MATH 140 Calculus 1 • McGill University, Montreal

Corrected exercises: optimization problems (MATH 140)

This is the corrected exercise set for optimization problems in MATH 140, Calculus 1, at McGill University, section 4.7 of Stewart. It is where the whole course is used at once: a figure becomes a function, the chain and product rules produce the derivative, and the extreme value results of the previous chapters decide the answer. Every number is exact and chosen to be done by hand, and each solution names the theorem that licenses its conclusion, because an optimum given without its justification earns only part of the marks.

The thread running through the whole set: a critical number is only a CANDIDATE, and the domain, read from the geometry before anything is differentiated, decides. On a closed interval the closed interval method compares the candidates with BOTH endpoints; on an open or unbounded domain the sign of the derivative on the WHOLE domain proves the global extremum. The same gesture is posed on every exercise, with its variants: the critical number outside the domain, the endpoint that wins, the endpoint that is a real object and must be compared exactly.

The traps named in the solutions: answering with the critical number instead of the quantity asked, keeping a critical number that describes an impossible box, stopping at a second derivative test that only proves a local extremum, forgetting that the closed interval method is unavailable on an open domain, giving the minimum of d2d^2 as the distance, squaring an equation without checking signs, and reporting the unconstrained optimum when a capacity or a land limit moves the answer to an endpoint.

10 corrected exercises • 100 points • 150 minutes

Revision sheet for this chapter → Every MATH 140 chapter →

Self-checking set Type your answers under each question: the page tells you right or wrong before you open the solution. With an account, every answer right on the first try earns points.

Course recap

  • • Method: figure and variables, constraint, objective in ONE variable, domain from the situation, candidates, then a justification of the absolute extremum, then the answer to the question asked.
  • • Closed interval method: ff continuous on [a,b][a, b]; compare ff at the critical numbers in (a,b)(a, b) and at aa and bb. The largest value is the absolute maximum.
  • • First derivative test for absolute extreme values: if f′>0f' > 0 for all x<cx < c and f′<0f' < 0 for all x>cx > c in the domain, f(c)f(c) is the absolute maximum (reverse signs for a minimum).
  • • One critical number: on an interval, a single critical number that is a local extremum is the absolute one. f′′<0f'' < 0 on the whole domain also licenses a maximum.
  • • Since t\sqrt t and t2t^2 (for t≥0t \ge 0) are increasing, minimize d2d^2 instead of dd, and maximize A2A^2 instead of A≥0A \ge 0: same point, different value.
  • • Classics: can of fixed volume h=2rh = 2r; Norman window h=rh = r; square sheet box x=a6x = \frac{a}{6}; cylinder in a cone r=2R3r = \frac{2R}{3}, h=H3h = \frac{H}{3}.

Part A: the basics (/50)

Exercise 1: The method on a divided pasture: constraint, objective, domain, global maximum

Every optimization problem is solved in the same five moves: name the variables on a figure; write the CONSTRAINT that links them; write the OBJECTIVE function in one variable; give its DOMAIN, read from the situation; then find the absolute extremum AND justify that it is absolute. The derivative only supplies candidates. The last move is where marks are lost.

A farmer has 600600 m of fencing to enclose a rectangular pasture along a straight river, and to split it into two pens by one fence perpendicular to the river, as in the figure. No fence is needed along the river. Let xx be the length of each of the three fences perpendicular to the river and yy the length of the side parallel to it, in metres.

river: no fence on this sidexxxypen 1pen 2
  • a) Write the constraint, the total area AA as a function of xx alone, and the domain of xx. Explain why the two endpoints of the domain may be included.
  • b) Find the critical number of AA and use the closed interval method to find the dimensions of the pasture of largest area, and that area.
  • c) The farmer's land only extends 8080 m back from the river, so x≤80x \le 80. Find the dimensions of the largest pasture now.
  • d) A classmate answers b) with: A′′(x)=−6<0A''(x) = -6 < 0, so x=100x = 100 is a maximum. What exactly does this prove, and why is the maximum global here anyway?
  • e) With LL metres of fencing and nn pens (so n+1n + 1 fences perpendicular to the river), prove that the optimal side parallel to the river is always y=L2y = \frac{L}{2}.

Type your answers, the page tells you right or wrong 0/9

a)
Domain of xx ,
b)
c)
d)
e)
Show the solution

Answers

  • a) 3x+y=6003x + y = 600, A(x)=x(600−3x)=600x−3x2A(x) = x(600 - 3x) = 600x - 3x^2, domain [0,200][0, 200]
  • b) x=100x = 100 m, y=300y = 300 m, A=30 000A = 30\,000 m2^2 (two pens of 100100 by 150150 m)
  • c) A′>0A' > 0 on [0,80][0, 80]: x=80x = 80 m, y=360y = 360 m, A=28 800A = 28\,800 m2^2
  • d) A′′(100)<0A''(100) < 0 only gives a LOCAL maximum; it is global because it is the only critical number in [0,200][0, 200] and A(0)=A(200)=0A(0) = A(200) = 0.
  • e) x=L2(n+1)x = \frac{L}{2(n+1)}, so y=L−(n+1)x=L2y = L - (n+1)x = \frac{L}{2}

a) Counting the fence on the figure: three pieces of length xx and one of length yy, so the constraint is 3x+y=6003x + y = 600, that is y=600−3xy = 600 - 3x. The objective is the total area A=xyA = xy, and substituting the constraint, A(x)=x(600−3x)=600x−3x2A(x) = x(600 - 3x) = 600x - 3x^2. The domain comes from the geometry, not from the formula: x≥0x \ge 0 and y≥0y \ge 0, and y=600−3x≥0y = 600 - 3x \ge 0 gives x≤200x \le 200. So x∈[0,200]x \in [0, 200]. The endpoints are degenerate pastures of area 00 (no depth, or all the fence used across the river), and including them costs nothing: it makes the domain a CLOSED interval, on which the continuous polynomial AA is guaranteed an absolute maximum by the Extreme Value Theorem. That is the licence for the closed interval method.

b) A′(x)=600−6xA'(x) = 600 - 6x, which is 00 for x=100x = 100, inside [0,200][0, 200]. There is no point where A′A' fails to exist. Closed interval method: compare the values at the critical number and at BOTH endpoints: A(0)=0A(0) = 0, A(200)=0A(200) = 0, A(100)=100×300=30 000A(100) = 100 \times 300 = 30\,000. The absolute maximum is 30 00030\,000 m2^2, reached for x=100x = 100 m and y=600−300=300y = 600 - 300 = 300 m. The answer the question asks for is the DIMENSIONS and the AREA: a pasture 300300 m along the river and 100100 m deep, split into two pens of 100100 m by 150150 m, total area 30 00030\,000 m2^2. Writing only x=100x = 100 answers a question nobody asked.

c) The domain is now [0,80][0, 80]. The only critical number, 100100, is OUTSIDE it and is not a candidate. On [0,80][0, 80], A′(x)=600−6x≥600−480=120>0A'(x) = 600 - 6x \ge 600 - 480 = 120 > 0, so AA is increasing and its maximum is at the right endpoint: x=80x = 80 m, y=600−240=360y = 600 - 240 = 360 m, A=80×360=28 800A = 80 \times 360 = 28\,800 m2^2. The closed interval method gives the same answer: A(0)=0<A(80)A(0) = 0 < A(80). Reporting the unconstrained answer x=100x = 100 here designs a pasture that does not fit on the land, and costs the whole question. The optimum of a constrained problem is often ON the boundary, with a derivative that is not 00 there.

d) A′′(100)=−6<0A''(100) = -6 < 0 is the second derivative test: it proves that x=100x = 100 is a LOCAL maximum, that is, larger than the nearby values, and nothing about far away values. Two arguments make it global here, and one of them must be written. Either the closed interval method of b), which compares with the endpoints; or the fact that x=100x = 100 is the ONLY critical number on the interval [0,200][0, 200] and is a local maximum, so AA increases before it and decreases after it, and no other value can beat it. A local maximum with no justification of this kind is worth part of the marks only, because the next exercise with a larger endpoint will make the same sentence false.

e) Now (n+1)x+y=L(n + 1)x + y = L, so A(x)=x(L−(n+1)x)=Lx−(n+1)x2A(x) = x(L - (n+1)x) = Lx - (n+1)x^2 on [0,Ln+1]\left[0, \frac{L}{n+1}\right]. A′(x)=L−2(n+1)x=0A'(x) = L - 2(n+1)x = 0 gives x=L2(n+1)x = \frac{L}{2(n+1)}, inside the interval, and AA vanishes at both endpoints, so this is the absolute maximum. Then y=L−(n+1)L2(n+1)=L2y = L - (n+1)\frac{L}{2(n+1)} = \frac{L}{2}. Check with b): L=600L = 600, n=2n = 2 gives x=100x = 100 and y=300y = 300. Whatever the number of pens, half the fence runs along the river side: a result worth remembering to check a numerical answer in two seconds.

