3 coins

Calculus Level 3

The centers of three identical coins form the angle that's colored green above.

What angle maximizes the area of the blue convex hull?

Note: You can imagine the perimeter of the convex hull as a rubber band stretched around the three coins.

60 degrees 90 degrees 120 degrees 135 degrees 150 degrees 180 degrees

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

8 solutions

Jeremy Galvagni
Apr 29, 2018

Relevant wiki: Extrema - Problem Solving - Medium

The white sectors together make a whole circle regardless of the angle. We can ignore them.

The orange rectangles also do not depend on the angle we can ignore them too.

Let the coins have radius 1 and the angle be 2 θ 2\theta then the triangle has area 2 sin 2 θ 2\sin{2\theta} and the blue rectangle has area 4 sin θ 4\sin{\theta} so the area as a function of θ \theta is

A ( θ ) = 2 sin 2 θ + 4 sin θ A(\theta)=2\sin{2\theta}+4\sin{\theta}

A ( θ ) = 4 cos 2 θ + 4 cos θ A'(\theta)=4\cos{2\theta}+4\cos{\theta}

Solving A ( θ ) = 0 A'(\theta)=0

4 cos 2 θ + 4 cos θ = 0 4\cos{2\theta}+4\cos{\theta}=0

2 cos 2 θ + cos θ 1 = 0 2\cos^{2}\theta+\cos{\theta}-1=0

cos θ = 1 2 o r cos θ = 1 \cos\theta = \frac{1}{2} or \cos\theta=-1

θ = 60 o r 180 \theta=60 or 180

2 θ = 120 o r 360 2\theta=120 or 360

The solution that makes sense is 2 θ = 120 d e g r e e s 2\theta=\boxed{120 degrees}

Note this picture is not the exact solution but it is close. The middle circle would have to be tangent to the line joining the other circles' centers.

You need to address that the solution to A ( θ ) = 0 A'(\theta) = 0 is the global maximum and not just an extrema.

Eli Ross Staff - 3 years, 1 month ago

Log in to reply

Just make a quick graph, then. I don't mean to sound rude but this isn't a calculus exam.

Jeremy Galvagni - 3 years, 1 month ago

Log in to reply

Jeremy, I also found a simple formula that can directly calculate the changing area would be assuming the radius of all the circles are 1. Here it is:

Area = 4cos(x)(sin(x)+1)-pi The way I approached it was a bit different than yours but your answer almost looks like mine so I decided I could share this one with you and see what you thought about it. If you want me to explain how I got this answer I am more than glad to share that as well.

Also, if you don't know what I mean by the changing area I mean the section where the area changes depending on the angles given. Look for John McLaughlin's post below to see what I mean (it's the blue area).

xXTheUniverseXx . - 3 years ago

Log in to reply

@Xxtheuniversexx . Oh yeah and the angle you plug in would be the CAB angle or the ACB angle (doesn't matter since both angles will always be the same).

xXTheUniverseXx . - 3 years ago

The way to remedy this would be to show that A ( 60 ) A(60) is greater than A ( 30 ) A(30) and A ( 90 ) A(90) , correct?

John Ross - 3 years, 1 month ago

Log in to reply

Or the second derivative test: for θ \theta between 0 0 and π / 2 , \pi/2, the second derivative 8 sin 2 θ 4 sin θ -8\sin 2\theta - 4\sin \theta is always negative, so the local maximum must be a global maximum.

Patrick Corn - 3 years, 1 month ago

Right. And you find the minimum which is at $\theta = 90$.

Chongchong He - 1 year, 8 months ago

So the answer is 120 degrees?

Clayton Birchenough - 3 years, 1 month ago

Used the same rules. I made a geogebra figures that helped me solve it. If you want to have a look : geogebra You can move the blue dot for various angles

Thomas Lesgourgues - 3 years, 1 month ago

Log in to reply

This is how I checked my solution was correct.

Jeremy Galvagni - 3 years, 1 month ago

I used the same approach, based on the insight that that only the rectangle-triangle region ever changed in area and that the three set of circle regions defined by the various lines each preserved a full circle's area regardless of angle, but used the algebraic expressions from the Pythagorean theorem, area as a function of the altitude of the triangle (ranging from 0 at 180 to √3 at 60) instead of the trig functions. Evaluating the derivative of that function (using unit circles) produced a maximum at h = 1 (and minimum at h = 0) corresponding to an angle of 120 degrees.

Robert DeLisle - 3 years ago

Does anyone know how to create this gif animation in geometers sketchpad. I can animate the circles but I am unsure how to wrap the "elastic band" around them.

Michael Vamvadelis - 3 years ago

Log in to reply

Construct the relevant radii as perpendicular to the segments joining the centers of the circles to get the points. Then construct the three segments and three arcs and make them the same color and thick.

Can you make animated gifs in gsp?

Jeremy Galvagni - 3 years ago

Log in to reply

Thanks for the tips. I managed to get it right. I haven't been able to get the gif part just yet but this link from the GSP website is my starting point.

http://www.dynamicgeometry.com/Technical Support/FAQ/Sketchpad and the Internet/Web_Options.html

Michael Vamvadelis - 3 years ago
Luke Limbo
May 9, 2018

The maximum and minimum angle measure of this are 180 and 60 respectively. So for the optimum area, it will be the median of these measures. Thus, (180+60)/2 = 120.

Excellent! I too solve such problems dealing with geometry through this very intuitive method........

