The mathematically best design ignoring constraints is often infeasible in practice. Add a real constraint, and the actual optimum usually isn't the unconstrained peak — it's wherever the constraint boundary cuts it off.
Optimization finds the best value of a design variable (or set of variables) according to some objective — minimum cost, maximum strength, minimum weight — subject to real-world constraints that limit what values are actually achievable. A crucial, sometimes counterintuitive result: once constraints are added, the true optimum is very often not the mathematically ideal unconstrained peak, but sits exactly at whichever constraint boundary limits it.
The unconstrained mathematical optimum of an objective function represents the best possible value if any design were physically achievable — but real engineering problems always operate under real limits: material availability, manufacturing tolerances, cost budgets, physical space, code-required minimums or maximums. When the true unconstrained optimum falls outside those limits, it's simply not a usable answer, regardless of how mathematically 'best' it appears in isolation.
As demonstrated above, when a constraint actually excludes the unconstrained optimum, the best feasible design is typically found right at the constraint boundary itself — moving further into the feasible region away from the boundary only makes the objective worse (since the true peak lies on the excluded side), so the best available point is exactly where the constraint stops you. This 'active constraint' behavior is a fundamental and recurring pattern in constrained optimization problems.
Nearly every real engineering optimization problem — minimizing material cost subject to a strength requirement, maximizing efficiency subject to a size limit, minimizing weight subject to a safety factor — is a constrained optimization problem, not an unconstrained one. Recognizing which constraints are actually 'active' (binding the solution) versus merely present but not limiting is central to understanding why a real optimized design looks the way it does, and which requirement is actually driving that design.
Because constraints represent genuine physical, economic, or regulatory limits — a design value outside the feasible region simply cannot be built, afforded, or legally implemented, regardless of how mathematically optimal it appears. The constrained optimum represents the actual best achievable design given real-world limits, which is the only kind of 'optimal' that matters for a real engineering decision.
An active (binding) constraint is one that actually limits the achievable optimum — the true optimal solution sits exactly at that constraint's boundary. An inactive constraint is present in the problem but doesn't actually restrict the solution, since the optimum would satisfy it anyway even without it being explicitly enforced.
No — as shown when the constraint isn't binding above, if the unconstrained optimum already satisfies all constraints, it remains the true optimum without needing to sit at any boundary. The optimum sits at a constraint boundary specifically when that constraint excludes the otherwise-best unconstrained solution, forcing the real answer to the edge of what's actually achievable.
Try our STEM Learning Studio
More calculators, simulators, and guides for this discipline.