Tick the exercises you have done or want to review: a free account, no password, keeps your ticks from one visit to the next and tells you which chapter to tackle next. Create your space, an email is enough.

Exercise 2: The open box cut from a sheet: the domain comes from the geometry

A rectangular sheet of cardboard measures 1616 cm by 1010 cm. A square of side xx cm is cut from each corner and the four flaps are folded up along the dashed lines to make a box WITHOUT a lid. We want the box of largest volume.

The figure shows the sheet, the four squares to remove and the fold lines. Every critical number must be tested against the geometry before it is used.

1610xxbase: (16 - 2x) by (10 - 2x)fold along the dashed linescut out the four corner squares
  • a) Express the length, width and height of the box in terms of xx, then the volume V(x)V(x) expanded, and the domain of xx.
  • b) Compute V′(x)V'(x), factor it, and find the critical numbers. Which one is rejected, and why?
  • c) Use the closed interval method to find the largest volume and the dimensions of that box.
  • d) A student computes V′′(203)>0V''\left(\frac{20}{3}\right) > 0 and writes that the box of MINIMUM volume has x=203x = \frac{20}{3}. Correct this.
  • e) The same construction from a SQUARE sheet of side aa. Prove that the best cut is x=a6x = \frac{a}{6}, find the maximum volume in terms of aa, and apply it to a=18a = 18 cm.

Type your answers, the page tells you right or wrong 0/10

a)
Domain of xx ,
b)
c)
d)
e)
Show the solution

Answers

  • a) (16−2x)(16 - 2x) by (10−2x)(10 - 2x) by xx; V(x)=4x3−52x2+160xV(x) = 4x^3 - 52x^2 + 160x; domain [0,5][0, 5]
  • b) V′(x)=4(3x−20)(x−2)V'(x) = 4(3x - 20)(x - 2); x=2x = 2 kept, x=203x = \frac{20}{3} rejected (>5> 5)
  • c) V(0)=V(5)=0V(0) = V(5) = 0, V(2)=144V(2) = 144: the box 1212 cm by 66 cm by 22 cm, 144144 cm3^3
  • d) 203\frac{20}{3} is outside [0,5][0, 5] (a negative width); the minimum volume is 00, at the endpoints.
  • e) V=x(a−2x)2V = x(a - 2x)^2, V′=(a−2x)(a−6x)V' = (a - 2x)(a - 6x); x=a6x = \frac{a}{6}, Vmax⁡=2a327V_{\max} = \frac{2a^3}{27}; for a=18a = 18: x=3x = 3, V=432V = 432 cm3^3

a) Removing a square of side xx at each end of a side shortens it by 2x2x: the base is (16−2x)(16 - 2x) by (10−2x)(10 - 2x), and the folded flaps give the height xx. So V(x)=x(16−2x)(10−2x)=4x(8−x)(5−x)=4x(40−13x+x2)=4x3−52x2+160xV(x) = x(16 - 2x)(10 - 2x) = 4x(8 - x)(5 - x) = 4x(40 - 13x + x^2) = 4x^3 - 52x^2 + 160x. Domain: every dimension must be ≥0\ge 0, so x≥0x \ge 0, x≤8x \le 8 and x≤5x \le 5. The SHORT side decides: x∈[0,5]x \in [0, 5]. The endpoints are flat boxes of volume 00; they are included to make the interval closed.

b) V′(x)=12x2−104x+160=4(3x2−26x+40)=4(3x−20)(x−2)V'(x) = 12x^2 - 104x + 160 = 4(3x^2 - 26x + 40) = 4(3x - 20)(x - 2), as expanding (3x−20)(x−2)=3x2−26x+40(3x - 20)(x - 2) = 3x^2 - 26x + 40 confirms. The critical numbers are x=2x = 2 and x=203x = \frac{20}{3}. Since 203>5\frac{20}{3} > 5, it is NOT in the domain: cutting squares of side 203≈6.7\frac{20}{3} \approx 6.7 cm from a sheet 1010 cm wide is impossible, the two cuts would overlap. The formula V(x)V(x) knows nothing about cardboard; only the domain does. The candidate x=2x = 2 is kept.

c) Closed interval method on [0,5][0, 5]: V(0)=0V(0) = 0, V(5)=5×6×0=0V(5) = 5 \times 6 \times 0 = 0, V(2)=2×12×6=144V(2) = 2 \times 12 \times 6 = 144. The largest volume is 144144 cm3^3, for the box of base 1212 cm by 66 cm and height 22 cm. Check that VV is continuous on [0,5][0, 5] (a polynomial) before invoking the method: that sentence is the licence, and it is one line. A quick plausibility check: V(1)=1×14×8=112V(1) = 1 \times 14 \times 8 = 112 and V(3)=3×10×4=120V(3) = 3 \times 10 \times 4 = 120, both below 144144.

d) V(203)=203⋅83⋅(−103)=−160027V\left(\frac{20}{3}\right) = \frac{20}{3} \cdot \frac{8}{3} \cdot \left(-\frac{10}{3}\right) = -\frac{1600}{27}, a NEGATIVE volume, which says it all: the point is outside the domain and describes no box. The second derivative test was applied to a point that should never have been tested. The minimum volume on [0,5][0, 5] is 00, at the two endpoints, and it is a degenerate box. The general rule: reject a critical number outside the domain BEFORE any test, and say why in words.

e) V(x)=x(a−2x)2V(x) = x(a - 2x)^2 on [0,a2]\left[0, \frac{a}{2}\right]. By the product and chain rules, V′(x)=(a−2x)2+x⋅2(a−2x)(−2)=(a−2x)[(a−2x)−4x]=(a−2x)(a−6x)V'(x) = (a - 2x)^2 + x \cdot 2(a - 2x)(-2) = (a - 2x)\left[(a - 2x) - 4x\right] = (a - 2x)(a - 6x). Critical numbers: x=a2x = \frac{a}{2} (an endpoint, V=0V = 0) and x=a6x = \frac{a}{6}. With V(0)=V(a2)=0V(0) = V\left(\frac{a}{2}\right) = 0, the maximum is V(a6)=a6(2a3)2=a6⋅4a29=2a327V\left(\frac{a}{6}\right) = \frac{a}{6}\left(\frac{2a}{3}\right)^2 = \frac{a}{6} \cdot \frac{4a^2}{9} = \frac{2a^3}{27}. For a=18a = 18: x=3x = 3 cm and V=2×583227=432V = \frac{2 \times 5832}{27} = 432 cm3^3, which checks directly: 3×12×12=4323 \times 12 \times 12 = 432. Factoring V′V' by the common factor (a−2x)(a - 2x) instead of expanding is the gesture that keeps the numbers small.

Exercise 3: The can of fixed volume: an open domain, and no endpoint to test

A closed cylindrical can must hold 250π250\pi cm3^3. The figure shows the can and its net: two disks of radius rr and a rectangle of width 2πr2\pi r (the circumference) and height hh. We want to use as little metal as possible.

Here the domain of rr is OPEN: no radius is forbidden, but no radius is an endpoint either. The closed interval method is then unavailable, and the global minimum needs another licence.

rhside: 2πr by hhtopbottom
  • a) Write the area of metal AA in terms of rr and hh, the volume constraint, and A(r)A(r) in one variable.
  • b) Give the domain of rr and the behaviour of A(r)A(r) as r→0+r \to 0^+ and as r→∞r \to \infty. Why can the closed interval method not be used?
  • c) Find the critical number and justify that it gives the ABSOLUTE minimum. State the dimensions of the can and the least area of metal.
  • d) The top and bottom are cut from a metal that costs twice as much per cm2^2 as the side. Find the radius of the cheapest can and the ratio hr\frac{h}{r}.
  • e) For a can of any fixed volume VV made of a single metal, prove that the optimal can has h=2rh = 2r.

