Lattice properties

In fcaR: Formal Concept Analysis

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(fcaR)

Introduction

Formal Concept Analysis (FCA) connects data analysis with Order Theory and Lattice Theory. Beyond simply extracting concepts, it is often useful to analyze the algebraic structure of the resulting Concept Lattice.

This vignette introduces a new set of features in fcaR, Algebraic Properties: Efficiently check if a concept lattice is distributive, modular, semimodular, or atomic.

1. Checking lattice properties

The ConceptLattice class now provides methods to verify standard lattice-theoretic properties. These checks are implemented in optimized C++ for performance.

Let's use the built-in planets dataset as an example:

fc <- FormalContext$new(planets)
fc$find_concepts()

# Check properties
print(paste("Is Distributive?", fc$concepts$is_distributive()))
print(paste("Is Modular?",      fc$concepts$is_modular()))
print(paste("Is Semimodular?",  fc$concepts$is_semimodular()))
print(paste("Is Atomic?",       fc$concepts$is_atomic()))

Understanding the properties

• Distributive: A lattice is distributive if the operations of join ($\vee$) and meet ($\wedge$) distribute over each other. Distributive lattices are isomorphic to rings of sets (Birkhoff's Representation Theorem).
• Modular: A generalization of distributivity. In a modular lattice, if $x \le z$, then $x \vee (y \wedge z) = (x \vee y) \wedge z$. All distributive lattices are modular.
• Upper Semimodular: If $x$ covers $x \wedge y$, then $x \vee y$ covers $y$. This property ensures the lattice is graded (all maximal chains have the same length).
• Atomic: Every element (except the bottom) lies above some atom (an element that covers the bottom). In FCA, atoms often correspond to object concepts.

Example: A non-distributive lattice ($M_3$)

The "Diamond" lattice ($M_3$) is the smallest non-distributive lattice. Let's create it manually to verify our checks.

# Context for M3 (The Diamond)
# 3 objects, 3 attributes. Objects have 2 attributes each.
I_m3 <- matrix(c(
  0, 1, 1,
  1, 0, 1,
  1, 1, 0
), nrow = 3, byrow = TRUE)

fc_m3 <- FormalContext$new(I_m3)
fc_m3$find_concepts()

# M3 is Modular but NOT Distributive
print(paste("M3 Distributive:", fc_m3$concepts$is_distributive()))
print(paste("M3 Modular:",      fc_m3$concepts$is_modular()))

2. Arrow Relations

Arrow relations highlight the structural relationships between objects and attributes in a binary formal context. They are particularly useful for identifying irreducible elements and understanding the lattice structure.

For a pair $(g, m) \notin I$: - $g \swarrow m$ iff $g'$ is maximal among intents not containing $m$. - $g \nearrow m$ iff $m'$ is maximal among extents not containing $g$. - $g \updownarrow m$ iff both $g \swarrow m$ and $g \nearrow m$.

In fcaR, you can compute these relations using the calculate_arrow_relations() method. Once computed, the print() and to_latex() methods will display the arrows:

# Use the planets dataset
fc <- FormalContext$new(planets)
fc$calculate_arrow_relations()

# The print method now shows the arrows
print(fc)

fcaR documentation built on July 27, 2026, 5:06 p.m.