Relations, Orders and Equivalence Classes
The basic vocabulary for describing how elements of a set relate to each other — equivalence relations group things that are 'the same' in some sense, while order relations rank things by 'bigger' or 'smaller'.
A relation on a set is simply a rule for which pairs of elements are considered "related" — nothing more than a chosen collection of pairs. Two families of relations show up constantly in quantitative work: equivalence relations, which capture "these two things count as the same for our purposes," and order relations, which capture "this thing ranks above that one."
An equivalence relation must satisfy three properties: reflexivity (every element relates to itself), symmetry (if relates to , then relates to ), and transitivity (if relates to and relates to , then relates to ). "Same sign" on the real numbers is an equivalence relation; it partitions all numbers into disjoint groups called equivalence classes — here, the positives, the negatives, and zero — where every element in a class is treated as interchangeable with every other element of that class for the purpose at hand.
An order relation instead requires reflexivity, transitivity, and antisymmetry (if relates to and relates to , then and must actually be equal), and it need not rank every pair — a "partial order" allows some elements to simply be incomparable. Ordinary on the real numbers is a "total order" because any two numbers are comparable; but portfolio dominance orderings (one portfolio dominates another only if it beats it on every risk measure simultaneously) are typically only partial orders, since two portfolios can each beat the other on a different measure and so are genuinely incomparable rather than tied.
Equivalence relations (reflexive, symmetric, transitive) partition a set into interchangeable classes; order relations (reflexive, transitive, antisymmetric) rank elements, though a partial order may leave some pairs genuinely incomparable rather than tied or ranked.
Related concepts
Practice in interviews
Further reading
- Halmos, Naive Set Theory