Despace tutorial and examples

Installation

!pip install despace --upgrade

Usage

from despace import SortND
import numpy as np  # for generating random data

Use cases

SortND is a python class for sorting N-D array-like data and visualize the sorted data if the number of dimensions is <=3.

1. Sort 2D coordinates

data = np.random.rand(100, 2)  # Let's generate some random data

We can either initiate the SortND class with data and then use the sort method, or initiate the SorND class withtou providing data, but then directly call with data.

s = SortND(data)
s.sort()
s.plot(show_plot=True, plot_arrow=True)

True

The data is colored from blue to red as the index goes higher. The arrows coonect two successive data points.

If we don’t want the arrows that show the relative indices in the sorted data, we can set plot_arrow to False

s.plot(show_plot=True, plot_arrow=False)

True

Let’s look at the sorted data:

s.sorted_coords[0:20,:]

array([[1.20941257e-01, 5.91526257e-02],
       [3.25718514e-02, 9.74565620e-02],
       [1.62132588e-01, 1.35027698e-01],
       [3.25022682e-01, 7.93149650e-03],
       [1.75661503e-01, 7.94614740e-02],
       [2.07350441e-01, 1.59170200e-01],
       [2.38184967e-01, 2.97525569e-01],
       [7.02506519e-03, 3.43268630e-01],
       [1.73103492e-02, 3.49722952e-01],
       [3.19409889e-01, 2.07180786e-01],
       [2.70387646e-01, 4.61570909e-01],
       [2.73479793e-01, 4.65325404e-01],
       [3.74223225e-01, 1.43901996e-02],
       [3.44131364e-01, 1.17916049e-01],
       [3.61386002e-01, 6.23778659e-02],
       [4.16027140e-01, 1.15405545e-06],
       [4.42175562e-01, 2.49551881e-02],
       [4.55579364e-01, 1.82072067e-01],
       [4.17344346e-01, 2.39805423e-01],
       [3.46449546e-01, 4.55420742e-01]])

2. Sort 3D coordinates

data = np.random.rand(1000, 3)
s(data)
s.plot(show_plot=True)

True

3. Space filling curves

Morton space filling curve

data = [[i, j] for i in range(8) for j in range(8)]
s.sort(data)
s.plot(show_plot=True, plot_arrow=True)

True

If one dimension length is odd:

data = [[i, j] for i in range(8) for j in range(9)]
s.sort(data)
s.plot(show_plot=True, plot_arrow=True)

True

The path can cross itself in the ‘odd’ dimension.

If both dimension lengths are odd:

data = [[i, j] for i in range(9) for j in range(9)]
s.sort(data)
s.plot(show_plot=True, plot_arrow=True)

True

Since the order is random if two points have the same value in either of their components, we see move data cross-over in both directions!

Hilbert space filling curve (2D)

If we set sort_type='Hilbert':

data = [[i, j] for i in range(16) for j in range(16)]
s.sort(data, sort_type='Hilbert')
s.plot(show_plot=True, plot_arrow=True)

True