Relaxation (Solving Easier Versions of Hard Problems)

A mathematical optimization technique where you solve a simplified ('relaxed') version of a hard problem by removing constraints, then use that solution to guide the solution of the original problem. In computer science, relaxing an integer programming problem to allow fractional solutions makes it solvable, and the relaxed solution provides bounds on the optimal solution. Applied broadly: when a problem is too complex to solve directly, remove the hardest constraint, solve the easier version, then reintroduce the constraint incrementally.

When to use it

When a problem has so many constraints that no approach seems viable; when analysis paralysis comes from trying to satisfy everything simultaneously; when creative solutions require understanding what's possible before what's practical; when strategic planning needs to separate 'what we'd do ideally' from 'what constraints prevent.'

How it can help

When facing an intractable problem, ask: what constraint makes this hard? Remove it, solve the easier version, then figure out how to reintroduce the constraint. 'How should we redesign the organization?' is intractable. 'How should we redesign the organization if budget weren't a constraint?' is easier and produces insights. 'How should we redesign the organization if we could hire anyone?' produces different insights. Each relaxation reveals structure in the problem that the full-constraint version hides. Then progressively reintroduce constraints to move from ideal to feasible.