larger data sets, just had to buy a new computer to decrease the increased model run times. That is not the case anymore and so model run times keep increasing with added model complexity and data.
Because CPU cores are not getting much faster anymore, hardware vendors have been adding additional CPU cores to their CPU’s. One obvious way to solve the issue of increasing model run times is to make models use the multiple CPU cores. This requires a reimplementation of the above mentioned environmental modelling algorithms.
This project is about parallizing one or more environmental modelling algorithms. Some of these algorithms are very easy to parallize, and some are not. In this project you will look into parallizing one or more algorithms from the latter category. You will design one or more approaches to parallize the algorithm and, depending on your interest and background, test these approaches by implementing them.
This work is highly relevant, because the results may be used in a new implementation of our own library of modelling algorithms. Faster algorithms will have obvious benefits for the modellers and you can make a very concrete contribution to this.
Supervision: You will be supervised by a team of experienced modellers and software engineers. They will provide you with a description of the sequential version of each algorithm and help you getting up to speed quickly. This team will at least consist of Dr. Derek Karssenberg and Drs. Kor de Jong.
Number of students: 1-3
Prerequisite: preferably courses in spatio-temporal modelling, geoinformatics, computer science
Program/track: any track if you have the required prerequisite