Geometry is the description of the physical shape (volume) of an object within a Mantid instrument and the distances and rotations between them.
In Mantid we separate the Geometry of the shape of an object from the Geometry of it’s position. This is done primarily to save on memory usage but also to improve performance. Many operations within Mantid need to know where for example a detector is, but do not need to know what shape it is. By keeping the Geometry and Position separate we can keep the performance high. Also while in any one instrument we may have over 10,000 detector pixels all with different locations they will all have the same shape, therefore by keeping the shape separate we can greatly reduce the amount of memory required.
Both of the forms of Geometry share certain basic concepts. These are that co-ordinates are stored as 3D Vectors (the class in Mantid is called V3D), directions are described as similar unit 3D Vectors, and rotations are described using quaternions.
Category: Concepts