The Monte Carlo radiation transport programm AMOS (Allgemeine Monte Carlo Simulation) has been developed at the TU Dresden since 1989. Originally, it was designed for neutron transport. But in the middle of the nineties, the redesign and optimisation for pure photon transport was carried out. When specialised tasks from radiobiology arose in 2001, the development of low energy electron transport was initiated which was later coupled with the existing photon kernel. The extension of applications on neutron fields required the re-implementation of neutron transport which was accomplished in 2012. Today, especially the complex coupling between the different kinds of radiation are worked on.
By using C++, all benefits of object-oriented programming can be employed. For the developer, this makes the code flexible and easily extendable with regard to the introduction of new particles or interactions. The user is provided with a programm which needs only limited input parameters to launch a full simulation.
Unlike other programm systems, AMOS has no pure group or point data structure. While the particle attributes (energy, direction) are written quasi continuous, the interaction parameters are provided as group data. The latter improves the access speed and, due to the large memory of modern computers, the physical data is still represented in detail.
The focus of AMOS are tasks from medical and technical applications of x-ray and nuclide radiation erasing within medical imaging and tumor therapy as well as witin radiation protection and detector physics. This restricts the relevant energy interval, in which the simulation of photons and electrons has to be carried out, to from 1 keV to few MeV. The lower limit results from the available interactino data. Below 1 keV, their uncertainty is such that the simulation of particles in this energy region cannot improve the simulation result.
Beside the precise simulation of the different particles, it is essential to be able to realise a most complete spectrum of possible geometrical structures. In AMOS it is possible to apply two different geometry models and even to combine them (see Geometry). Meanwhile, one has to bear in mind that the AMOS geometry kernel is not suface-related but instead uses a position-based algorithm to locate the particle and to draw samples for the interaction-free path. Using the concept of the minimal mean free path, in general the shortest mean free path of the world is used to draw a sample for a transport path length. Then, the ratio between minimal and actual mean free path yields the probability for a real interaction.
The advantage of this concept is, that no intersections between particle path and geometry surfaces have to be determined. Such calculations are generally very time consuming, especially because they have to be carried out all the time. The drawback of the minimal mean free path concept comes up when the geometry consists of (small) regions with high Z material (very short mean free path) and at the same time large section with low Z material as air, which itself has a large mean free path. Then, in the low-Z material large numbers of ... are processed which slows down the simulation extremely.
This negative feature can be compensated by the concept of world cuboids. Its purpose is to separate the world in main regions in which the mean free paths of the locally occuring materials do not differ in the way described above. These world cuboids then would have to be distinguished by a surface-based algorithm - but this routine would not have to be called alle the time but rather seldom. Thus, the benefit of both methods would be combined.