Class 12 Maths, Worked in Full Past papersGet updates
Chapter 12 · Linear Programming

Example 4: Z=−50x+20yZ = -50x + 20y on an unbounded region: no minimum

ExamplesUnbounded feasible region, the open half-plane test · Section 12.2.2
The question

Determine graphically the minimum value of the objective function

Z=−50x+20y…(1)Z = -50x + 20y \qquad \ldots(1)

subject to the constraints:

2x−y≥−5…(2)3x+y≥3…(3)2x−3y≤12…(4)x≥0, y≥0…(5)\begin{aligned} 2x - y &\ge -5 &&\ldots(2) \\ 3x + y &\ge 3 &&\ldots(3) \\ 2x - 3y &\le 12 &&\ldots(4) \\ x \ge 0,\ y &\ge 0 &&\ldots(5) \end{aligned}

After its solution the book adds:

In the above example, can you say whether z=−50x+20yz = -50x + 20y has the maximum value 100 at (0,5)(0, 5)⁠? For this, check whether the graph of −50x+20y>100-50x + 20y > 100 has points in common with the feasible region. (Why?)

Exam answer

(i) Draw the lines (graph):

2x−y=−52x - y = -5 through (−52,0)\left(-\frac{5}{2}, 0\right) and (0,5)(0, 5)⁠;

3x+y=33x + y = 3 through (1,0)(1, 0) and (0,3)(0, 3)⁠;

2x−3y=122x - 3y = 12 through (6,0)(6, 0) and (0,−4)(0, -4)⁠.

At (0,0)(0, 0)⁠: 0≥−50 \ge -5 is true, so shade towards the origin for (2); 0≥30 \ge 3 is false, so shade away from the origin for (3); 0≤120 \le 12 is true, so shade towards the origin for (4). With x,y≥0x, y \ge 0⁠, the feasible region (shaded) is unbounded.

CornerZ=−50x+20y(0,5)100(0,3)60(1,0)−50(6,0)−300←smallest\begin{array}{c|c} \text{Corner} & Z = -50x + 20y \\ \hline (0, 5) & 100 \\ (0, 3) & 60 \\ (1, 0) & -50 \\ (6, 0) & -300 \leftarrow \text{smallest} \end{array}

The region is unbounded, so we check the open half-plane −50x+20y<−300-50x + 20y < -300⁠, i.e. −5x+2y<−30-5x + 2y < -30⁠. Draw the line −5x+2y=−30-5x + 2y = -30 dashed (graph), through (6,0)(6, 0) and (10,10)(10, 10)⁠.

The point (9,2)(9, 2) is feasible, as 16≥−516 \ge -5⁠, 29≥329 \ge 3⁠, 12≤1212 \le 12⁠. It gives −45+4=−41<−30-45 + 4 = -41 < -30⁠, so this half-plane has points in common with the feasible region.

∴\therefore ZZ has no minimum value.

(ii) The largest corner value is 100, at (0,5)(0, 5)⁠. As the region is unbounded, 100 is the maximum only if the open half-plane −50x+20y>100-50x + 20y > 100⁠, i.e. −5x+2y>10-5x + 2y > 10⁠, has no point in common with the region. That is why we check it.

On the region, (2) gives y≤2x+5y \le 2x + 5⁠, and x≥0x \ge 0⁠, so

−5x+2y≤−5x+4x+10=10−x≤10.\begin{aligned} -5x + 2y &\le -5x + 4x + 10 \\ &= 10 - x \le 10. \end{aligned}

So no feasible point has −5x+2y>10-5x + 2y > 10⁠: the region lies on one side of the dashed line −5x+2y=10-5x + 2y = 10 and meets it only at (0,5)(0, 5) (graph).

∴\therefore ZZ has the maximum value 100 at (0,5)(0, 5)⁠.

Want the reason behind every step, the basics and the checks?

The one idea this problem is built on

When the feasible region goes on for ever, the corner table alone cannot be trusted: ZZ may keep falling (or rising) as we move far out. The smallest corner value mm is the minimum only if no feasible point gives a smaller ZZ⁠, that is, only if the open half-plane Z<mZ < m misses the region. Here it does not miss, so there is no minimum. For the maximum the test is Z>MZ > M⁠, and that half-plane does miss the region, so the maximum is M=100M = 100⁠.