Type your answers, the page tells you right or wrong 0/8

b)
Domain of rr ,
c)
d)
e)
Show the solution

Answers

  • a) A=2πr2+2πrhA = 2\pi r^2 + 2\pi rh, πr2h=250π\pi r^2 h = 250\pi, A(r)=2πr2+500πrA(r) = 2\pi r^2 + \frac{500\pi}{r}
  • b) (0,∞)(0, \infty); A→∞A \to \infty at both ends; no endpoint to evaluate
  • c) A′<0A' < 0 on (0,5)(0, 5), A′>0A' > 0 on (5,∞)(5, \infty): r=5r = 5 cm, h=10h = 10 cm, A=150πA = 150\pi cm2^2
  • d) r3=1252r^3 = \frac{125}{2}, r=523r = \frac{5}{\sqrt[3]{2}} cm; hr=4\frac{h}{r} = 4
  • e) r3=V2πr^3 = \frac{V}{2\pi} and hr=Vπr3=2\frac{h}{r} = \frac{V}{\pi r^3} = 2

a) The net has two disks of area πr2\pi r^2 and a rectangle of area 2πr⋅h2\pi r \cdot h, so A=2πr2+2πrhA = 2\pi r^2 + 2\pi rh. The constraint is the volume: πr2h=250π\pi r^2 h = 250\pi, so h=250r2h = \frac{250}{r^2}. Substituting, A(r)=2πr2+2πr⋅250r2=2πr2+500πrA(r) = 2\pi r^2 + 2\pi r \cdot \frac{250}{r^2} = 2\pi r^2 + \frac{500\pi}{r}. Solve the constraint for the variable that is easy to isolate, here hh, never for rr, which would bring a square root.

b) Any r>0r > 0 gives a can, with h=250r2>0h = \frac{250}{r^2} > 0, so the domain is (0,∞)(0, \infty): open at 00 and unbounded. As r→0+r \to 0^+, 500πr→∞\frac{500\pi}{r} \to \infty (a tall thin needle with an enormous side); as r→∞r \to \infty, 2πr2→∞2\pi r^2 \to \infty (a huge flat coin). So A(r)→∞A(r) \to \infty at both ends. There is no endpoint at which to evaluate AA, and the Extreme Value Theorem, which requires a closed interval, does not apply: nothing guarantees a minimum in advance. The justification must come from the derivative on the WHOLE domain.

c) A′(r)=4πr−500πr2=4π(r3−125)r2A'(r) = 4\pi r - \frac{500\pi}{r^2} = \frac{4\pi(r^3 - 125)}{r^2}. On (0,∞)(0, \infty) the denominator is positive, so the sign of A′A' is that of r3−125r^3 - 125: negative for 0<r<50 < r < 5, zero at r=5r = 5, positive for r>5r > 5. Hence AA is decreasing on (0,5](0, 5] and increasing on [5,∞)[5, \infty): A(5)A(5) is smaller than EVERY other value, the first derivative test for absolute extreme values. (Equivalently, A′′(r)=4π+1000πr3>0A''(r) = 4\pi + \frac{1000\pi}{r^3} > 0 on the whole domain, so the only critical number is the absolute minimum.) The can: r=5r = 5 cm, h=25025=10h = \frac{250}{25} = 10 cm, and A(5)=50π+100π=150πA(5) = 50\pi + 100\pi = 150\pi cm2^2. Answering the minimum is 55 confuses the radius with the area: the question asked for dimensions and area.

d) The cost is proportional to 2⋅2πr2+2πrh2 \cdot 2\pi r^2 + 2\pi rh (each disk counts twice), so we minimize C(r)=4πr2+500πrC(r) = 4\pi r^2 + \frac{500\pi}{r} on (0,∞)(0, \infty). C′(r)=8πr−500πr2=4π(2r3−125)r2C'(r) = 8\pi r - \frac{500\pi}{r^2} = \frac{4\pi(2r^3 - 125)}{r^2}, negative then positive around r3=1252r^3 = \frac{125}{2}, so the absolute minimum is at r=523=5432r = \frac{5}{\sqrt[3]{2}} = \frac{5\sqrt[3]{4}}{2} cm. For the shape, use the constraint instead of computing hh: hr=250r3=250125/2=4\frac{h}{r} = \frac{250}{r^3} = \frac{250}{125/2} = 4. The cheap can is twice as tall, relative to its radius, as the can of c): a dearer lid pushes the design toward a smaller lid.

e) With h=Vπr2h = \frac{V}{\pi r^2}, A(r)=2πr2+2VrA(r) = 2\pi r^2 + \frac{2V}{r} and A′(r)=4πr−2Vr2=4πr3−2Vr2A'(r) = 4\pi r - \frac{2V}{r^2} = \frac{4\pi r^3 - 2V}{r^2}, negative before and positive after r3=V2πr^3 = \frac{V}{2\pi}: the absolute minimum. Then hr=Vπr3=Vπ⋅2πV=2\frac{h}{r} = \frac{V}{\pi r^3} = \frac{V}{\pi} \cdot \frac{2\pi}{V} = 2, so h=2rh = 2r: the height equals the diameter. c) is the case V=250πV = 250\pi: h=10=2×5h = 10 = 2 \times 5. Real cans are taller than this because the lid is thicker metal and the seams cost too, which is exactly the effect measured in d).

Exercise 4: The closest point on a curve: minimize the square of the distance

The distance from A(a,b)A(a, b) to a point P(x,y)P(x, y) is d=(x−a)2+(y−b)2d = \sqrt{(x - a)^2 + (y - b)^2}. Differentiating a square root is legal but heavy. Since t↦tt \mapsto \sqrt t is increasing on [0,∞)[0, \infty), the distance dd and its square D=d2D = d^2 are smallest at the SAME point: we minimize DD, and take the square root only at the very end, for the value.

The figure shows the curve y=xy = \sqrt x, the point A(3,0)A(3, 0) and a point P(x,x)P(x, \sqrt x) of the curve.

123456123y = √xA(3, 0)P(x, √x)dx
  • a) Find the point of the curve y=xy = \sqrt x closest to A(3,0)A(3, 0), justify that the minimum is absolute, and give the least distance.
  • b) Same curve, point B(14,0)B\left(\frac{1}{4}, 0\right). A student solves D′(x)=0D'(x) = 0 and finds x=−14x = -\frac{1}{4}. Find the closest point.
  • c) For the point (a,0)(a, 0) with a>0a > 0, find the closest point of the curve, depending on aa.
  • d) Find the point of the branch xy=4xy = 4, x>0x > 0, closest to the origin, and the least distance.
  • e) In a), check that the segment APAP is perpendicular to the tangent line at the closest point PP, and explain why a student who answers 114\frac{11}{4} for the distance loses a mark.

Type your answers, the page tells you right or wrong 0/11

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) D(x)=x2−5x+9D(x) = x^2 - 5x + 9; P(52,52)P\left(\frac{5}{2}, \sqrt{\frac{5}{2}}\right), dmin⁡=112d_{\min} = \frac{\sqrt{11}}{2}
  • b) D′=2x+12>0D' = 2x + \frac{1}{2} > 0 on [0,∞)[0, \infty): the origin, at distance 14\frac{1}{4} (x=−14x = -\frac{1}{4} rejected)
  • c) x=a−12x = a - \frac{1}{2} if a≥12a \ge \frac{1}{2}; the origin if 0<a<120 < a < \frac{1}{2}
  • d) (2,2)(2, 2), at distance 222\sqrt 2
  • e) Slopes 125/2\frac{1}{2\sqrt{5/2}} and −252-2\sqrt{\frac{5}{2}}, product −1-1; 114\frac{11}{4} is DD, the distance is 112\frac{\sqrt{11}}{2}.

