QonGPU

Program from my bachelorthesis. Its main purpose is to simulate a time resolved 1D ionization with some python tools to analyze the data.

Method

The solver algorithms are described in my bachelorthesis, which I will link this project to as soon as I have the time to upload a complete corrected version of it.

Dependencies

1. CMake >= 3.0
2. Boost >= 1.5.6
3. GCC >= 4.9.2
4. HDF5/HDF5lite >= 1.86
5. Python 3: Matplotlib & Numpy for data analysis
6. CUDA comptible with GCC >= 4.9.2

Building

Since this is single-node HPC application only Linux is currently supported.

cd /location_to_QonGPU/
./configure.sh

The script will take care of all the build folders etc. The results are saved per default in a HDF5 file called res.h5 and is located in the build folder. The way the timesteps are saved is that one timesteps n creates 2 HDF5 datasets, one for the imaginary and one for the real part of the wave-function. These datasets will always have the name dsetNreal and dsetNimg, where N is the number of timesteps calculated. The stationary states as well as the energies of each stationary state are saved inside the HDF5 file build/sim1.h5. All the simulation parameters are set inside the main.cu file.