Given, to find, and the plan
Given
Z=−50x+20yZ = -50x + 20y and the constraints 2x−y≥−52x - y \ge -5⁠, 3x+y≥33x + y \ge 3⁠, 2x−3y≤122x - 3y \le 12⁠, x≥0x \ge 0⁠, y≥0y \ge 0⁠.
To find
(i) The minimum value of ZZ⁠, if there is one. (ii) Whether ZZ has the maximum value 100 at (0,5)(0, 5)⁠, and why the half-plane −50x+20y>100-50x + 20y > 100 is the thing to check.
Plan
Part (i): draw the three lines and choose their sides; shade the unbounded region; find its four corners; tabulate ZZ⁠; apply the open half-plane test to the smallest value. Part (ii): apply the same test to the largest value.

Basics you need

1Lines and sides

Draw ax+by=cax + by = c from its axis points (an axis point may have a negative coordinate); test (0,0)(0, 0) to choose the side. Basics: where a line meets the axes · Basics: which side

2Bounded or unbounded

A region is bounded if a circle can be drawn round it; otherwise it is unbounded and goes on for ever in some direction. Basics: bounded or not

3Corners and the table

Corners are where two boundary lines meet; ZZ is worked out at each. Call the largest value MM and the smallest mm⁠. Basics: a corner solves two equations

4The open half-plane test (unbounded region)

mm is the minimum of Z=ax+byZ = ax + by if the open half-plane ax+by<max + by < m has no point in common with the region; otherwise ZZ has no minimum. Likewise MM is the maximum if ax+by>Max + by > M has no common point; otherwise there is no maximum. "Open" means the line itself is left out. Basics: lines of equal Z

Part (i): the minimum of ZZ

  1. Draw the line 2x−y=−52x - y = -5 and choose its side

    Put y=0y = 0⁠: 2x=−52x = -5⁠, so x=−52x = -\frac52⁠. Put x=0x = 0⁠: −y=−5-y = -5⁠, so y=5y = 5⁠. The line joins (−52,0)\left(-\frac52, 0\right) and (0,5)(0, 5)⁠; it also passes through (1,7)(1, 7)⁠.

    Test (0,0)(0, 0) in (2): 0≥−50 \ge -5 is true, so (2) keeps the side of the line that contains the origin. In the first quadrant this is the side below the line: (2) says y≤2x+5y \le 2x + 5⁠.

    The xx-axis point is to the left of the origin, outside the first quadrant, but it still fixes the line. The test point needs care with the minus signs: 2(0)−0=02(0) - 0 = 0⁠, and 00 is greater than −5-5⁠.
  2. Draw the line 3x+y=33x + y = 3 and choose its side

    Its axis points are (1,0)(1, 0) and (0,3)(0, 3)⁠. Test (0,0)(0, 0) in (3): 0≥30 \ge 3 is false, so (3) keeps the side away from the origin.

    "At least 3" fails at the origin, so the kept side is the far one.
  3. Draw the line 2x−3y=122x - 3y = 12 and choose its side

    Put y=0y = 0⁠: x=6x = 6⁠. Put x=0x = 0⁠: −3y=12-3y = 12⁠, so y=−4y = -4⁠. The line joins (6,0)(6, 0) and (0,−4)(0, -4)⁠; it also passes through (9,2)(9, 2)⁠.

    Test (0,0)(0, 0) in (4): 0≤120 \le 12 is true, so (4) keeps the origin's side, which is the side above the line: (4) says y≥2x−123y \ge \frac{2x - 12}{3}⁠.