a) D(x)=(x−3)2+(x−0)2=x2−6x+9+x=x2−5x+9D(x) = (x - 3)^2 + (\sqrt x - 0)^2 = x^2 - 6x + 9 + x = x^2 - 5x + 9, on the domain [0,∞)[0, \infty) of the curve. D′(x)=2x−5D'(x) = 2x - 5, zero at x=52x = \frac{5}{2}, inside the domain. The domain is not closed and bounded, so justify on the WHOLE domain: D′<0D' < 0 on [0,52)\left[0, \frac{5}{2}\right) and D′>0D' > 0 on (52,∞)\left(\frac{5}{2}, \infty\right), so DD decreases then increases and D(52)D\left(\frac{5}{2}\right) is its absolute minimum. D(52)=254−252+9=25−50+364=114D\left(\frac{5}{2}\right) = \frac{25}{4} - \frac{25}{2} + 9 = \frac{25 - 50 + 36}{4} = \frac{11}{4}. The closest point is P(52,52)=(52,102)P\left(\frac{5}{2}, \sqrt{\frac{5}{2}}\right) = \left(\frac{5}{2}, \frac{\sqrt{10}}{2}\right), and the least distance is 114=112\sqrt{\frac{11}{4}} = \frac{\sqrt{11}}{2}. The endpoint x=0x = 0 gives D=9D = 9, larger, as it must.

b) D(x)=(x−14)2+x=x2+x2+116D(x) = \left(x - \frac{1}{4}\right)^2 + x = x^2 + \frac{x}{2} + \frac{1}{16} and D′(x)=2x+12D'(x) = 2x + \frac{1}{2}. The root x=−14x = -\frac{1}{4} is NOT in the domain [0,∞)[0, \infty): the curve y=xy = \sqrt x has no point there. On the domain, D′(x)≥12>0D'(x) \ge \frac{1}{2} > 0, so DD is increasing and its minimum is at the endpoint x=0x = 0: the closest point is the origin, at distance D(0)=116=14\sqrt{D(0)} = \sqrt{\frac{1}{16}} = \frac{1}{4}. The student's equation had a solution and the problem still had its answer elsewhere, at an endpoint where D′≠0D' \ne 0.

c) D(x)=(x−a)2+xD(x) = (x - a)^2 + x, D′(x)=2x−2a+1D'(x) = 2x - 2a + 1, zero at x=a−12x = a - \frac{1}{2}. If a≥12a \ge \frac{1}{2}, this is in [0,∞)[0, \infty), D′D' changes from negative to positive there, and the closest point has x=a−12x = a - \frac{1}{2} (for a=3a = 3, x=52x = \frac{5}{2}, as in a)). If 0<a<120 < a < \frac{1}{2}, the root is negative, D′>0D' > 0 on the domain, and the closest point is the origin (b) is a=14a = \frac{1}{4}). Same method, two answers, and the switch happens exactly when the critical number leaves the domain.

d) On the branch, y=4xy = \frac{4}{x}, so D(x)=x2+16x2D(x) = x^2 + \frac{16}{x^2} on (0,∞)(0, \infty), an OPEN domain. D′(x)=2x−32x3=2(x4−16)x3D'(x) = 2x - \frac{32}{x^3} = \frac{2(x^4 - 16)}{x^3}, negative for 0<x<20 < x < 2 and positive for x>2x > 2. So x=2x = 2 gives the absolute minimum: the point (2,2)(2, 2), with D(2)=4+4=8D(2) = 4 + 4 = 8 and distance 8=22\sqrt 8 = 2\sqrt 2. The symmetry of the curve about y=xy = x predicted the point; the sign of D′D' is what proves it. Note that D→∞D \to \infty at both ends of the domain, as for the can of Exercise 3.

e) At x=52x = \frac{5}{2} the tangent slope is 12x=125/2\frac{1}{2\sqrt x} = \frac{1}{2\sqrt{5/2}}. The slope of APAP is 5/2−052−3=−252\frac{\sqrt{5/2} - 0}{\frac{5}{2} - 3} = -2\sqrt{\frac{5}{2}}. Their product is −25/225/2=−1-\frac{2\sqrt{5/2}}{2\sqrt{5/2}} = -1: the shortest segment from a point to a smooth curve meets it at a right angle, a two second check of any closest point answer. As for 114\frac{11}{4}: it is the minimum of DD, the SQUARE of the distance. The trick of minimizing DD changes the function, not the point; the value asked must be brought back through the square root.

Exercise 5: The rectangle inscribed in a semicircle: area, then perimeter

A rectangle has its base on the diameter of a semicircle of radius 55 and its two upper corners on the arc, as in the figure. By symmetry, its upper right corner is (x,y)(x, y) with x2+y2=25x^2 + y^2 = 25, y≥0y \ge 0, so the rectangle is 2x2x wide and y=25−x2y = \sqrt{25 - x^2} high.

The same figure answers two different questions, and the two answers differ: an optimum belongs to the objective function, not to the figure.

(x, y)x5y
  • a) Write the area A(x)A(x) of the rectangle and the domain of xx.
  • b) Find the rectangle of largest area, justifying that the maximum is absolute. Where does A′A' fail to exist, and does it matter?
  • c) Redo b) by maximizing A(x)2A(x)^2 instead of A(x)A(x). Why is this legitimate?
  • d) Find the rectangle of largest PERIMETER inscribed in the same semicircle. Compare with the endpoints exactly, without a calculator.
  • e) Compare the shapes of the two optimal rectangles of b) and d).

Type your answers, the page tells you right or wrong 0/10

a)
Domain of xx ,
b)
c)
d)
e)
Show the solution

Answers

  • a) A(x)=2x25−x2A(x) = 2x\sqrt{25 - x^2} on [0,5][0, 5]
  • b) x=52x = \frac{5}{\sqrt 2}: base 525\sqrt 2, height 522\frac{5\sqrt 2}{2}, area 2525; A′A' undefined at the endpoint 55 only
  • c) A2=4(25u−u2)A^2 = 4(25u - u^2) with u=x2u = x^2: u=252u = \frac{25}{2}, same rectangle
  • d) x=25x = 2\sqrt 5, y=5y = \sqrt 5, P=105P = 10\sqrt 5, larger than P(0)=10P(0) = 10 and P(5)=20P(5) = 20
  • e) Base twice the height for the area, four times the height for the perimeter.

a) The rectangle is 2x2x by y=25−x2y = \sqrt{25 - x^2}, so A(x)=2x25−x2A(x) = 2x\sqrt{25 - x^2}. The corner must lie on the arc in the first quadrant: 0≤x≤50 \le x \le 5. At x=0x = 0 the rectangle is a vertical segment, at x=5x = 5 a horizontal one, both of area 00. AA is continuous on the closed interval [0,5][0, 5].

b) Product rule, and chain rule with inner function 25−x225 - x^2: A′(x)=225−x2+2x⋅−2x225−x2=2(25−x2)−2x225−x2=50−4x225−x2A'(x) = 2\sqrt{25 - x^2} + 2x \cdot \frac{-2x}{2\sqrt{25 - x^2}} = \frac{2(25 - x^2) - 2x^2}{\sqrt{25 - x^2}} = \frac{50 - 4x^2}{\sqrt{25 - x^2}} for 0≤x<50 \le x < 5. A′(x)=0A'(x) = 0 gives x2=252x^2 = \frac{25}{2}, x=52x = \frac{5}{\sqrt 2} (the negative root is outside the domain). A′A' does not exist at x=5x = 5: that point is a critical number in the wide sense, but it is an endpoint, already on the list of the closed interval method. Values: A(0)=0A(0) = 0, A(5)=0A(5) = 0, A(52)=2⋅52⋅252=2⋅252=25A\left(\frac{5}{\sqrt 2}\right) = 2 \cdot \frac{5}{\sqrt 2} \cdot \sqrt{\frac{25}{2}} = 2 \cdot \frac{25}{2} = 25. The largest rectangle has base 2x=522x = 5\sqrt 2, height 52=522\frac{5}{\sqrt 2} = \frac{5\sqrt 2}{2} and area 2525, the square of the radius.

c) A(x)≥0A(x) \ge 0 on [0,5][0, 5], and t↦t2t \mapsto t^2 is increasing on [0,∞)[0, \infty), so AA and A2A^2 are largest at the same xx. With u=x2∈[0,25]u = x^2 \in [0, 25]: A2=4x2(25−x2)=4(25u−u2)A^2 = 4x^2(25 - x^2) = 4(25u - u^2), a downward parabola in uu with vertex u=252u = \frac{25}{2}, inside [0,25][0, 25]. So x=52x = \frac{5}{\sqrt 2} again, with no square root to differentiate. The same trick as the distance in Exercise 4, with the same warning: the maximum of A2A^2 is 625625, the area is 625=25\sqrt{625} = 25. It is legitimate only because A≥0A \ge 0: squaring a quantity that changes sign can move its maximum.

