Selection

Definition

Selection (symbol: σ, sigma) is the relational algebra operation that picks a SUBSET OF ROWS from a relation based on a condition — it never removes or adds columns, only rows. σ(gpa >= 3.5)(Students) reads as "select the rows of Students where gpa >= 3.5."

How It Works

Selection maps directly onto SQL's WHERE clause:

sql
SELECT id, name, department, gpa FROM students WHERE gpa >= 3.5 ORDER BY id;

The algebraic notation σcondition(Relation) and the SQL WHERE clause express exactly the same idea: filter rows, keep the full row shape (same columns) for every row that passes. Selection is one of the two most fundamental relational algebra operations, alongside Projection (7.2) — nearly every real query is built by combining these two with joins.

Edge Cases and Pitfalls

• Selection is often confused with Projection because both "narrow down" a relation — but Selection narrows ROWS (keeps all columns, fewer rows), while Projection narrows COLUMNS (keeps all matching rows, fewer columns). They answer different questions: "which rows?" vs. "which columns?"
• A selection condition involving NULL follows three-valued logic (see Chapter 8's NULL Semantics) — σ(department = NULL) is not how you select rows with a missing department; that requires the SQL-specific IS NULL test, which doesn't have a clean pure-relational-algebra equivalent in the classic (NULL-free) formulation of the algebra.
• Multiple selection conditions can be combined with AND/OR, corresponding to combining SQL WHERE conditions the same way: σ(gpa >= 3.5 AND department = 'Engineering')(Students).

Key Takeaways

• Selection (σ) filters rows by a condition; SQL's WHERE clause is its direct equivalent.
• It never changes which columns appear — only which rows survive.
• Selection and Projection are the two foundational, complementary "narrowing" operations of relational algebra.