The original worked lessons remain the main explanations. Use these focused reminders to check the conditions before you start.
Optimisation with feasible domains
Eliminate a variable and state the allowed interval.
Reject infeasible stationary values and justify a global optimum.
Return every requested dimension, value and unit.
Build a one-variable objective and state its feasible domain. Differentiate; test all allowed stationary candidates and relevant endpoints. A local second-derivative test alone does not compare every feasible value. For integer inputs, compare the neighbouring allowed integers around a continuous optimum.