neumann-a / stochasticphysics

A C++ "Framework" to perform stochastic simulations. Currently mainly focused on the simulation of the magnetisation behavior of nanoparticles in a makrospin approximation

CI status

dev branch

Introduction

This project is aimed to perform stochastic simulations of physical processes. Currently the project is focused to perform those simulations for single domain magnetic particles.

---

Getting Started

Installation prerequisite

1. CMake
2. MATLAB (Linux: Needs to be on PATH for CMake to find it!)
3. C++ compiler e.g. clang(-cl) from LLVM (don't use MSVC if you care about performance; Also MSVC is not tested any longer).

Installation process

1. Get the code (executed from <basedir>)
   • git clone https://github.com/Neumann-A/StochasticPhysics.git <srcdir>
   • cd <srcdir>
   • setup git submodules via git submodule init and git submodule update
2. setup vcpkg.
   1. git clone https://github.com/microsoft/vcpkg.git (either in <srcdir> or <basedir>)
   2. (Optional) Bootstrap vcpkg. cd vcpkg
      • Windows: bootstrap.bat -disableMetrics -win64
      • Linux: ./bootstrap.sh -disableMetrics (optional: -useSystemBinaries)
   3. (Optional) Install dependencies via vcpkg
      • vcpkg install serar pcg-cpp eigen3 boost-random --overlay-ports=<srcdir>/vcpkg-ports
        • On Windows: use --triplet x64-windows or --triplet x64-windows-static
3. configure with cmake e.g. cmake -G "Ninja" -DCMAKE_BUILD_TYPE=Release -S <path-to-source> -B <path-to-build>. Available build options are: VCPKG_TARGET_TRIPLET, BUILD_TESTING, BUILD_BENCHMARKS
   • You might need to set -DVCPKG_TARGET_TRIPLET=<triplet> and -DVCPKG_HOST_TRIPLET=<triplet>
   • Other options (the default should be fine) Simulation_PCG, Simulation_Boost_Random,Simulation_WITH_GSL_Solvers (requires GSL), Simulation_WITH_ImplicitMidpoint
   • You can also look at the setup of the CI piplines for more info. These are located in the directory .gihub/worflows/ci.(linux|windows).yml
4. build/install with cmake e.g. cmake --build [<options>] -B <path-to-build>
   • --target StoPhysApp_MultiArch or StochasticPhysics
   • --config Release
   • You can also look at the setup of the CI piplines for more info. These are located in the directory .gihub/worflows/ci.(linux|windows).yml

Running test

• Same as installation process
• Tests are located in the Tests subfolder
• Not all tests are designed to be succesful.
• List of availale test targets:
  • Random_State_Init_Test
  • Implicit_solver_Test
  • Neel_Problem_Test
  • Neel_Spherical_Test
  • BrownAndNeel_Relaxation_Test (TODO: remove Relaxation)
  • NeelSpherical_BrownEuler_Relaxation_Test (TODO: remove Relaxation)

Running Simulations

Execute either StochasticPhysics or StoPhysApp_MultiArch it will run a dummy simulation generating a pair of *.ini configfiles. Those configfiles can be modified and passed to the applications via:

• StochasticPhysics -parfile:<SimulationSettings>
• StoPhysApp_MultiArch --parameter_file=<SimulationSettings>

The main difference between the two applications is that the former is a single architecture executable while the later is a multi architecture executable (AVX, AVX2, AVX512). The single architecture executable will probably be removed in the future.

Example for other config files can be found in the examples folder.

Running the executables on Linux might require extra setup of LD_LIBRARY_PATH. The Code is optimized and tested mainly on Windows systems.

Citing

Since there is currently no publication covering the internals of the code, citing is done by referencing the github url and the used git commit id. This is done

Publications/Poster/Talks using this Code

Universität zu Lübeck

• A. Neumann & T. M. Buzug
  Stochastic Simulations of Magnetic Particles: Comparison of Different Methods
  8th International Workshop on Magnetic Particle Imaging 213, Hamburg (Deutschland) (2018).

• A. Neumann, S. Draack, F. Ludwig & T. M. Buzug
  Parameter estimations of magnetic particles: A comparison between measurements and simulations
  9th International Workshop on Magnetic Particle Imaging 79, New York (USA) (2019)

• T. Klemme, T. M. Buzug & A. Neumann
  Exploring parameters of magnetic particles in 1D field excitation
  9th International Workshop on Magnetic Particle Imaging 189, New York (USA) (2019)

  T. Klemme, T. M. Buzug & A. Neumann
  Exploring Parameters of Magnetic Particles in 1D Field Excitation
  International Journal on Magnetic Particle Imaging, 6(2), 2004001, (2020)

• A. Neumann, & T. M. Buzug
  Simulations of magnetic particles with arbitrary anisotropies
  International Journal on Magnetic Particle Imaging, 6(2) Suppl. 1, 2009032, (2020)

Others

None yet

Regarding API/ABI stability

No gurantees on API or ABI stability are given.

stochasticphysics's Issues

Issue with saved excitation field amplitudes when saving single particle data

Saved excitation fields are not equal for different particles, when "Save_single_simulations" is enabled and there are less Number_of_simulators then Number_of_simulations

Field_Issue_cfg_files.zip