rafsaf / tw-complex

Algorithms for TW

TW Complex

Repo with algorithms to divide ally villages into front and back in TW.

Underneath it is a problem of dividing a set of 2D points A according to the min_radius and max_radius distances from a set of other 2D points B, which can be solved most simply by counting the distances from each point in the first set A to all points in the second set B one by one.

• Instalation
• Basic usage
• Examples
• Running locally

Instalation

pip install tw_complex

Basic usage

from tw_complex import CDistBrute
import numpy as np

# The code for Example 1 below

points1 = np.random.rand(10000, 2) + [2, 0]
points2 = np.random.rand(15000, 2)
min_radius = 1.4
max_radius = 2

precise_front, precise_back = CDistBrute(
    ally_villages=points1,
    enemy_villages=points2,
    min_radius=min_radius,
    max_radius=max_radius,
).result()

Examples (before -> after)

Example 1

Ally: 10000 points
Enemy: 15000 points
min_radius: 1.4
max_radius: 2

Example 2

Ally: 2500 points
Enemy: 6000 points
min_radius: 4
max_radius: 10

Example 3

Ally: 20000 points
Enemy: 20000 points
min_radius: 20
max_radius: 60

Example 4

Ally: 20000 points
Enemy: 20000 points
min_radius: 10
max_radius: 120

Running locally

You will need to have poetry installed.

git clone https://github.com/rafsaf/tw-complex.git
cd tw-complex
poetry install

Code lives in tw-complex folder, and you may also test algorithms running in main folder

# In main folder
# eg. ~/Desktop/tw-complex

pytest

For CDistAndKNN it looks like

# tests/test_cdist.py

from tw_complex import CDistAndKNN
import tests.utils as utils


def test_CDistAndKNN():
    utils.run_all_tests(CDistAndKNN, "CDistAndKNN", _precision=0.8, draw=True)

# Go with `draw=False` if you do not want to use pyplot to show diagrams

Some hardcoded tests are located in tests/utils.py, it uses brute force for calculating exact result, then compare it to given algorithm using basic maths. You can even compare it to brute force itself (eg. using diffrent _precision). For new test there should be another file in tests/test_name_of_file_in_tw_complex_folder.py with pretty much the same content as above.