Modelization and Computational Science Lab

Houston and Rennes 21st July 2008

Dear Virtual Prairies Explorers,

          We would like to report on our progress with the Virtual Prairie Project that you have kindly accepted to run on your system. We have more or less finished the first phase of the project (modeling of the individual plant) and have been browsing the parameter space of our individual based model of clonal plant growth. The first campaign of simulations ran from April to May, and brought us some usual information on a better heuristic to browse the parameter space of our model.

          The second campaign of simulations just ended last week: we managed to run 22 million simulations thanks to you and your 771 fellows!

          We have submitted our second conference paper to an IEEE conference that summarizes our findings. Basically we show that for achieving the best growth, there is no single optimum in plants but rather multiple strategies. This would explain why some natural prairies are composed of species of plants of different growing patterns. Because we deal with very large samples of data, classification techniques and statistical analyses are not easy to manipulate successfully, and this is currently our focus. Nevertheless this work would not have been possible without you.

          The publications corresponding to our current work will be available on the web site of the project once they get reviewed and hopefully accepted. This project has got the attention of CNRS, Cemagref and ANR that are national French research institutions and have decided to provide some funding to support the research.

          In the second phase of the project we will simulate a prairie with thousands of individual plants. We are now finishing the verification of the code. The simulations will run for a season with hundred of clonal plants: we expect to see interesting emerging strategies in our simulation.

          Malek Smaoui has been working very hard to get that code prepared for our third campaign of simulations that should start anytime now: we will need probably 10 000 volunteers!

          Aurelie Celant from CPE Lyon has joined our group for the summer and is preparing some interactive computer tools to analyze on the fly the results of volunteer computing. The web server of the project will be organized to illustrate these improvements by early September.

          Marie-Lise Benot from University of Rennes has started another campaign of experiments in garden studying the growth of plants in real. This work will help to evaluate if the data coming from the simulations meet real plant growth.

          This experience with BOINC continues to be for us quite exciting and stressful. We are still in the learning process and our simulation has already been far beyond the size of any standard simulation in Ecology. We would like to thank David Anderson, from the Berkeley Space Lab and PI of the BOINC project, for his invaluable help.

          Well that's all for now! We'd like to thank you all, and we will do our best to take advantage of these BOINC simulations. Next log should come in about a month...

                                                                                            Be the force of the Silicon with you!

                                                                                            Marc Garbey and Cendrine Mony

VIP Project