BMI Visualize!

Visualize the output of the BMI-enabled Diffusions model as a function of time.

Setup

Check that these packages are installed on your system:

pandas
matplotlib
bokeh
xarray
holoviews

Some imports to start, including holoviews:



In [1]:

    
import os
import numpy as np
import holoviews as hv
hv.extension('bokeh', 'matplotlib')

Load and run the Diffusion model through its BMI

Start by importing the BmiDiffusion class from the bmi-live package and creating an instance:



In [2]:

    
from bmi_live import BmiDiffusion
x = BmiDiffusion()

Locate a configuration file for this example and use it to initialize the model:



In [3]:

    
from bmi_live import data_directory
cfg_file = os.path.join(data_directory, 'visualize_diffusion.yaml')
x.initialize(cfg_file)

Get a reference to the temperature variable exposed by the model:



In [4]:

    
temp = x.get_value_ptr('plate_surface__temperature')

Add an impulse to the initial temperature field:



In [5]:

    
temp[200] = 100.0

Because temp is a reference, there's no need to call set_value to update its value.

Get the id of the grid on which the temperature variable is located:



In [6]:

    
grid_id = x.get_var_grid('plate_surface__temperature')
grid_id









    Out[6]:





0

Use this grid id to get the shape of the grid:



In [7]:

    
grid_shape = np.ndarray(x.get_grid_rank(grid_id), int)
x.get_grid_shape(grid_id, grid_shape)
grid_shape









    Out[7]:





array([20, 30])

Set the number of time steps that we'll advance the model:



In [8]:

    
n_steps = 35

Make a variable to hold the values of the temperature field over time:



In [9]:

    
temperature = np.zeros((*grid_shape, n_steps))

What is the shape of this variable?



In [10]:

    
temperature.shape









    Out[10]:





(20, 30, 35)

Load the initial perturbed temperature field into the temperature variable:



In [11]:

    
temperature[:,:,0] = temp.reshape(grid_shape)

Advance the model through time, adding to the temperature array at each step:



In [12]:

    
for t in range(n_steps):
    x.update()
    temperature[:,:,t] = temp.reshape(grid_shape)

Note that because temp is a reference, it's automatically updated on each time step.

Visualize output from the Diffusion model

Load the temperature array into a holoviews dataset.



In [13]:

    
ds = hv.Dataset((np.arange(n_steps), np.arange(grid_shape[1]), np.arange(grid_shape[0]),
                 temperature), ['Time','x', 'y'], 'Temperature')

Examine the dataset:



In [14]:

    
type(ds.data), list(ds.data.keys())









    Out[14]:





(dict, ['Time', 'x', 'y', 'Temperature'])

Visualize the dataset as an image with a slider to step through time:



In [15]:

    
%opts Image (cmap='Reds')[colorbar=True]
%output size=150
ds.to(hv.Image, ['x', 'y'], dynamic=True)