d) The perimeter is P(x)=2(2x)+2y=4x+225−x2P(x) = 2(2x) + 2y = 4x + 2\sqrt{25 - x^2} on [0,5][0, 5]. P′(x)=4−2x25−x2P'(x) = 4 - \frac{2x}{\sqrt{25 - x^2}}, zero when 225−x2=x2\sqrt{25 - x^2} = x. Both sides are ≥0\ge 0 on the domain, so squaring is safe: 4(25−x2)=x24(25 - x^2) = x^2, x2=20x^2 = 20, x=25x = 2\sqrt 5, and then y=25−20=5y = \sqrt{25 - 20} = \sqrt 5. Closed interval method: P(0)=10P(0) = 10 (the vertical segment, counted twice), P(5)=20P(5) = 20 (the diameter, counted twice) and P(25)=85+25=105P(2\sqrt 5) = 8\sqrt 5 + 2\sqrt 5 = 10\sqrt 5. Exact comparison: 105>2010\sqrt 5 > 20 because 5>2\sqrt 5 > 2, since 5>45 > 4. The maximum perimeter is 10510\sqrt 5, for the rectangle 454\sqrt 5 by 5\sqrt 5. Here an endpoint value is NOT zero, and skipping the comparison would leave the answer unproved.

e) For the area, base 525\sqrt 2 and height 522\frac{5\sqrt 2}{2}: the base is TWICE the height, so the rectangle is two squares side by side. For the perimeter, base 454\sqrt 5 and height 5\sqrt 5: the base is FOUR times the height, a flatter rectangle, because the perimeter rewards width more than the area does. The same figure, two objective functions, two answers: the first line of every solution must say which quantity is optimized.

Part B: problems and reasoning (/50)

Exercise 6: The Norman window: an endpoint whose value is not zero

A Norman window is a rectangle surmounted by a half-disk whose diameter is the top side of the rectangle. Its frame, the whole outline of the window, measures 88 m. With the notation of the figure, the rectangle is 2r2r wide and hh high, and the half-disk has radius rr. We want to let in as much light as possible.

In the previous exercises the endpoints of the domain were degenerate figures of area 00. Here one of them is a perfectly good window, and its value must be compared with the candidate, exactly and without a calculator.

2rhrclear glasshalf-disk
  • a) Write the constraint on the perimeter, the area AA of the window as a function of rr alone, and the domain of rr.
  • b) Find the critical number, then apply the closed interval method. Compare the values exactly.
  • c) Show that, at the optimum, the height of the rectangle equals the radius, and give the largest area.
  • d) The half-disk is now made of stained glass, which lets through only half as much light per m2^2 as clear glass. Find the radius that maximizes the light, justifying the absolute maximum, and the ratio hr\frac{h}{r}.
  • e) Compare the two optimal windows of c) and d), and explain the difference in one sentence.

Type your answers, the page tells you right or wrong 0/7

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) 2h+2r+πr=82h + 2r + \pi r = 8; A(r)=8r−4+π2r2A(r) = 8r - \frac{4 + \pi}{2}r^2 on [0,82+π]\left[0, \frac{8}{2 + \pi}\right]
  • b) r=84+πr = \frac{8}{4 + \pi}; A(0)=0A(0) = 0, A(82+π)=32π(2+π)2<324+πA\left(\frac{8}{2+\pi}\right) = \frac{32\pi}{(2+\pi)^2} < \frac{32}{4 + \pi}
  • c) h=r=84+πh = r = \frac{8}{4 + \pi} m, Amax⁡=324+πA_{\max} = \frac{32}{4 + \pi} m2^2
  • d) r=168+3πr = \frac{16}{8 + 3\pi} m (vertex of a downward parabola, inside the domain); hr=4+π4\frac{h}{r} = \frac{4 + \pi}{4}
  • e) The stained glass window is taller (h>rh > r): light is moved from the dim half-disk to the clear rectangle.

a) The outline is two vertical sides hh, the bottom 2r2r and the half-circle πr\pi r: 2h+2r+πr=82h + 2r + \pi r = 8, so h=8−(2+π)r2h = \frac{8 - (2 + \pi)r}{2}. The area is A=2rh+πr22A = 2rh + \frac{\pi r^2}{2}, and substituting, A(r)=r(8−(2+π)r)+πr22=8r−2r2−πr2+πr22=8r−4+π2r2A(r) = r\left(8 - (2 + \pi)r\right) + \frac{\pi r^2}{2} = 8r - 2r^2 - \pi r^2 + \frac{\pi r^2}{2} = 8r - \frac{4 + \pi}{2}r^2. Domain: r≥0r \ge 0 and h≥0h \ge 0, that is (2+π)r≤8(2 + \pi)r \le 8, so r∈[0,82+π]r \in \left[0, \frac{8}{2 + \pi}\right]. At the right endpoint, h=0h = 0: the window is a half-disk alone, a real window with a real area.

b) A′(r)=8−(4+π)r=0A'(r) = 8 - (4 + \pi)r = 0 for r=84+πr = \frac{8}{4 + \pi}, which is in the domain since 4+π>2+π4 + \pi > 2 + \pi. Values: A(0)=0A(0) = 0; at the candidate, A=644+π−4+π2⋅64(4+π)2=644+π−324+π=324+πA = \frac{64}{4 + \pi} - \frac{4 + \pi}{2} \cdot \frac{64}{(4 + \pi)^2} = \frac{64}{4 + \pi} - \frac{32}{4 + \pi} = \frac{32}{4 + \pi}; at the right endpoint, only the half-disk remains: A=π2(82+π)2=32π(2+π)2A = \frac{\pi}{2}\left(\frac{8}{2 + \pi}\right)^2 = \frac{32\pi}{(2 + \pi)^2}. Exact comparison: 324+π>32π(2+π)2\frac{32}{4 + \pi} > \frac{32\pi}{(2 + \pi)^2} is equivalent to (2+π)2>π(4+π)(2 + \pi)^2 > \pi(4 + \pi), that is 4+4π+π2>4π+π24 + 4\pi + \pi^2 > 4\pi + \pi^2, that is 4>04 > 0. True. The candidate wins, and the comparison took one line of algebra instead of a calculator.

c) With r=84+πr = \frac{8}{4 + \pi}: h=8−(2+π)84+π2=82⋅(4+π)−(2+π)4+π=84+π=rh = \frac{8 - (2 + \pi)\frac{8}{4 + \pi}}{2} = \frac{8}{2} \cdot \frac{(4 + \pi) - (2 + \pi)}{4 + \pi} = \frac{8}{4 + \pi} = r. The height of the rectangle equals the radius of the half-disk, whatever the perimeter (the 88 factors out). The largest area is 324+π\frac{32}{4 + \pi} m2^2, about 4.54.5 m2^2 using π≈3.14\pi \approx 3.14, for a window 164+π\frac{16}{4 + \pi} m wide. Stating h=rh = r is a check worth writing: a result this clean rarely comes out of a wrong computation.

d) The light is proportional to L=2rh+12⋅πr22=r(8−(2+π)r)+πr24=8r−(2+3π4)r2L = 2rh + \frac{1}{2} \cdot \frac{\pi r^2}{2} = r\left(8 - (2 + \pi)r\right) + \frac{\pi r^2}{4} = 8r - \left(2 + \frac{3\pi}{4}\right)r^2, on the same domain. L′(r)=8−(4+3π2)r=0L'(r) = 8 - \left(4 + \frac{3\pi}{2}\right)r = 0 for r=84+3π/2=168+3πr = \frac{8}{4 + 3\pi/2} = \frac{16}{8 + 3\pi}. Justification by concavity on the WHOLE domain: L′′(r)=−(4+3π2)<0L''(r) = -\left(4 + \frac{3\pi}{2}\right) < 0 everywhere, so LL is a downward parabola and its only critical number is the absolute maximum on any interval that contains it; and it is in the domain, since 168+3π≤82+π\frac{16}{8 + 3\pi} \le \frac{8}{2 + \pi} amounts to 32+16π≤64+24π32 + 16\pi \le 64 + 24\pi. Then h=8(8+3π)−16(2+π)2(8+3π)=16+4π8+3πh = \frac{8(8 + 3\pi) - 16(2 + \pi)}{2(8 + 3\pi)} = \frac{16 + 4\pi}{8 + 3\pi}, and hr=16+4π16=4+π4\frac{h}{r} = \frac{16 + 4\pi}{16} = \frac{4 + \pi}{4}.

e) In c), hr=1\frac{h}{r} = 1; in d), hr=1+π4>1\frac{h}{r} = 1 + \frac{\pi}{4} > 1, and the radius 168+3π\frac{16}{8 + 3\pi} is smaller than 84+π\frac{8}{4 + \pi} (compare 16(4+π)=64+16π16(4 + \pi) = 64 + 16\pi with 8(8+3π)=64+24π8(8 + 3\pi) = 64 + 24\pi). Since the half-disk now pays full price in frame but returns only half the light, the optimum shrinks it and spends the frame on a taller clear rectangle. Changing the objective moved the optimum, as in Exercise 5, and the method, candidate then global justification, did not change at all.

