From 21e22e59ad16f1d54994b63a371a4af327db359d Mon Sep 17 00:00:00 2001 From: Antoine Cyril David Hoffmann <antoine.hoffmann@epfl.ch> Date: Fri, 23 Dec 2022 21:32:44 +0000 Subject: [PATCH] Update README.md --- README.md | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-) diff --git a/README.md b/README.md index a828e1cd..8a1241b3 100644 --- a/README.md +++ b/README.md @@ -23,14 +23,17 @@ along with this program. If not, see <https://www.gnu.org/licenses/>. # How to compile and run GYACOMO -1. Be sure to have correct library paths in local/dirs.inc for the different libraries, see INSTALATION.txt for a tutorial to install the required libraries. -2. Compile from /gyacomo using make, the binary will be located in /gyacomo/bin (you can also compile a debug version using make dbg) -4. The fort.90 file should contain the parameters for a typical CBC to test the compilation. One can run it by calling the executable /bin/gyacomo in the directory where the fort.90 is located. -5. It is possible to run GYACOMO in parallel using MPI: mpirun -np N ./bin/gyacomo Np Ny Nz where N=Np x Ny x Nz is the number of processes and Np Ny Nz are the parallel dimensions in Hermite polynomials, binormal direction and parallel direction, respectively -6. You can stop your simulation without corrupting the output file by creating a blank file call "mystop", using e.g. "touch mystop" in the directory where the simulation is running. (the file will be removed once read) -7. It is also possible to put simulations ID in order to chain them. The parameter Job2load allows you to tell which output file should be read in order to restart a simulation. E.g. I run a first simulation with job2load = -1, it creates a outputs_00.h5 then I create a new fort.90 which I call fort_01.90 where job2load = 0. I run then GYACOMO, indicating that I want it to read the fort_00.90 using 0 as a last argument, i.e. "./gyacomo 0" or "mpirun -np N ./gyacomo Np Ny Nz 0", which will start from the latest 5D state saved in outputs_00.h5. A new output file has also been created, output_01.h5. -8. You can obtain various plots and gifs using gyacomo/wk/gyacomo_analysis.m once the simulation is done. The directory where the results are located must be given in the scripts. It is not a function (yet...) -// Comment : For some collision operators (Sugama and Full Coulomb) you have to run COSOlver from B.J.Frei first in order to generate the required matrices in gyacomo/iCa folder. // +To compile and run GYACOMO, follow these steps: + + +1. Make sure the correct library paths are set in local/dirs.inc. Refer to INSTALATION.txt for instructions on installing the required libraries. +2. Go to the /gyacomo directory and run make to compile the code. The resulting binary will be located in /gyacomo/bin. You can also compile a debug version by running make dbg. +3. The file fort.90 should contain the parameters for a typical CBC. To test the compilation, navigate to the directory where fort.90 is located and run the executable /bin/gyacomo. +4. GYACOMO can be run in parallel using MPI by running mpirun -np N ./bin/gyacomo Np Ny Nz, where N = Np x Ny x Nz is the number of processes and Np Ny Nz are the parallel dimensions in Hermite polynomials, binormal direction, and parallel direction, respectively. +5. To stop the simulation without corrupting the output file, create a blank file called "mystop" using touch mystop in the directory where the simulation is running. The file will be removed once it is read. +6. It is possible to chain simulations by using the parameter "Job2load" in the fort.90 file. For example, to restart a simulation from the latest 5D state saved in outputs_00.h5, create a new fort.90 file called fort_01.90 and set "Job2load" to 0. Then run GYACOMO with the command ./gyacomo 0 or mpirun -np N ./gyacomo Np Ny Nz 0. This will create a new output file called output_01.h5. +7. To generate plots and gifs using the simulation results, use the script gyacomo/wk/gyacomo_analysis.m and specify the directory where the results are located. Note that this script is not currently a function. +Note: For some collision operators (Sugama and Full Coulomb), you will need to run COSOlver from B.J.Frei to generate the required matrices in the gyacomo/iCa folder before running GYACOMO. -- GitLab