Linear programming model
A linear programming problem asks to maximise or minimise a linear objective subject to linear equality/inequality constraints, often with nonnegativity restrictions.
Decision variables
Choose variables for controllable quantities (e.g. products made) and attach correct units. Express every resource limit with an inequality in the correct direction.
Objective function
Profit, cost, time or yield is expressed as Z=ax+by. Its value is evaluated at feasible points; an optimal solution must satisfy every constraint.
Feasible region
The intersection of half-planes satisfying all constraints is the feasible region. It may be empty, bounded or unbounded.
Boundary lines and test points
To graph ax+by≤c, first draw ax+by=c, then test a point not on the line to determine the allowed side.
Corner points and graphical optimum
When an optimum exists for a linear objective over a polygonal feasible region, at least one corner point is optimal. Comparing corner-point values is central to the graphical method.
Bounded feasible regions
For a bounded, nonempty polygonal feasible region, finite maximum and minimum values exist for a continuous linear objective and are achieved at one or more corner points.
Unbounded feasible regions
An unbounded feasible region does not automatically imply no optimum; an objective may still attain a maximum or minimum. Test the objective's improving direction rather than assuming.
Multiple optima and degeneracy
If the objective line coincides with a feasible boundary segment, infinitely many points on that segment may be optimal; equal values at adjacent vertices signal this situation.
Interpreting solutions
Check feasibility, nonnegative quantities and practical constraints. A solution with fractional counts may need discrete/integer analysis beyond the continuous graphical model.