Exercise 7: Row, then walk: the fastest route, and when the endpoint wins

A woman is in a rowboat at AA, 33 km from the nearest point CC of a straight shore. She wants to reach the village BB, on the shore 88 km from CC, as fast as possible. She rows at 33 km/h and walks at 55 km/h. She will land at a point PP between CC and BB, at distance xx km from CC, as in the figure.

Time is distance over speed, leg by leg. The domain is a closed interval, and both endpoints are genuine routes: row straight to CC, or row all the way to BB.

A3 kmCxPB8 kmrow at 3 km/hwalk at 5 km/h
  • a) Write the total time T(x)T(x) in hours and its domain.
  • b) Find the critical number of TT, justifying each step of the resolution.
  • c) Apply the closed interval method, comparing the three values exactly, and describe the fastest route and its duration.
  • d) Same boat, same speeds, but the village is only 22 km from CC. Find the fastest route.
  • e) Show that the landing point does not depend on the distance CBCB as long as CB≥94CB \ge \frac{9}{4} km, and express the optimality condition with the angle θ=∠CAP\theta = \angle CAP.

Type your answers, the page tells you right or wrong 0/8

a)
Domain of xx ,
b)
c)
d)
e)
Show the solution

Answers

  • a) T(x)=9+x23+8−x5T(x) = \frac{\sqrt{9 + x^2}}{3} + \frac{8 - x}{5} on [0,8][0, 8]
  • b) T′(x)=x39+x2−15=0T'(x) = \frac{x}{3\sqrt{9 + x^2}} - \frac{1}{5} = 0 gives x=94x = \frac{9}{4}
  • c) T(0)=135T(0) = \frac{13}{5}, T(94)=125T\left(\frac{9}{4}\right) = \frac{12}{5}, T(8)=733T(8) = \frac{\sqrt{73}}{3}: land 94\frac{9}{4} km from CC, total 125\frac{12}{5} h =2= 2 h 2424 min
  • d) T′<0T' < 0 on [0,2][0, 2]: row straight to the village, T(2)=133T(2) = \frac{\sqrt{13}}{3} h
  • e) x=94x = \frac{9}{4} comes from x9+x2=35\frac{x}{\sqrt{9 + x^2}} = \frac{3}{5}, free of CBCB; that is sin⁡θ=35\sin\theta = \frac{3}{5}, the ratio of the speeds.

a) Rowing covers AP=9+x2AP = \sqrt{9 + x^2} km (Pythagoras in the right triangle ACPACP) at 33 km/h, and walking covers PB=8−xPB = 8 - x km at 55 km/h. So T(x)=9+x23+8−x5T(x) = \frac{\sqrt{9 + x^2}}{3} + \frac{8 - x}{5}, for x∈[0,8]x \in [0, 8]. Landing beyond BB or before CC only adds distance, so nothing is lost by this domain.

b) Chain rule, inner function 9+x29 + x^2: T′(x)=13⋅2x29+x2−15=x39+x2−15T'(x) = \frac{1}{3} \cdot \frac{2x}{2\sqrt{9 + x^2}} - \frac{1}{5} = \frac{x}{3\sqrt{9 + x^2}} - \frac{1}{5}. T′(x)=0T'(x) = 0 means 5x=39+x25x = 3\sqrt{9 + x^2}. Both sides are ≥0\ge 0 for x≥0x \ge 0, so squaring creates no false root on the domain: 25x2=81+9x225x^2 = 81 + 9x^2, 16x2=8116x^2 = 81, x=94x = \frac{9}{4} (−94-\frac{9}{4} is outside the domain, and would not even satisfy the unsquared equation). T′T' exists everywhere on [0,8][0, 8].

c) T(0)=33+85=135T(0) = \frac{3}{3} + \frac{8}{5} = \frac{13}{5}. At x=94x = \frac{9}{4}: 9+8116=225169 + \frac{81}{16} = \frac{225}{16}, so AP=154AP = \frac{15}{4} and T=1512+2320=2520+2320=125T = \frac{15}{12} + \frac{23}{20} = \frac{25}{20} + \frac{23}{20} = \frac{12}{5}. T(8)=733T(8) = \frac{\sqrt{73}}{3}. Exact comparisons: 125<135\frac{12}{5} < \frac{13}{5} is clear, and 125<733\frac{12}{5} < \frac{\sqrt{73}}{3} amounts to 36<57336 < 5\sqrt{73}, that is 1296<18251296 < 1825 after squaring positive numbers. The fastest route: row to the point 94\frac{9}{4} km from CC, then walk the remaining 234\frac{23}{4} km, in 125\frac{12}{5} h, that is 22 h 2424 min. A decimal like T(8)≈2.85T(8) \approx 2.85 is fine as a comment, never as the comparison itself.

d) Now T(x)=9+x23+2−x5T(x) = \frac{\sqrt{9 + x^2}}{3} + \frac{2 - x}{5} on [0,2][0, 2], with the same derivative. Its only zero, 94\frac{9}{4}, is outside [0,2][0, 2]. Since x9+x2\frac{x}{\sqrt{9 + x^2}} increases with xx, T′(x)<T′(94)=0T'(x) < T'\left(\frac{9}{4}\right) = 0 on [0,2][0, 2]: TT is decreasing and the minimum is at the endpoint x=2x = 2. She rows straight to the village, in T(2)=133T(2) = \frac{\sqrt{13}}{3} h. Check against the other endpoint: T(0)=1+25=75T(0) = 1 + \frac{2}{5} = \frac{7}{5}, and 133<75\frac{\sqrt{13}}{3} < \frac{7}{5} amounts to 513<215\sqrt{13} < 21, that is 325<441325 < 441. Solving T′=0T' = 0 and answering x=94x = \frac{9}{4} would send her past the village.

e) The distance CBCB appears in TT only through the term −x5-\frac{x}{5} plus a constant, so it disappears from T′T': the equation x39+x2=15\frac{x}{3\sqrt{9 + x^2}} = \frac{1}{5} does not involve it. As long as the root 94\frac{9}{4} lies in [0,CB][0, CB], it is the landing point. In the triangle ACPACP, sin⁡θ=CPAP=x9+x2\sin\theta = \frac{CP}{AP} = \frac{x}{\sqrt{9 + x^2}}, and the condition reads sin⁡θ=35\sin\theta = \frac{3}{5}, the ratio of the rowing speed to the walking speed: check with x=94x = \frac{9}{4}, 9/415/4=35\frac{9/4}{15/4} = \frac{3}{5}. Light obeys the same rule when it crosses from air to water, which is why the fastest path bends.

Exercise 8: Five statements to correct

Each statement below was written on a MATH 140 assignment, and each is false. Say what is wrong, give a counterexample, and write the correct statement.

  • a) Once A′(x)=0A'(x) = 0 is solved, the problem is finished: in the pasture problem, the maximum area is x=100x = 100.
  • b) A local maximum of the objective function is its absolute maximum.
  • c) Minimizing the square of a distance may give a different point from minimizing the distance itself.
  • d) On a closed interval, the absolute maximum of a differentiable function is at a point where its derivative is 00.
  • e) Every optimization problem has an answer, since a continuous function always reaches a maximum and a minimum.

