Glossary · approach
Column Generation
A large-scale optimization method that generates candidate decisions (columns) on demand instead of enumerating all of them upfront.
Delayed Column Generation
Column generation is the method used when listing every possible decision (e.g. every possible cutting pattern, every possible vehicle route) in advance is impractical. It starts from a small initial set, solves the problem, then mathematically asks 'is there a new decision that would improve the solution' and adds it if so. Developed by Gilmore and Gomory in 1961 for the cutting stock problem, it is today the workhorse for large-scale problems like vehicle routing (VRP), crew scheduling, and air-traffic assignment.
Örnek
A textile shop has, in theory, millions of possible cutting patterns. Column generation does not enumerate them — it discovers improvement-bearing patterns one at a time, delivering a near-optimal solution in practical time.