You have 3 free guides left 😟
Unlock your guides
You have 3 free guides left 😟
Unlock your guides

5.4 Galois connections and closure operators

4 min readaugust 7, 2024

Galois connections link two ordered sets through paired functions, forming a powerful tool in lattice theory. They generalize to categories as , with lower adjoints preserving joins and upper adjoints preserving meets.

Closure and interior operators are closely related to Galois connections. Closure operators extend, preserve order, and are idempotent, while interior operators contract, preserve order, and are idempotent. Both have applications in formal concept analysis and data mining.

Galois Connections

Definition and Properties

Top images from around the web for Definition and Properties
Top images from around the web for Definition and Properties
  • establishes a relationship between two partially ordered sets (posets) through a pair of monotone functions
    • Consists of two functions, typically denoted as ff and gg, that map between the posets in opposite directions
    • Satisfies the condition: for all elements xx in the first poset and yy in the second poset, f(x)yf(x) \leq y if and only if xg(y)x \leq g(y)
  • Adjoint functors generalize the concept of Galois connections to categories
    • Given two categories CC and DD, an adjunction between them consists of a pair of functors F:CDF: C \to D and G:DCG: D \to C
    • FF is called the left adjoint and GG is called the right adjoint
  • Lower adjoint (or left adjoint) preserves joins (least upper bounds) and maps elements to their largest possible image in the codomain
    • Formally, for a Galois connection (f,g)(f, g) between posets PP and QQ, ff is the lower adjoint if f(X)=f(X)f(\bigvee X) = \bigvee f(X) for any subset XPX \subseteq P

Upper Adjoint and Residuated Mappings

  • Upper adjoint (or right adjoint) preserves meets (greatest lower bounds) and maps elements to their smallest possible image in the codomain
    • Formally, for a Galois connection (f,g)(f, g) between posets PP and QQ, gg is the upper adjoint if g(Y)=g(Y)g(\bigwedge Y) = \bigwedge g(Y) for any subset YQY \subseteq Q
  • Residuated mapping is a generalization of the upper adjoint in a Galois connection
    • Given a partially ordered set PP and a QQ, a function f:PQf: P \to Q is residuated if there exists a function g:QPg: Q \to P such that for all xPx \in P and yQy \in Q, f(x)yf(x) \leq y if and only if xg(y)x \leq g(y)
    • The function gg is called the residual of ff and is an upper adjoint

Closure and Interior Operators

Closure Operators

  • on a partially ordered set PP is a function c:PPc: P \to P that satisfies three properties:
    1. Extensive: xc(x)x \leq c(x) for all xPx \in P
    2. Monotone: if xyx \leq y, then c(x)c(y)c(x) \leq c(y) for all x,yPx, y \in P
    3. Idempotent: c(c(x))=c(x)c(c(x)) = c(x) for all xPx \in P
  • Fixed points of a closure operator form a complete lattice
    • An element xPx \in P is a fixed point of cc if c(x)=xc(x) = x
    • The set of all fixed points of cc is denoted as Fix(c)\text{Fix}(c)

Interior Operators

  • on a partially ordered set PP is a function i:PPi: P \to P that satisfies three properties:
    1. Contractive: i(x)xi(x) \leq x for all xPx \in P
    2. Monotone: if xyx \leq y, then i(x)i(y)i(x) \leq i(y) for all x,yPx, y \in P
    3. Idempotent: i(i(x))=i(x)i(i(x)) = i(x) for all xPx \in P
  • Closure and interior operators are dual notions
    • If cc is a closure operator on PP, then the function i(x)={yPc(y)x}i(x) = \bigvee\{y \in P \mid c(y) \leq x\} is an interior operator on PP
    • Conversely, if ii is an interior operator on PP, then the function c(x)={yPxi(y)}c(x) = \bigwedge\{y \in P \mid x \leq i(y)\} is a closure operator on PP

Applications

Concept Lattices and Formal Concept Analysis

  • Concept lattices arise from the application of Galois connections to formal concept analysis (FCA)
    • FCA is a mathematical framework for analyzing the relationships between objects and their attributes
    • Starts with a formal context (G,M,I)(G, M, I), where GG is a set of objects, MM is a set of attributes, and IG×MI \subseteq G \times M is a binary relation indicating which objects have which attributes
  • Galois connection between the power sets of objects and attributes induces a concept lattice
    • For a formal context (G,M,I)(G, M, I), define two functions f:2G2Mf: 2^G \to 2^M and g:2M2Gg: 2^M \to 2^G as follows:
      • f(A)={mM(g,m)I for all gA}f(A) = \{m \in M \mid (g, m) \in I \text{ for all } g \in A\}
      • g(B)={gG(g,m)I for all mB}g(B) = \{g \in G \mid (g, m) \in I \text{ for all } m \in B\}
    • The pair (f,g)(f, g) forms a Galois connection between (2G,)(2^G, \subseteq) and (2M,)(2^M, \supseteq)
    • The fixed points of the composite functions fgf \circ g and gfg \circ f are called formal concepts and form a complete lattice, known as the concept lattice of the formal context
  • Concept lattices have applications in various domains, such as:
    • Knowledge representation and ontology engineering
    • Data mining and machine learning (e.g., association rule mining, clustering)
    • Information retrieval and recommender systems
    • Software engineering (e.g., feature modeling, code analysis)
© 2024 Fiveable Inc. All rights reserved.
AP® and SAT® are trademarks registered by the College Board, which is not affiliated with, and does not endorse this website.


© 2024 Fiveable Inc. All rights reserved.
AP® and SAT® are trademarks registered by the College Board, which is not affiliated with, and does not endorse this website.

© 2024 Fiveable Inc. All rights reserved.
AP® and SAT® are trademarks registered by the College Board, which is not affiliated with, and does not endorse this website.
Glossary
Glossary