Particle-based Simulations for Hard and Soft Matter
Course material (slides and tutorials) can be found in the "Files" section! For material on Pasimodo and SPH, please visit http://www.sciencedirect.com/science/article/pii/S2210983814000170 or www.itm.uni-stuttgart.de.
If you have questions regarding the ESPResSo tutorials / presentations, please contact our mailing list firstname.lastname@example.org.
Particle based simulations have become a standard tool in all areas of computer simulations. While originating from molecular mechanics, where classical methods like Monte-Carlo or Molecular Dynamics are routinely used to study the static and dynamic properties of atomistic many-body systems, particle-based simulations are now even used as a very promising route to mesh-free methods for differential equations. Such methods promise better adaptiveness, but most importantly include crack formation in a natural way. The size and time scales spanned are enormous, ranging from a few nanometers and -seconds in molecular systems to meters and hours when simulating the dynamics of liquids like fuel in a tank truck. Yet, the underlying computational methods and algorithms are very similar.
This similarity makes it possible to use state of the art algorithms from very different areas in order to improve the efficiency of particles simulations and allow for simulations of larger systems and longer time scales. We exploit these synergies in the DFG funded collaborative research center SFB 716 at the University of Stuttgart, where engineers, natural scientists, and HPC computer scientists work together on pushing the limits of dynamical, particle-based simulations. The codes in this collaboration span the range from Molecular Dynamics (ITAP Molecular Dynamics, IMD, imd.itap.physik.uni-stuttgart.de) over mesoscopic particle simulations (Extensible Simulation Package for Research on Soft matter, ESPResSo, www.espressomd.org) to macroscopic simulations of fluids and rigid objects (PASIMODO. www.itm.uni-stuttgart.de/research/pasimodo/pasimodo_en.php).
While simulations of large particle systems become easier, they produce at the same time enormous amounts of data, that has to be analyzed. Specialized visualization software is required that can handle such large data sets. The visualization platform MegaMol (svn.vis.uni-stuttgart.de/trac/megamol), another project within the SFB 716, is capable of doing this besides other visual analytics tasks.
In this tutorial, the participants learn about the basic methods and algorithms of particle-based simulations, such as domain-decomposition, Verlet lists and integrators for Newton's equations of motion, up to more state-of-the-art techniques used in current simulations codes for soft and hard matter. This includes discrete element simulations used for granular matter, MD techniques, and methods to treat hydrodynamic interactions such as Lattice Boltzmann and SPH methods. In hands-on sessions, they will learn how to use real life simulation codes in order to study systems on all scales, namely all codes mentioned above. In particular, this also includes sessions on visualization of particles.
Since simulations are often designed and performed by PhD students, there is a continuous need for teaching the underlying techniques of computer simulations. Usually, students either write their own code, or at best get skilled in using a particular software. In this tutorial, the students will get a bird's eye view on a variety of techniques used in particle simulations,, so that they understand the underlying basic mechanisms. Our intention is to provide the students with the knowledge that should enable them to chose the appropriate method applicable to their problem, as well as to give them new insights beyond their own field of expertise. The school is unique in the sense that it tries to bridge the traditional boundaries of computational engineering, material science, and soft and hard condensed matter in the field of classical particle based simulation methods.
Another aim of this summer school is to give the developers of the software packages feedback on ongoing research, so that design decisions better match the needs of the community.
We plan to have 4 days of lectures in the morning and hands-on sessions in the afternoon. Occasionally, there will be two tracks of lectures and hands-on sessions. In particular IMD and ESPResSo are relatively similar packages, so that for some participants, it might be more interesting to get a deeper insight into one of the two packages. The last day, Friday, will be dedicated to presentations of successful applications of the software packages, and we would like to give participants the chance to present their own (planned) projects.
Axel Arnold (Institute for Computational Physics, University of Stuttgart) - Organiser & speaker
Peter Eberhard (Institute of Engineering and Computational Mechanics, University of Stuttgart) - Organiser
Johannes Roth (FMQ, Universität Stuttgart) - Organiser & speaker
Daniel Weiskopf (Visualization Research Center (VISUS), University of Stuttgart) - Organiser & speaker