    Again one axis point, (0,−4)(0, -4)⁠, is outside the first quadrant. Rewriting (4) as 3y≥2x−123y \ge 2x - 12 (adding 3y−123y - 12 to both sides) shows directly that the kept side is above the line.
  4. Shade the feasible region: it is unbounded

    In the first quadrant, the region lies below 2x−y=−52x - y = -5⁠, beyond 3x+y=33x + y = 3⁠, and above 2x−3y=122x - 3y = 12⁠. It is shaded below; the darker part and the dashed line are used in Part (i), Step 7.

    x y O 2 4 8 10 12 14 −2 2 6 8 10 (−5/2, 0) (0, −4) (0, 3) (1, 0) (6, 0) 2x − y = −5 3x + y = 3 2x − 3y = 12 (9, 2) (0, 5) −5x + 2y = −30 Z < −300
    The feasible region of Example 4 (shaded, drawn with equal scales). It has corners (0,5)(0, 5)⁠, (0,3)(0, 3)⁠, (1,0)(1, 0)⁠, (6,0)(6, 0) and is open towards the top right. The dashed line is −5x+2y=−30-5x + 2y = -30⁠; the darker part of the region lies in the open half-plane −5x+2y<−30-5x + 2y < -30⁠, where Z<−300Z < -300⁠. It contains (9,2)(9, 2)⁠.

    The region is unbounded: every point (t,t)(t, t) with t≥1t \ge 1 is in it (t≥−5t \ge -5⁠, 4t≥34t \ge 3⁠, −t≤12-t \le 12⁠), and these points go as far from O as we like, so no circle can enclose the region.

    Between the two upward-sloping lines y=2x+5y = 2x + 5 and y=2x−123y = \frac{2x - 12}{3} there is a strip that keeps widening to the right, because the first line is steeper. Nothing closes it off, which is what "unbounded" means.
  5. Find the corner points

    (0,5)(0, 5)⁠: line (2) meets the yy-axis. (0,3)(0, 3)⁠: line (3) meets the yy-axis. (1,0)(1, 0)⁠: line (3) meets the xx-axis. (6,0)(6, 0)⁠: line (4) meets the xx-axis. Each obeys every constraint; for example (6,0)(6, 0)⁠: 12≥−512 \ge -5⁠, 18≥318 \ge 3⁠, 12≤1212 \le 12⁠.

    No other crossing point is in the region. Lines (2) and (3) meet at (−25,215)\left(-\frac25, \frac{21}{5}\right)⁠, lines (2) and (4) at (−274,−172)\left(-\frac{27}{4}, -\frac{17}{2}\right)⁠, lines (3) and (4) at (2111,−3011)\left(\frac{21}{11}, -\frac{30}{11}\right)⁠: each has a negative coordinate. (−52,0)\left(-\frac52, 0\right) and (0,−4)(0, -4) are outside the first quadrant, and O breaks (3).

    An unbounded region still has corners; it just also has edges that run off for ever (here along lines (2) and (4)). Checking all the crossing points makes sure the list of corners is complete.
  6. Work out ZZ at each corner

    Corner pointZ=−50x+20yZ = -50x + 20y
    (0,5)(0, 5)0+100=1000 + 100 = 100 ← largest
    (0,3)(0, 3)0+60=600 + 60 = 60
    (1,0)(1, 0)−50+0=−50-50 + 0 = -50
    (6,0)(6, 0)−300+0=−300-300 + 0 = -300 ← smallest

    So m=−300m = -300⁠, at (6,0)(6, 0)⁠.

    If the region were bounded, Theorem 2 would make −300-300 the minimum at once. It is unbounded, so −300-300 may or may not be the minimum, and the next step decides.
  7. The open half-plane test for m=−300m = -300

    Following step 3(ii) of the Corner Point Method, we look at the open half-plane

    −50x+20y<−300,i.e.−5x+2y<−30\begin{aligned} -50x + 20y &< -300, \\ \text{i.e.}\quad -5x + 2y &< -30 \end{aligned}

    (dividing by 10). Its boundary −5x+2y=−30-5x + 2y = -30 passes through (6,0)(6, 0) and (10,10)(10, 10)⁠; it is drawn dashed because the line itself is not included.

    The feasible point (9,2)(9, 2) is in this half-plane: −5(9)+2(2)=−41<−30-5(9) + 2(2) = -41 < -30⁠, and indeed Z(9,2)=−450+40=−410<−300Z(9, 2) = -450 + 40 = -410 < -300⁠. So the half-plane has points in common with the feasible region (the darker part of the graph), and ZZ has no minimum value.

