The Simplex Method

Ideas from Solving LPs Graphically

1.  If there is an optimal solution, it is a CPF solution (CPFS).

2.  There are a finite number of CPFS.

3.  CPFS are the intersection of n (# variables) defining equations.

4.  Adjacent CPFS have n-1 defining equations in common.

5.   If a CPFS is optimal, it is better than all (adjacent) CPFS.

6.  If a CPFS is better than all (adjacent) CPFS, it is optimal.


Basis of the Simplex Algorithm

1. Start at any CPFS and move to a better one.

2. Stop if you reach a CPFS better than all adjacent CPFS.


The Simplex Algorithm

1. Convert to standard form.

2. Find (if possible) a bfs.

3. Determine whether the current bfs is optimal.

4. If not, determine which nonbasic variable should become a basic (entering) variable and which basic variable should become a nonbasic (leaving) variable.

5. Use EROs to find the new bfs with better objective function value.

6. Go back to step 3.