dufuz 0.0.1

Discrete numeric fuzzy sets in Python algorithms.

dufuz

Incorporating discrete numeric fuzzy sets in Python algorithms. These sets are more general than fuzzy numbers.

Dependencies: numpy, torch, matplotlib
Contact: Manios Krasanakis ([e-mail])
License: Apache 2

:rocket: Quickstart

First create a discrete environment for spawning and executing operations on numeric fuzzy sets. Provide a GPU torch device to the environment to parallelize execution. The device is used as one logical core.

import torch
from dufuz import DiscreteEnvironment
from dufuz import tnorm

env = DiscreteEnvironment(tnorm=tnorm.lukasiewicz,
                          tol=0.01, breadth=1,
                          device=torch.device('cuda:0'))

You can write algorithms involving normal Python operations. If-then-else statements that involve fuzzy comparisons take the form condition.choose(result if true, result if false) and fuzzy boolean arithmetics use the &,| operations.

As a demonstration. the following code implements the bubblesort algorithm for a list of fuzzy numbers.

def bubblesort(values):
    for i in range(len(values)):
        for j in range(i+1, len(values)):
            vali = values[i]
            valj = values[j]
            comparison = vali < valj
            values[i] = comparison.choose(vali, valj)
            values[j] = comparison.choose(valj, vali)

The list can be defined to hold triangle fuzzy numbers per:

values = list(range(8))
values = env.number(values)