Example 5: Minimise : no feasible region
Minimise
subject to the constraints:
Draw the lines (graph):
through and ;
through and .
At : is false, so for (1) shade the side of away from the origin; is true, so for (2) shade the side of containing the origin. With , (2) and (3) give the triangle with corners , , .
The two shaded parts have no common point (graph). Indeed, for any point obeying (2) and (3),
so (1) fails at every such point.
There is no feasible region, hence no feasible solution: has no minimum value.
Want the reason behind every step, the basics and the checks?
Before looking for corners, there must be a region. Here the constraints contradict each other: with keeps only a small triangle near the origin, where is at most 5, while demands at least 8. No point obeys all the constraints, so there is nothing to minimise over.
- Given
- and the constraints , , , .
- To find
- The minimum of , or show that there is none. The book's solution also asks: is there any feasible region, and why?
- Plan
- (1) and (2) Draw each line and choose its side. (3) Find the region kept by (2) and (3) and the largest value of on it. (4) Show that no point obeys (1) as well. (5) Conclude.
Basics you need
Draw each line from its axis points and test to choose its side. Basics: where a line meets the axes · Basics: which side
A point is feasible only if it obeys every constraint. If the kept sides have no common point, there is no feasible region. Basics: where the shadings overlap
On a bounded region, a linear expression such as is largest at a corner (Theorem 2, Section 12.2.2). Basics: corners
If and , then ; and adding the same amount to both sides keeps an inequality true (Class XI, Linear Inequalities).
Solution
Draw the line and choose its side
Its axis points are and . Test in (1): is false, so (1) keeps the side of the line away from the origin.
"At least 8" fails at the origin, so the kept side is the far one: points such as , or .Draw the line and choose its side
Put : . Put : . The line joins and . Test in (2): is true, so (2) keeps the origin's side.
"" with a positive number on the right keeps the origin's side.The region kept by (2) and (3)
In the first quadrant, (2) keeps the triangle with corners , and . The largest value of on it is at a corner:
Corner of the triangle ← largest So every point of the triangle has .
The triangle is bounded and is linear, so its largest value on the triangle is at a corner, just as for in earlier examples. This table shows how far short of 8 the triangle falls.No point obeys all the constraints
Example 5, drawn with equal scales. Gold: the part of the first quadrant with . Violet: the triangle , with corners , , . The two shaded parts have no common point. The graph shows a gap between the violet triangle and the gold side. To prove it for every point: if , and , then, because ,
But (1) needs , and . So no point satisfies (1), (2) and (3) together.
This answers the book's questions "Is there any feasible region? Why is so?": no, because the constraints (2) and (3) force , while (1) forces . The two lines do meet, at , but that point has , outside the first quadrant.Conclude
There is no feasible region, and hence no feasible solution. With no feasible point at all, has no minimum value (and no maximum either).
An optimal solution must be a feasible point, and there are none. So the Corner Point Method has nothing to work on: no region, no corners, no table of . The book closes with a Remark: whenever a feasible region exists it is convex, and the optimum (when there is one) is at a corner.
The constraints have no common point: (2) and (3) give , while (1) needs . So
and has no minimum value. (Textbook answer: the problem is having no feasible region and hence no feasible solution.)
Points that obey (1) break (2): gives ; gives ; gives ✓.
Points that obey (2) and (3) break (1): the corner has , and has ✓.
A computer check of the 13 whole-number points that obey (2) and (3) found the largest to be 5, at ; none reaches 8 ✓.
- Using the crossing point of the two lines as a corner. has , so it breaks (3).
- Writing a minimum such as at . The origin breaks (1); a minimum must be at a feasible point.
- Shading towards the origin for . The test is false.
- Ending without a conclusion. Say it in words: no feasible region, so no feasible solution and no minimum.
- Always check that the kept sides overlap before looking for corners.
- A one-line combination of the constraints, here , proves that a region is empty.
- So far: a bounded region gives both optima (Examples 1 to 3), an unbounded one needs the half-plane test (Example 4), and an empty one gives no solution.
Every fact used, and where it comes from
| Fact | Used in | Source |
|---|---|---|
| A line from its axis points; a test point chooses the side | Steps 1, 2 | Class XI, Linear Inequalities (Basics, Basics) |
| A linear expression on a bounded region is largest at a corner | Step 3 | Theorem 2, Section 12.2.2 (Basics) |
| Feasible region: common region of all the constraints | Step 4 | Section 12.2.2 (Basics) |
| Inequalities can be chained and added | Step 4 | Class XI, Linear Inequalities |
| No feasible region means no feasible solution | Step 5 | Example 5, Section 12.2.2 (Lesson) |
| Remarks: the feasible region is convex; optima occur at corners | Step 5 | Remarks after Example 5, Section 12.2.2 (Basics) |
| All values | Steps, Check | Recomputed with exact fractions in the site's verification script |
Tell us if something on this page is wrong or unclear, or if you would like a fuller explanation, another method or a new feature. Send feedback on this problem →