Add parallel solve for sample_flow_at_points

[misi9170]

As identified by @Bartdoekemeijer in #1009 , the ParFlorisModel did not have a parallelized version of sample_flow_at_points(). Instead, this method was simply inherited from FlorisModel, so ran in serial. This works, but does not take advantage of parallelization.

This PR address the issue by adding a sample_flow_at_points() method to ParFlorisModel. This can be checked by using the script that @Bartdoekemeijer provided in #1009:

import numpy as np
from floris import ParFlorisModel

if __name__ == "__main__":
    # Load the parallel floris model and create a 200-turbine farm
    interface = None #"multiprocessing", "pathos", "concurrent"
    fmodel = ParFlorisModel("../inputs/gch.yaml", max_workers=8, print_timings=True, interface=interface)
    n_turbs = 200 # Or try something smaller 
    fmodel.set(
        layout_x=5 * 126.0 * np.arange(n_turbs),
        layout_y=np.zeros(n_turbs),
    )

    # Cover a large grid of findices
    wd_grid, ws_grid = np.meshgrid(
        np.arange(0.0, 360.0, 3.0),
        np.arange(1.0, 30.01, 1.0)
    )
    fmodel.set(
        wind_directions=wd_grid.flatten(),
        wind_speeds=ws_grid.flatten(),
        turbulence_intensities=0.06 * np.ones_like(ws_grid.flatten()),
    )

    # Sample wind speeds at a couple of points in the flow
    wind_speeds_at_points = fmodel.sample_flow_at_points(
        x=[500.0, 750.0, 1000.0, 1250.0, 1500.0],
        y=[  0.0,   0.0,    0.0,    0.0,    0.0],
        z=[ 90.0,  90.0,   90.0,   90.0,   90.0]
    )
    print(wind_speeds_at_points)

Additional supporting information

In adding the sample_flow_at_points() method, I realized the main run() method had a lot of unnecessarily repeated code, so I've cleaned that up a bit (and added a private method for printing timing output).

Test results, if applicable

• Existing tests (for main run() functionality) pass
• New test added for sample_flow_at_points method
• No change to output of examples/009_parallel_models.py

[paulf81]

Hi @misi9170 I reviewed the code and it looks good to me. The new test is a good reassurance it's working properly on all tests passed on my end

[Bartdoekemeijer]

Awesome, thank you Misha!!