Example

The example libyt/example/amr-example demonstrates how to implement libyt in an adaptive mesh refinement (AMR) grid simulation. The example has a set of pre-calculated data. It assigns the data to MPI processes randomly to simulate data distributed on different processes. (Though in real world, data won’t distribute randomly.)

The steps related to implementation of libyt is commented like this:

// ==========================================
// libyt: 0. include libyt header
// ==========================================

// ==========================================
// libyt: [Optional] ...
// ==========================================

The example initializes libyt, loads data to libyt in every simulation time step in the iterative process and uses yt to do in situ Python analysis, and finalizes it before terminating the simulation.

Building and Running the Example

Using CMake

1. Follow Install.

2. Enter example folder. Assume we build the project under libyt/build:

   cd libyt                        # go to libyt project root folder
   cd build/example/amr-example    # go to example folder
   

Using Make

1. Follow Install.

2. Go to libyt/example/amr-example folder.

   cd libyt/example/amr-example
   

3. Compile the code:

   • Serial Mode (using GCC):

     make OPTIONS=-DSERIAL_MODE LIBYT_PATH=<libyt-install-prefix>
     

   • Parallel Mode (using MPI):

     make MPI_PATH=<mpi-install-prefix> LIBYT_PATH=<libyt-install-prefix> 
     

     Make sure you are using the same MPI to compile libyt and the example.

Running the Example

1. Run the example:

   • Serial Mode (using GCC):

     ./example
     

   • Parallel Mode (using MPI):

     OMPI_MCA_osc=sm,pt2pt mpirun -np 2 ./example
     

     OMPI_MCA_osc=sm,pt2pt is for using one-sided MPI communications.

2. The output results we get from the first step:

   • Density projection along z-axis, FigName000000000_Projection_z_density.png:

     ProjectionPlot

     

     Python

     def yt_inline_ProjectionPlot( fields ):
         ds = yt_libyt.libytDataset()
         prjz = yt.ProjectionPlot(ds, 'z', fields)
     
         if yt.is_root():
             prjz.save()
     

   • Density profile with respect to x coordinate, FigName000000000_1d-Profile_x_density.png:

     ProfilePlot

     

     Python

     def yt_inline_ProfilePlot():
         ds = yt_libyt.libytDataset()
         profile = yt.ProfilePlot(ds, "x", ["density"])
     
         if yt.is_root():
             profile.save()
     

   • Slice plot of the reciprocal of the density field, FigName000000000_Slice_z_InvDens.png:

     SlicePlot

     

     Python

     def yt_derived_field_demo():
         ds = yt_libyt.libytDataset()
         slc = yt.SlicePlot(ds, "z", ("gamer", "InvDens"))
     
         if yt.is_root():
             slc.save()
     

   • Particle plot of the level of a grid. Each grid has one particle with its position set at the center of the grid with value equals to grid level. The result is FigName000000000_Particle_z_Level.png. We can only see some tiny dots in the figure, since there is only one particle in each grid. (The plot tries to demonstrate the particle functionality.)

     ParticlePlot

     

     Python

     def yt_inline_ParticlePlot():
         ds = yt_libyt.libytDataset()
         par = yt.ParticlePlot(ds, "particle_position_x", "particle_position_y", "Level", center = 'c')
     
         if yt.is_root():
             par.save()
     

   After loading simulation data using libyt API, we call Python functions defined in inline_script.py using yt_run_Function/yt_run_FunctionArguments and use yt to analyze the data and plot figures. For how to use yt see:

   • Using yt

---

What’s Next

Change Python Script Name

The example uses inline_script.py for in situ Python analysis. To change the Python script name, simpy change param_libyt.script and do not include file extension .py in example.cpp.

// file: example/example.cpp
yt_param_libyt param_libyt;
param_libyt.verbose = YT_VERBOSE_INFO;   // libyt log level
param_libyt.script = "inline_script";    // inline python script, excluding ".py"
param_libyt.check_data = false;          // check passed in data or not

To know more about writing inline Python script, you can refer to Using Python for In Situ Analysis in libyt.

Other Python Entry Points for Interactive Data Analysis

To try interactive Python prompt:

1. Go to amr-example build folder.

2. Create LIBYT_STOP. The file is signal for libyt to decide whether to launch the file-based Python prompt.

3. Run the program. See more in Interactive Python Prompt.

To try file-based Python prompt:

1. Go to amr-example build folder.

2. Create LIBYT_RELOAD. The file is signal for libyt to decide whether to launch the file-based Python prompt.

3. Run the program. See more in Reload Script.

To try Jupyter Notebook / JupyterLab access:

1. Go to amr-example build folder.

2. Create LIBYT_JUPYTER. The file is signal for libyt to decide whether to launch Jupyter kernel used for Jupyter Notebook / JupyterLab access.

3. Run the program. See more in Jupyter Notebook Access.