Repository files navigation

Fuzzy sets and fuzzy logic theory implementations.

import FuzzyKitenumFunding{case adequate, marginal, inadequate}enumStaffing{case small, large}enumRisk{case low, normal, high}letfunding:SimpleLinguisticVariable<Funding,AnyFuzzySet>=[.inadequate:.init(membershipFunction:.leftOpen(slopeStart:15, slopeEnd:35)),.marginal:.init(membershipFunction:.triangular(minimum:21, peak:41, maximum:61)),.adequate:.init(membershipFunction:.rightOpen(slopeStart:55, slopeEnd:75)),]letstaffing:SimpleLinguisticVariable<Staffing,AnyFuzzySet>=[.small:.init(membershipFunction:.leftOpen(slopeStart:29, slopeEnd:69)),.large:.init(membershipFunction:.rightOpen(slopeStart:37, slopeEnd:77)),]letrisk:SimpleLinguisticVariable<Risk,AnyFuzzySet>=[.low:.init(membershipFunction:.leftOpen(slopeStart:20, slopeEnd:40)),.normal:.init(membershipFunction:.triangular(minimum:20, peak:50, maximum:80)),.high:.init(membershipFunction:.rightOpen(slopeStart:60, slopeEnd:80)),]letØ= AnyFuzzySet<Double>.emptyletruleBase=FuzzyRuleBase{    funding.is(.adequate) || staffing.is(.small)--> risk.is(.low)    funding.is(.marginal) && staffing.is(.large)--> risk.is(.normal)    funding.is(.inadequate) || Ø--> risk.is(.high)}letflc=FuzzyLogicController(rules: ruleBase, settings:.init(implication:.mamdani))flc.consequenceGrade(for:50, usingSingletonFact:(8.8,42))  // result is 0.675

Using the Swift Package Manager, don't forget to add the package as a dependency to yourPackage.swift file:

dependencies: [+   .package(url: "https://github.com/allexks/FuzzyKit", .upToNextMajor(from: "0.1.0")),],

To be able to use everything from this package, you can import everything at once using this helper module:

.target(    name: "...",    dependencies: [+       .product(name: "FuzzyKit", package: "FuzzyKit"),    ]),

import FuzzyKit

Or alternatively, import only the specific modules needed:

.target(    name: "...",    dependencies: [+       .product(name: "FuzzySets", package: "FuzzyKit"),+       .product(name: "FuzzyNumbers", package: "FuzzyKit"),+       .product(name: "FuzzyRelations", package: "FuzzyKit"),+       .product(name: "FuzzyLogic", package: "FuzzyKit"),    ]),

import FuzzySetsimport FuzzyNumbersimport FuzzyRelationsimport FuzzyLogic

API Reference automatically collected withjazzy is publishedhere with each new release.

This abstraction requires a fuzzy set to provide agrade(forElement:) method which accepts a parameter of anassociatedtype Universe and returns its membershipGrade in the set. There are 3 provided concrete implementations in this module:

1. struct AnyFuzzySet - allows type erasure. It only stores aMembershipFunction and has non-mutable methods.

2. struct IterableFuzzySet - stores aMembershipFunction as well as aSequence of elements of the associated typeUniverse. ImplementsSequence so that it can easily be iterated over them. The elements of the iteration over anIterableFuzzySet arestructs containinggrade andelement properties. It has non-mutable methods only. Includessupport,core andheight computed properties.

3. struct DiscreteMutableFuzzySet - it is "discrete" because it doesn't stores aMembershipFunction but instead keeps its elements and their grade in aDictionary, and it is "mutable" because it contains mutable equivalents of all other methods that operate over the set (includingsubscript). A default value of0 is returned for the grade of an element that is not in the dictionary (a different default value can be provided as well). Includessupport,core andheight computed properties. A bonus feature is its debug print using Zadeh's notation.

Convertions between the 3 types are easy and possible using theeraseToAnyFuzzySet,makeIterable andmakeDiscreteMutable methods defined on them.

All 3 concrete types implement it. It requires the following methods that operate on fuzzy sets:

• alphaCut(_:alpha:)

• complement(method:)

• intersection(_:method:)

• union(_:method:)

• difference(_:method:)

• symmetricDifference(_:method:)

• power(_:n:)

• appliedCustomFunction(_:function:)

MIT