Type your answers, the page tells you right or wrong 0/7

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) False: x=100x = 100 is a candidate and a length. Justify the global maximum, then answer 300300 m by 100100 m, 30 00030\,000 m2^2.
  • b) False: x3−6x2+9x+1x^3 - 6x^2 + 9x + 1 on [0,5][0, 5] has a local max 55 at x=1x = 1 and f(5)=21f(5) = 21. True if it is the ONLY critical number of an interval.
  • c) False: t\sqrt t is increasing, so dd and d2d^2 have the same minimizer; only the VALUE changes.
  • d) False: with the cap x≤80x \le 80, the maximum is at x=80x = 80 where A′=120A' = 120. The maximum is at a critical number OR an endpoint.
  • e) False: the can of fixed volume has no MAXIMUM area, A→∞A \to \infty. The Extreme Value Theorem needs a CLOSED interval.

a) FALSE, twice. First, A′(x)=0A'(x) = 0 only produces a candidate: without the closed interval method or a sign argument on the whole domain, nothing says it is the maximum, and the next problem will have a larger endpoint. Second, x=100x = 100 is a length, not an area, and the question asked for the dimensions and the largest area. Correct statement: the critical number is a candidate; after justifying that it gives the absolute maximum, the answer is the quantity asked, here a pasture of 300300 m by 100100 m of area 30 00030\,000 m2^2.

b) FALSE. f(x)=x3−6x2+9x+1f(x) = x^3 - 6x^2 + 9x + 1 on [0,5][0, 5]: f′(x)=3(x−1)(x−3)f'(x) = 3(x - 1)(x - 3), a local maximum f(1)=5f(1) = 5, a local minimum f(3)=1f(3) = 1, and yet f(5)=125−150+45+1=21>5f(5) = 125 - 150 + 45 + 1 = 21 > 5. A local maximum is only larger than its NEIGHBOURS. Correct statement: if cc is the ONLY critical number of ff on an interval and ff has a local maximum at cc, then f(c)f(c) is the absolute maximum on that interval (with two critical numbers, as here, the endpoints must be compared).

c) FALSE. For D=d2≥0D = d^2 \ge 0, d=Dd = \sqrt D, and t↦tt \mapsto \sqrt t is increasing on [0,∞)[0, \infty): whenever D(x1)<D(x2)D(x_1) < D(x_2), d(x1)<d(x2)d(x_1) < d(x_2). So dd and DD are smallest at the same point. What differs is the VALUE: in Exercise 4 a), min⁡D=114\min D = \frac{11}{4} while the least distance is 112\frac{\sqrt{11}}{2}. Correct statement: minimizing d2d^2 gives the same point as minimizing dd; the least distance is the square root of the least value of d2d^2. The same holds for A2A^2 when A≥0A \ge 0.

d) FALSE. In the pasture with the land cap x≤80x \le 80, A(x)=600x−3x2A(x) = 600x - 3x^2 is largest at x=80x = 80, where A′(80)=120≠0A'(80) = 120 \ne 0. A maximum at an endpoint has no reason to have a horizontal tangent; f(x)=xf(x) = x on [0,1][0, 1] is the simplest example. Correct statement: on a closed interval, a continuous function reaches its absolute maximum at a critical number (where f′=0f' = 0 or f′f' does not exist) OR at an endpoint, which is why the closed interval method lists both.

e) FALSE. The Extreme Value Theorem needs a continuous function on a CLOSED and bounded interval. For the can of Exercise 3, A(r)=2πr2+500πrA(r) = 2\pi r^2 + \frac{500\pi}{r} on the open domain (0,∞)(0, \infty) tends to ∞\infty at both ends: the can using the MOST metal does not exist, whatever the calculus says. Correct statement: on a closed interval the optimum exists; on an open or unbounded domain it must be proved, by the sign of the derivative on the whole domain or by the behaviour at the ends, and it may fail to exist, in which case that is the answer.

Exercise 9: Pricing concert tickets: revenue, a sold-out hall, and profit

A concert organizer surveys demand. At 5050 dollars a ticket, 500500 tickets would sell, and each 11 dollar cut in the price sells 2020 more tickets (each 11 dollar increase loses 2020). The figure shows the resulting demand line and the capacity of the first hall considered, 600600 seats.

Price against demand is the classic trade-off: a higher price earns more per ticket and sells fewer tickets. The revenue is R=pqR = pq, price times quantity sold.

10203040506070802004006008001000120014001600q = 1500 - 20p(50, 500)capacity: 600 seatsp (dollars)q (tickets)
  • a) Show that the number of tickets sold at price pp is q=1500−20pq = 1500 - 20p, and give the domain of pp for which the model makes sense.
  • b) In a very large hall, find the price that maximizes the revenue, the number of tickets sold and the revenue, with a justification of the absolute maximum.
  • c) In the 600600-seat hall, find the price that maximizes the revenue, and the revenue.
  • d) In a 10001000-seat hall, each spectator costs 1010 dollars (staff, security) and the evening costs 50005000 dollars in fixed charges. Find the price that maximizes the PROFIT, and the profit.
  • e) Redo d) with qq as the variable, and interpret the condition you obtain.

Type your answers, the page tells you right or wrong 0/10

a)
Domain of pp ,
b)
c)
d)
e)
Show the solution

Answers

  • a) q=500+20(50−p)=1500−20pq = 500 + 20(50 - p) = 1500 - 20p; 0≤p≤750 \le p \le 75
  • b) R(p)=1500p−20p2R(p) = 1500p - 20p^2; p=37.5p = 37.5 dollars, q=750q = 750, R=28 125R = 28\,125 dollars
  • c) q≤600q \le 600 gives p≥45p \ge 45; R′<0R' < 0 on [45,75][45, 75]: p=45p = 45, sold out, R=27 000R = 27\,000 dollars
  • d) Π(p)=(p−10)(1500−20p)−5000\Pi(p) = (p - 10)(1500 - 20p) - 5000 on [25,75][25, 75]; p=42.5p = 42.5 dollars, q=650q = 650, Π=16 125\Pi = 16\,125 dollars
  • e) R′(q)=75−q10=10=C′(q)R'(q) = 75 - \frac{q}{10} = 10 = C'(q): q=650q = 650; stop where the last ticket brings in exactly what it costs.

a) Lowering the price from 5050 to pp is a cut of 50−p50 - p dollars, which sells 20(50−p)20(50 - p) extra tickets: q=500+20(50−p)=1500−20pq = 500 + 20(50 - p) = 1500 - 20p. The formula also covers increases (p>50p > 50 gives 50−p<050 - p < 0). Check: p=50p = 50 gives q=500q = 500, the survey point on the figure. The model needs p≥0p \ge 0 and q≥0q \ge 0, that is p≤75p \le 75: domain [0,75][0, 75].

b) R(p)=p(1500−20p)=1500p−20p2R(p) = p(1500 - 20p) = 1500p - 20p^2, continuous on [0,75][0, 75]. R′(p)=1500−40p=0R'(p) = 1500 - 40p = 0 for p=37.5p = 37.5. Closed interval method: R(0)=0R(0) = 0 (free tickets), R(75)=0R(75) = 0 (nobody comes), R(37.5)=37.5×750=28 125R(37.5) = 37.5 \times 750 = 28\,125. The best price is 37.5037.50 dollars, which sells 750750 tickets for a revenue of 28 12528\,125 dollars. Note that the answer is a price, a number of tickets and an amount of money: three quantities, all asked.

c) The hall adds a constraint: q≤600q \le 600, that is 1500−20p≤6001500 - 20p \le 600, p≥45p \ge 45. The domain becomes [45,75][45, 75], and the candidate 37.537.5 is outside it. On [45,75][45, 75], R′(p)=1500−40p≤1500−1800=−300<0R'(p) = 1500 - 40p \le 1500 - 1800 = -300 < 0, so RR is decreasing and its maximum is at the left endpoint p=45p = 45: exactly 600600 tickets, a sold-out hall, and R=45×600=27 000R = 45 \times 600 = 27\,000 dollars. The solution figure shows the parabola cut by the capacity line. Charging 37.5037.50 dollars here would sell only 600600 of the 750750 tickets demanded, for 22 50022\,500 dollars: the unconstrained answer is not just suboptimal, it is 45004500 dollars worse.

