[geometry] periodic boundary conditions
Hi, I am starting to investigate the boost.geometry library. One of the applications I have already for it is in the context of particle simulations. I still have not much idea of the architecture of the library but it seem very powerful. So for example I was wondering if geometry has facilities for distances and points living in 3D periodic boundary conditions. Topologically it is a torus but with euclidean distance. If somebody could point where to look I will appreciate it. Thank you, Alfredo
hi Alfredo,
I am starting to investigate the boost.geometry library. One of the applications I have already for it is in the context of particle simulations.
Thanks for your interest.
I still have not much idea of the architecture of the library but it seem very powerful. So for example I was wondering if geometry has facilities for distances and points living in 3D periodic boundary conditions. Topologically it is a torus but with euclidean distance.
Boost.Geometry supports euclidean distances between points or points-lines in 3D. I've no experience in particle simulations, so don't know what to expect, but I'm curious to your findings.
If somebody could point where to look I will appreciate it.
There are several places to look in more detail. The architecture of the library is described in two papers written for BoostCon'09 and BoostCon'10, they can be downloaded here: http://trac.osgeo.org/ggl/wiki/BoostCon On that website there are also some other links a.o. to the Doxygen documentation. Regards, Barend
On Jul 23, 10:32 am, Barend Gehrels
Boost.Geometry supports euclidean distances between points or points-lines in 3D. I've no experience in particle simulations, so don't know what to expect, but I'm curious to your findings.
just to complete the idea. in particle simulations more often that not we use periodic boundary conditions. the trick is that the particles are free to move in 3D (this helps to simulate difussion in liquids), but the interactions (distances) are computed as if all the particles are still in the original box. so, the keys are euclidean distances and torus topology.
The architecture of the library is described in two papers written for BoostCon'09 and BoostCon'10, they can be downloaded here:http://trac.osgeo.org/ggl/wiki/BoostCon
Thank you for the references. Alfredo
On 24/07/2010 4:20 AM, alfC wrote:
just to complete the idea. in particle simulations more often that not we use periodic boundary conditions. the trick is that the particles are free to move in 3D (this helps to simulate difussion in liquids), but the interactions (distances) are computed as if all the particles are still in the original box.
so, the keys are euclidean distances and torus topology.
No it doesn't. Boost.Geometry's design is based around, providing your own data types, the composites of those such as points etc (if they're not already defined), and any practical operations you may need in your computations such as distance measures, areas etc (If they're not already defined). PCB'ed geometries are very difficult to get right in a generic sense.
participants (3)
-
alfC -
Arash Partow -
Barend Gehrels