L’IPHC | Ressources techniques » Electronique » Equipe technique "Systèmes de Mesure et d’Acquisition" » MGS » Frequently Asked Questions
Dernière mise à jour : mercredi 5 avril 2006, par
For representing a volume in 3D, we have basically three options :
fixed grid, where each point of a 3D matrix is inside or outside the volume, the grid is square.
adaptative grid, where you can refine the grid at the border of the volume to have a close match to the simulated volume. The grid is square but the size of the squares can change.
mesh grid, where the grid is triangular, the approximation of the simulated geometry is better.
Unfortunately the better the method, the harder it is to build a volume and to apply the simulation. MGS strenghts lie in its versatility : any type of detector can be modelled in a very short time. You can check in the developer documentation, how simple it is to describe a shape (see planar.m or agata__angle.m).
As for the simulation of Poisson theorem using SOR (Successive Relaxation Method), etc., it seems that no example working for any shape in 3D is available (its is 2D or for a specific shape).
Matlab is the best tool regarding numerical simulation. It is particularly suited for matrices computation. Similar free tools like Scilab and Octave appear to have limitations with very large matrices, so we will not translate MGS into any of these langages.
MGS will not be translated in Java, C# or C++,... :
Matlab has highly optimized libraries for matrices computations (written in C and in assembly), so a full homemade version is foreseen to be slower
Matlab provides very easy to use visualization commands
Developing with Matlab allows us to focus on the simulation mathematics and not on the language (hence it is faster to develop and test new ideas)