    The half-plane Z<−300Z < -300 contains exactly the points where ZZ is smaller than the best corner value. A feasible point there shows −300-300 is not the smallest value. In fact ZZ has no smallest value at all: along the edge 2x−3y=122x - 3y = 12⁠, at (6+3t,2t)(6 + 3t, 2t)⁠, Z=−300−110tZ = -300 - 110t⁠, which falls below any number as tt grows. The book prints the two tests of step 3(ii) as items "4. (a)" and "4. (b)" just below step 3(ii); the test used here is the one for the minimum, (b), and the book's text of Example 4 points to it as step 3(ii) of the Corner Point Method.

Part (ii): the book's follow-up, and why the half-plane is checked

  1. Why we check −50x+20y>100-50x + 20y > 100

    The largest corner value is M=100M = 100⁠, at (0,5)(0, 5)⁠. Because the region is unbounded, this is not automatically the maximum: ZZ might be larger at some point far out in the region. The test of the Corner Point Method (printed as item 4(a), just below step 3(ii)) says: MM is the maximum if the open half-plane −50x+20y>100-50x + 20y > 100 has no point in common with the feasible region; otherwise ZZ has no maximum value.

    This answers the book's "(Why?)". The open half-plane Z>100Z > 100 is exactly the set of points where ZZ beats the best corner. If a feasible point lay in it, 100 would not be the maximum. If none does, then Z≤100Z \le 100 at every feasible point, and Z=100Z = 100 is actually reached at (0,5)(0, 5)⁠, so 100 is the maximum.
  2. Draw the half-plane

    Dividing by 10, the half-plane is −5x+2y>10-5x + 2y > 10⁠. Its boundary −5x+2y=10-5x + 2y = 10 passes through (−2,0)(-2, 0) and (0,5)(0, 5)⁠, and the half-plane is the side above and to the left of it (the origin gives 0>100 > 10⁠, false).

    x y O 2 4 8 10 12 14 −2 2 6 8 10 (0, 3) (1, 0) (6, 0) 2x − y = −5 3x + y = 3 2x − 3y = 12 (0, 5) Z > 100 −5x + 2y = 10
    The same feasible region (blue, equal scales) with the dashed line −5x+2y=10-5x + 2y = 10 through (−2,0)(-2, 0) and (0,5)(0, 5)⁠. The open half-plane −5x+2y>10-5x + 2y > 10⁠, where Z>100Z > 100⁠, is the light rose side. It does not meet the region, which touches the dashed line only at (0,5)(0, 5)⁠.
    The dashed line is steeper (slope 52\frac52⁠) than the edge 2x−y=−52x - y = -5 (slope 2), and both pass through (0,5)(0, 5)⁠. So to the right of the yy-axis the dashed line runs above the region's upper edge.
  3. Show that there is no common point

    Every feasible point satisfies (2), that is y≤2x+5y \le 2x + 5⁠, and x≥0x \ge 0⁠. Hence

    −5x+2y≤−5x+2(2x+5)=10−x≤10.\begin{aligned} -5x + 2y &\le -5x + 2(2x + 5) \\ &= 10 - x \\ &\le 10. \end{aligned}

    So −5x+2y>10-5x + 2y > 10 never holds on the feasible region: the open half-plane has no point in common with it.

    Multiplying y≤2x+5y \le 2x + 5 by the positive number 2 keeps the direction. The graph suggests the answer; this one line of algebra proves it for the whole infinite region.
  4. Conclude

    Yes: Z=−50x+20yZ = -50x + 20y has the maximum value 100 at (0,5)(0, 5). Multiplying the last result by 10, Z≤100−10x≤100Z \le 100 - 10x \le 100 at every feasible point, with equality only when x=0x = 0 and y=5y = 5⁠.

