spheres_from_coords

In cases where an external package is used to generate a sphere packing, this function can be used to convert the result to an ndimage.

import porespy as ps
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

[01:01:07] ERROR    PARDISO solver not installed, run `pip install pypardiso`. Otherwise,          _workspace.py:56
                    simulations will be slow. Apple M chips not supported.

df

The function is designed to accept the input coordinates and sphere radii as a pandas DataFrame, which is why this argument is called df…although as we’ll see this is somewhat flexible.

Let assume we have 4 spheres whose x, y, and z coordinates plus their respective radii are stored as column an Excel file with each row representing a sphere. This format can be directly read into a DataFrame using df = pd.read_excel or df = pd.read_csv, which is why the DataFrame is the format of choice.

Let’s enter a DataFrame by hand for this demonstration, then print it:

df = pd.DataFrame({'X': [10, 20, 40, 40], 'Y': [10, 30, 50, 10], 'Z': [0, 0, 0, 0], 'R': [5.0, 8.0, 17.5, 4.0]})
print(df)

    X   Y  Z     R
0  10  10  0   5.0
1  20  30  0   8.0
2  40  50  0  17.5
3  40  10  0   4.0

This array of spheres is only 2D as can be seen by all the ‘Z’ values being 0. The returned image will be in 2D.

im = ps.generators.spheres_from_coords(df)

The function also accepts numpy arrays with the same format as the above DataFrame:

arr = np.array(df)
print(arr)

[[10.  10.   0.   5. ]
 [20.  30.   0.   8. ]
 [40.  50.   0.  17.5]
 [40.  10.   0.   4. ]]

im = ps.generators.spheres_from_coords(arr)
plt.imshow(im);
../../../_images/80dd60ca985bd7416b5e835cf744d83ac2745e1506499a67d2a6ce296d0af861.png

It is also acceptable to pass a python dict:

dct = {'X': [10, 20, 40, 40], 'Y': [10, 30, 50, 10], 'Z': [0, 0, 0, 0], 'R': [5.0, 8.0, 17.5, 4.0]}
im = ps.generators.spheres_from_coords(dct)
plt.imshow(im);
../../../_images/80dd60ca985bd7416b5e835cf744d83ac2745e1506499a67d2a6ce296d0af861.png

It is also acceptable to leave the ‘X’ or ‘Y’ columns as 0’s. The function will ignore the column with all 0’s and return a 2D image.

dct = {'X': [10, 20, 40, 40], 'Y': [0, 0, 0, 0], 'Z': [10, 30, 50, 10], 'R': [5.0, 8.0, 17.5, 4.0]}
im = ps.generators.spheres_from_coords(dct)
plt.imshow(im);
../../../_images/80dd60ca985bd7416b5e835cf744d83ac2745e1506499a67d2a6ce296d0af861.png

Of course, it also works with 3D images:

dct = {'X': [10, 20, 40, 40], 'Y': [10, 30, 50, 10], 'Z': [4, 8, 12, 16], 'R': [5.0, 8.0, 17.5, 4.0]}
im = ps.generators.spheres_from_coords(dct)
plt.imshow(ps.visualization.sem(~im, axis=0).T);
../../../_images/d67caef3dbeb7bb094e1c92f3c471e59b4b9a78ecad2a4a2a4aa97bf46cce0c5.png

mode

The options are 'extended' which means the spheres are allowed to extend beyond the edge of the image, and contained meaning the sphere lie fully inside the image.

im1 = ps.generators.spheres_from_coords(df, mode='extended')
im2 = ps.generators.spheres_from_coords(df, mode='contained')
fig, ax = plt.subplots(1, 2)
ax[0].imshow(im1)
ax[1].imshow(im2);
../../../_images/5425505d4038613ecd0b1373a6f4b04deeffdfe6612a8ab6bf41415d1890caae.png

smooth

This controls the outer shape of the spheres.

im1 = ps.generators.spheres_from_coords(df, smooth=True)
im2 = ps.generators.spheres_from_coords(df, smooth=False)
fig, ax = plt.subplots(1, 2)
ax[0].imshow(im1)
ax[1].imshow(im2);
../../../_images/0109f457140d9153b25793ce99c4e3f8ad33cc876b9a45246b9c95d6d00b25e9.png