d) The capacity now requires 1500−20p≤10001500 - 20p \le 1000, that is p≥25p \ge 25, so the domain is [25,75][25, 75]. Profit is revenue minus cost: Π(p)=p(1500−20p)−10(1500−20p)−5000=(p−10)(1500−20p)−5000\Pi(p) = p(1500 - 20p) - 10(1500 - 20p) - 5000 = (p - 10)(1500 - 20p) - 5000. Product rule: Π′(p)=(1500−20p)+(p−10)(−20)=1700−40p\Pi'(p) = (1500 - 20p) + (p - 10)(-20) = 1700 - 40p, zero for p=42.5p = 42.5, inside [25,75][25, 75]. Values: Π(25)=15×1000−5000=10 000\Pi(25) = 15 \times 1000 - 5000 = 10\,000, Π(75)=65×0−5000=−5000\Pi(75) = 65 \times 0 - 5000 = -5000, Π(42.5)=32.5×650−5000=21 125−5000=16 125\Pi(42.5) = 32.5 \times 650 - 5000 = 21\,125 - 5000 = 16\,125. The best price is 42.5042.50 dollars, selling 650650 tickets for a profit of 16 12516\,125 dollars. The profit price is higher than the revenue price of b): each extra spectator now costs 1010 dollars, so selling fewer tickets at a higher price pays.

e) Solving q=1500−20pq = 1500 - 20p for pp: p=75−q20p = 75 - \frac{q}{20}, so R(q)=75q−q220R(q) = 75q - \frac{q^2}{20} and the cost is C(q)=10q+5000C(q) = 10q + 5000. Π′(q)=R′(q)−C′(q)=75−q10−10=0\Pi'(q) = R'(q) - C'(q) = 75 - \frac{q}{10} - 10 = 0 gives q=650q = 650, and then p=75−32.5=42.5p = 75 - 32.5 = 42.5: the same answer, which is the check. The condition R′(q)=C′(q)R'(q) = C'(q) says: keep selling tickets as long as one more ticket brings in more money than it costs, and stop when the two are equal. Π′′(q)=−110<0\Pi''(q) = -\frac{1}{10} < 0 on the whole domain, so this candidate is the absolute maximum without evaluating any endpoint, the second licence.

102030405060708048121620242832vertex p = 37.5p = 45: sold outp (dollars)R (thousands of dollars)

Exercise 10: A final exam problem: the largest cylinder inside a cone

A right circular cylinder is inscribed in a right circular cone of base radius 66 cm and height 1212 cm: the base of the cylinder sits on the base of the cone, and its top circle touches the cone all around. The figure is the vertical cross-section through the axis: a triangle, and inside it the rectangle 2r2r by hh.

This is the shape of a long final exam question: set up the constraint from the geometry, optimize, justify, then change the objective and generalize, with exact answers and every step named.

126rh
  • a) Use similar triangles to express hh in terms of rr, and give the domain of rr.
  • b) Find the dimensions of the cylinder of largest volume and that volume, justifying that the maximum is absolute.
  • c) What fraction of the volume of the cone does this cylinder fill?
  • d) Find instead the cylinder of largest LATERAL area S=2πrhS = 2\pi rh inscribed in the same cone.
  • e) For a cone of base radius RR and height HH, find the cylinder of largest volume and show that the fraction of c) does not depend on the cone.

Type your answers, the page tells you right or wrong 0/12

a)
Domain of rr ,
b)
c)
d)
e)
Show the solution

Answers

  • a) h12=6−r6\frac{h}{12} = \frac{6 - r}{6}, so h=12−2rh = 12 - 2r; domain [0,6][0, 6]
  • b) V(r)=12πr2−2πr3V(r) = 12\pi r^2 - 2\pi r^3, V′=6πr(4−r)V' = 6\pi r(4 - r); r=4r = 4 cm, h=4h = 4 cm, V=64πV = 64\pi cm3^3
  • c) 64π144π=49\frac{64\pi}{144\pi} = \frac{4}{9}
  • d) S(r)=2πr(12−2r)S(r) = 2\pi r(12 - 2r): r=3r = 3 cm, h=6h = 6 cm, S=36πS = 36\pi cm2^2
  • e) r=2R3r = \frac{2R}{3}, h=H3h = \frac{H}{3}, V=4πR2H27=49⋅πR2H3V = \frac{4\pi R^2 H}{27} = \frac{4}{9} \cdot \frac{\pi R^2 H}{3}

a) In the cross-section, the right half of the triangle has legs 66 (along the base) and 1212 (along the axis). Above the top of the cylinder, the small triangle with apex at the top of the cone has height 12−h12 - h and half-width rr, and it is similar to the big one: r6=12−h12\frac{r}{6} = \frac{12 - h}{12}. Hence 12−h=2r12 - h = 2r and h=12−2rh = 12 - 2r. Domain: r≥0r \ge 0 and h≥0h \ge 0, so r∈[0,6]r \in [0, 6]; at r=0r = 0 the cylinder is the axis, at r=6r = 6 it is the flat base, both of volume 00. Naming the similar triangles is the method mark: h=12−2rh = 12 - 2r written without them is a guess.

b) V(r)=πr2h=πr2(12−2r)=12πr2−2πr3V(r) = \pi r^2 h = \pi r^2(12 - 2r) = 12\pi r^2 - 2\pi r^3, continuous on [0,6][0, 6]. V′(r)=24πr−6πr2=6πr(4−r)V'(r) = 24\pi r - 6\pi r^2 = 6\pi r(4 - r): critical numbers 00 (an endpoint) and 44. Closed interval method: V(0)=0V(0) = 0, V(6)=0V(6) = 0, V(4)=π×16×4=64πV(4) = \pi \times 16 \times 4 = 64\pi. The same conclusion reads on the sign of V′V': positive on (0,4)(0, 4), negative on (4,6)(4, 6). The largest cylinder has radius 44 cm and height h=12−8=4h = 12 - 8 = 4 cm, and volume 64π64\pi cm3^3. Its height equals its radius, not its diameter: this cylinder is short and wide.

c) The cone has volume 13π×62×12=144π\frac{1}{3}\pi \times 6^2 \times 12 = 144\pi cm3^3, so the cylinder fills 64π144π=49\frac{64\pi}{144\pi} = \frac{4}{9} of it, a little less than half. The comparison is a sanity check too: a cylinder inscribed in a cone must fill less than the whole cone, and more than nothing.

d) S(r)=2πr(12−2r)=24πr−4πr2S(r) = 2\pi r(12 - 2r) = 24\pi r - 4\pi r^2 on [0,6][0, 6], with S(0)=S(6)=0S(0) = S(6) = 0. S′(r)=24π−8πr=0S'(r) = 24\pi - 8\pi r = 0 for r=3r = 3, and S(3)=2π×3×6=36πS(3) = 2\pi \times 3 \times 6 = 36\pi. The cylinder of largest lateral area has radius 33 cm and height 66 cm, a different cylinder from b): the area grows like rr and the volume like r2r^2, so the volume pushes toward wider cylinders. As in Exercises 5 and 6, the optimum belongs to the objective, and the first line of the answer names it.

e) Similar triangles give rR=H−hH\frac{r}{R} = \frac{H - h}{H}, so h=H(1−rR)h = H\left(1 - \frac{r}{R}\right) on [0,R][0, R], and V(r)=πH(r2−r3R)V(r) = \pi H\left(r^2 - \frac{r^3}{R}\right). V′(r)=πH(2r−3r2R)=πHr(2−3rR)V'(r) = \pi H\left(2r - \frac{3r^2}{R}\right) = \pi H r\left(2 - \frac{3r}{R}\right), zero at r=2R3r = \frac{2R}{3}, with V(0)=V(R)=0V(0) = V(R) = 0 and V>0V > 0 in between, so this is the absolute maximum. Then h=H(1−23)=H3h = H\left(1 - \frac{2}{3}\right) = \frac{H}{3} and V=π⋅4R29⋅H3=4πR2H27V = \pi \cdot \frac{4R^2}{9} \cdot \frac{H}{3} = \frac{4\pi R^2 H}{27}. The cone has volume πR2H3=9πR2H27\frac{\pi R^2 H}{3} = \frac{9\pi R^2 H}{27}, so the fraction is 49\frac{4}{9} for EVERY cone. With R=6R = 6 and H=12H = 12: r=4r = 4, h=4h = 4, as in b).

© Ahmed Squalli Houssaini. Exercise set published at www.letuteurscientifique.ca/en/exercices/math140-optimization. Free for personal and classroom use; republishing it elsewhere requires written permission (legal notice).

See also

Struggling with MATH 140?

I tutor first-year calculus at McGill and Concordia, in English or in French, in Montreal or online. Get in touch for a first session.

Site by Studio Squalli