    So on this unbounded region ZZ has a maximum but no minimum. Unboundedness only makes an optimum possible to lose; the half-plane test decides each case separately.
Result

The feasible region is unbounded, with corners (0,5)(0, 5)⁠, (0,3)(0, 3)⁠, (1,0)(1, 0)⁠, (6,0)(6, 0)⁠, where Z=100,60,−50,−300Z = 100, 60, -50, -300⁠.

(i) The smallest corner value −300-300 is not a minimum, because the open half-plane −5x+2y<−30-5x + 2y < -30 meets the region: ZZ has no minimum value. (Textbook answer: Z=−50x+20yZ = -50x + 20y has no minimum value subject to the given constraints.)

(ii) The open half-plane −5x+2y>10-5x + 2y > 10 has no point in common with the region, so ZZ has the maximum value 100 at (0,5)(0, 5). (The book asks this as a question and does not print the answer.)

Check it with numbers

Along the edge 2x−3y=122x - 3y = 12⁠: (9,2)(9, 2) gives Z=−410Z = -410⁠, (12,4)(12, 4) gives −600+80=−520-600 + 80 = -520⁠, and (36,20)(36, 20) gives −1800+400=−1400-1800 + 400 = -1400⁠. Each point is feasible, and each value fits Z=−300−110tZ = -300 - 110t with t=1,2,10t = 1, 2, 10⁠: there is no lowest value ✓.

For the maximum, Z≤100−10xZ \le 100 - 10x⁠: at (1,7)(1, 7)⁠, on the upper edge, Z=−50+140=90=100−10Z = -50 + 140 = 90 = 100 - 10 ✓; at (4,13)(4, 13)⁠, Z=−200+260=60=100−40Z = -200 + 260 = 60 = 100 - 40 ✓; at (0,4)(0, 4)⁠, Z=80≤100Z = 80 \le 100 ✓.

The corner (6,0)(6, 0) itself lies on the dashed line −5x+2y=−30-5x + 2y = -30⁠: −30=−30-30 = -30 ✓, and (0,5)(0, 5) lies on −5x+2y=10-5x + 2y = 10 ✓.

Common mistakes
  1. Stopping at the table. On an unbounded region, −300-300 is only a candidate; the answer "minimum −300-300" is wrong here.
  2. Using a closed half-plane. The test uses Z<mZ < m⁠, not Z≤mZ \le m⁠. The corner (6,0)(6, 0) itself gives Z=−300Z = -300⁠, so a "≤\le" test would always find a common point.
  3. Sign slips with 2x−y≥−52x - y \ge -5⁠. At the origin 0≥−50 \ge -5 is true, so the origin's side is kept.
  4. Deciding that an unbounded region has no maximum either. Part (ii) shows the maximum 100 exists; each test must be done.
What to take away
  1. Unbounded region: make the table, then test Z<mZ < m for a minimum and Z>MZ > M for a maximum, using open half-planes drawn dashed.
  2. One feasible point in the open half-plane is enough to show there is no optimum of that kind.
  3. To show there is no common point, a line of algebra (or a clear graph) for the whole region is needed, as in Part (ii), Step 3. Example 5 shows the last case: no feasible region at all.

Every fact used, and where it comes from

FactUsed inSource
A line from its axis points; a test point chooses the sidePart (i), Steps 1 to 3Class XI, Linear Inequalities (Basics, Basics)
Bounded: can be enclosed within a circle; otherwise unboundedPart (i), Step 4Section 12.2.2, footnote to Theorem 2 (Basics)
Corner point: where two boundary lines meetPart (i), Step 5Section 12.2.2, footnote to Theorem 1 (Basics)
Unbounded region: a maximum or minimum may not exist (Remark after Theorem 2)Part (i), Step 6Section 12.2.2
Open half-plane tests for mm and MM (printed as 4(a), 4(b) under step 3(ii))Part (i), Step 7; Part (ii), Step 1Corner Point Method, Section 12.2.2 (Lesson)
Multiplying an inequality by a positive number keeps its directionPart (ii), Step 3Class XI, Linear Inequalities
Textbook working, answer and the "(Why?)" questionResultExample 4, Section 12.2.2
All valuesSteps, CheckRecomputed with exact fractions in the site's verification script
Found an error, or want more?

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 →