Unbounded regions and when an optimum exists ·… | StemPy Lab
Course 19.2·Lesson 2 of 3
Unbounded regions and when an optimum exists
When the feasible region has no outer edge, the corner values alone do not prove an optimum. Take the best corner value M and draw the open half-plane that would beat it. If this half-plane has no point in common with the region, M is the optimum. If it has a common point, the optimum does not exist. For a cost on a region with ≥ rules, the minimum usually exists and the maximum does not.
about 25 min1 simulation
Lesson 2 of 3Show all lessonsHide
Try the code and the simulations beside the lesson.
What you will be able to do
say why an unbounded region needs an extra test;
use the half-plane test to decide whether a maximum or minimum exists;
see why a negative number in Z can destroy a minimum.
An open region
Think of a diet. A person must get at least 4 units of one vitamin from two foods, so x+y≥4. The person also needs at least 6 units of another, so x+3y≥6. Add x≥0 and y≥0. The feasible region is unbounded, and it has corners at (0,4), (3,1) and (6,0).
The cost is Z=3x+2y rupees, and the person wants the smallest cost. The values at the corners are Z=8, 11 and 18. The smallest is 8.
But is 8 really the minimum? The region never ends, so there may be a far-away point with a smaller Z. Here we feel that the answer is no, because going further out costs more. A feeling is not a proof. We need a test.
The half-plane test
Let M be the smallest corner value. Draw the open half-planeZ<M, which is the set of points with ax+by<M, without its boundary line.
If this half-plane has no point in common with the feasible region, then M is the minimum.
If it has a point in common, then Z has no minimum.
For a maximum, let M be the largest corner value and use the half-plane Z>M in the same way.
The unbounded region x + y at least 4, x + 3y at least 6, x and y at least 0, shaded light blue above and to the right of the corners (0, 4), (3, 1) and (6, 0). A red dashed line, 3x + 2y = 8, passes through (0, 4) and touches the region only there. The small triangle between this line and the origin, where 3x + 2y is less than 8, is shaded grey and shares no point with the region. Arrows at the top and right edges show that the region does not end. A key in the picture names the red line and the grey triangle.
For the diet, M=8 and the half-plane is 3x+2y<8. Every feasible point has x+y≥4. Then
3x+2y≥2x+2y=2(x+y)≥8.
So no feasible point has Z<8. The half-plane has nothing in common with the region, and the minimum is 8, at (0,4).
Now try the maximum. The largest corner value is 18. The half-plane 3x+2y>18 contains (10,0), which is feasible, since 10+0≥4 and 10≥6. So the half-plane meets the region, and there is no maximum. Pushing x higher makes Z as large as you like.
When a minimum fails
Keep the same region but take Z=x−y. The corner values are −4, 2 and 6, so M=−4. The half-plane is x−y<−4.
The point (0,5) is feasible, since 0+5≥4 and 15≥6. It gives Z=−5, which is below −4. So the half-plane meets the region, and there is no minimum. The point (0,100) gives Z=−100, and Z keeps falling as y grows.
The negative number in Z is the cause. Going up the y axis lowers Z without limit.
A rule of thumb
For a region such as the diet, where the rules use ≥ and the region reaches out to the right and upwards:
if a≥0 and b≥0, a minimum of ax+by exists, and a maximum does not, except in trivial cases;
if a or b is negative, the minimum does not exist.
Still use the half-plane test when you write the answer. It is the proof.
An unbounded region does not mean there is no answer. One of the two, the maximum or the minimum, may still exist.
Practise with new numbers
Try it in Python
The program checks the region by brute force. It looks at all whole-number points up to 40 in each direction and counts those with Z below the smallest corner value.
Common mistakes
Stopping at the smallest corner value. On an unbounded region this is only a candidate. Apply the test.
Saying that an unbounded region has no solution. It may have a minimum, a maximum, both or neither.
Using the wrong half-plane. For a minimum use Z<M. For a maximum use Z>M.
Using the closed half-plane. The points with Z=M are allowed to be in the region. Use the strict sign.
Remember these
On a bounded region the corner values decide. On an unbounded region they only give a candidate.
Candidate M is the optimum if the open half-plane that beats it has no feasible point.
A cost with positive numbers on a region with ≥ rules has a minimum but no maximum.
A negative number in Z can make the minimum fail.
References
NCERT, Mathematics Textbook for Class XII, Part II, Chapter 12, Linear Programming (Theorem 2, the unbounded feasible region).
Check yourself
Try each one on your own first. If you are stuck, open a hint. Then check with Show the answer.
On an unbounded region the smallest corner value of Z is 20. What must you check before you call 20 the minimum?
For the region x+y≥4, x≥0, y≥0, find Z=2x+y at the corners (4,0) and (0,4). Is 4 the minimum?
For the same region, does Z=x+y have a maximum?
For the same region, does Z=x−2y have a minimum?
Can a problem with an unbounded region have both a maximum and a minimum?
Enroll free in The Corner Point Method
Every lesson is open to read. Enrolling also lets you:
Mark lessons done and carry on where you stopped.
Take the tests of this lesson and its course.
Count the course towards the certificate for its chapter.