2. Getting started

2.1. Installation

Project assume to build under Visual Studio 2017 Community edition.

To build it just clone or download it from the GitHub and open solution.

Solution consists of four projects.

Table 2.1.1 List of projects
Project	Description
Geometry	Small internal use geometry base classes, like vectors, matrixes, planes, etc
MC	Core Monte Carlo library
MC.Tests	Some collection of tests. Valuable learning resource
MCSimulator	Complex executable, which uses full power of MC library and can be used standalone. Most practical tasks can be completely described by XML input files

2.2. Running simulations

MC library can be used to program custom tasks. Source code for MCSimulator fully demonstrates how to do that and can be considered as reference implementation. It is very like EGS4 workflow.

MCSimulator can be used as ready to use Monte Carlo simulator. It takes two XML input files to describe the task. First file describes simulation conditions. Second file describes simulation geometry.

Data/XML folder contains some input examples.

To run simulations you need to create a folder for the task and copy there MCSimulator.exe and input files. Then you need to have side by side to this folder data folder with PEGS4 input file (AcceleratorSimulator.pegs4dat for now).

Here are some examples to run simulations from the samples data:

MCSimulator S_C60_sphere.xml G_C60_sphere.xml
MCSimulator S_CK_6X.xml G_CyberKnife.xml
MCSimulator S_Ross.xml G_Ross.xml
MCSimulator S_Target.xml G_Target.xml

2.3. Output and visualization

MCSimulator produces two output files. Their names indicated in simulation conditions XML file. One file (.dat) contains numerical results of simulations. The other one is VRML graphical dump (.wrl). It can be viewed many VRML viewers. The recommended old style perfect one can be unzipped from this project ./Data/vrmlview.zip file.

2.4. Debugging

Debugging is a valuable tool to study code.

To run MCSimulator in debug build debug version. Click right mouse button over MCSimulator project in the Solution Explorer and select Set as StartUp Project.

MCSimulator by default uses all CPU cores. This will cause difficulties in stepping in the code due to jumping through different threads. You can switch to single thread by directly setting variable nThreads to one in the main function. See fragment of MCSimulator.cpp below.

int i;
//int nThreads = concurrency::GetProcessorCount();
int nThreads = 1;

The tricky part is providing program arguments and working directory. It should be done through the MCSimulator project properties. One possible scenario for the GitHub version of folders structures is shown in the next figure (Fig. 2.4.1). First time create folder MCwork where you will run debug simulation.

_images/DebugProperties.png — Fig. 2.4.1 Debug Properties.

Then start debugging.

Remember about the code execution performance. In the past the difference in speed between debug and release was of the order of 2. Modern compiler continuously improve and use hardware features including GPU even if programmer did not request that. Now the speed up can be up to an order or more depending on code.