Closure: Why Does an Operation Keep Results in the Same Domain?

Closure does not mean shutting things away. It describes a property of an operation: after the operation is applied, the result remains inside the specified domain. The integers are closed under addition because adding any two integers produces another integer. They are not closed under division, because dividing 1 by 2 takes us outside the integers.

Closure is therefore never a label belonging to the objects alone. We must name both the domain and the operation. The same integers are closed under addition but not under division. This matters because closure allows a rule to be used repeatedly: the result of one step is still an acceptable input for the next, so a complex process can be built from the same operation without leaving its working domain.

Closure is related to an invariant, but they are not the same. An invariant is a property that remains unchanged through a transformation. Closure merely requires the result to stay within a domain; the object itself may change. Nor does closure mean a closed system, which concerns exchange with an environment. The useful question is precise: within which domain, and under which operation, can the result continue to participate in the same relation?