T C Adityaa - 3 years, 1 month ago

I solved it by eyeballing the animation... can you justify why the mean is the solution?

a b - 3 years, 1 month ago

That intuition worked in this instance, but it assumes that the graph of area for angle is symmetric. In other problems, that assumption will be wrong.

Jerry Barrington - 3 years, 1 month ago

in order to use the mean angle value, you need to show that the function linking the angle to the area is linear (which is not obvious)

Nicolas Cormond - 3 years, 1 month ago

I love simple answers :)

Fernando Roque - 3 years, 1 month ago

Log in to reply

When they are actually answers, so do I.

Robert DeLisle - 3 years ago

Hand waving.

Robert DeLisle - 3 years ago

Log in to reply

eyes rolling

Mehdi K. - 2 years, 8 months ago
Brian Hopkins
May 9, 2018

I did no math for this, but I already knew from rock climbing that a 120-degree angle in your anchor creates maximum tension (that's very bad). So 120 degrees was likely to be the most 'stretched' angle for the blue band.

Very nice!

Vin Benzin - 3 years, 1 month ago

I think it’s brilliant that 2 seemingly unrelated activities could link together like this

Paddy Hunn - 3 years, 1 month ago

I don't know whether or not this is valid, in this case, tension is proportional to the length of the band, which is 2 π r + 4 r + 4 r sin ( θ 2 ) 2\pi r+4r+4r\sin(\frac{\theta}{2}) , where θ \theta is the vertex angle of the isosceles triangle formed by the centers of the circles, and that maximizes at θ = π \theta=\pi .

Also, If I'm imagining this correctly, the tension in an anchor angle θ \theta would be proportional to sec ( θ 2 ) \sec(\frac{\theta}{2}) , which sure is big at θ = 2 π 3 \theta=\frac{2\pi}{3} , but it goes to infinity as θ \theta aproaches π . \pi. But I know nothing about rock climbing, you might me talking about something else.

Pedro Cardoso - 3 years, 1 month ago

street smarts :)

Adrian Mišak - 3 years ago
John McLaughlin
May 12, 2018

Relevant wiki: Optimization

The light and dark gray areas are constant, so we must maximize the blue area.

The blue area is the area of this pentagon minus the three green sectors. If we set the circle radii to be 1, the area of a sector is half the angle, so we have:

A g r e e n = θ + 2 ( π θ 2 ) 2 = π A_{green} = \frac{\theta + 2(\pi-\frac{\theta}{2})}{2} = \pi

Therefore the problem reduces to maximizing the pentagon area, which is equivalent to maximizing one half as follows:

By basic trigonometry:

b = 2 cos θ 2 a 2 + b 2 = 4 b = 4 a 2 d b d a = 1 2 1 4 a 2 ( 2 a ) = a b \begin{aligned} b &= 2 \cos\frac{\theta}{2} \\ a^2 + b^2 &= 4 \\ b &= \sqrt{4-a^2} \\ \dfrac{\text{d}b}{\text{d}a} &= \frac{1}{2}\frac{1}{\sqrt{4-a^2}}(-2a) = -\frac{a}{b} \\ \end{aligned}

Now the area to maximize is the area of the rectangle plus the area of the triangle and the solution is where the gradient with respect to a a is zero:

A = a + a b 2 = a ( 1 + b 2 ) d A d a = 0 = 1 + b 2 + a ( 1 2 d b d a ) = 1 + b 2 a 2 2 b = 1 + b 2 4 b 2 2 b = b + b 2 2 2 + b 2 2 = b 2 + b 2 = ( b 1 ) ( b + 2 ) \begin{aligned} A &= a + \frac{ab}{2} \\ &= a(1 + \frac{b}{2}) \\ \dfrac{\text{d}A}{\text{d}a} = 0 &= 1 + \frac{b}{2} + a(\frac{1}{2}\dfrac{\text{d}b}{\text{d}a}) \\ &= 1 + \frac{b}{2} - \frac{a^2}{2b} \\ &= 1 + \frac{b}{2} - \frac{4-b^2}{2b} \\ &= b + \frac{b^2}{2} - 2 + \frac{b^2}{2} \\ &= b^2 + b - 2 \\ &= (b-1)(b+2) \end{aligned}

and since 0 b 2 0 \leq b \leq 2 we have b = 1 b = 1 . To confirm this is a maximum, we can easily show that the second derivative at b = 1 b=1 is 3 3 -3\sqrt{3} .

We could now calculate θ \theta using an arccos \arccos , however it's clear that the triangle above is the 1 , 2 , 3 1,2,\sqrt{3} triangle making θ 2 = 6 0 \frac{\theta}{2} = 60^\circ and θ = 12 0 \boxed{\theta = 120^\circ} .

Instead of using geometry like Jeremy Galvagni , I used processing and plotted all the points for the three circles, then ran a convex hull calculator to calculate the convex hull around all three circles, filled in that shape with red, and then counted up all the red pixels. I did this for every angle, and then found the angle that gave the greatest number of red pixels.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
// http://leebyron.com/mesh/
import megamu.mesh.*;

// points to hull up
float[][] points = new float[360*3][2];
int currentPoint = 0;

// radius of the cirlce
int r=100;

// the angle in the middle
float ang=120;

int inc=-1;

void setup() {
size(600,600);

}

int tmax=0;
float rmax=0;

void drawCircle(float x, float y, float r, int start, int stop) {
  for (int i=0; i < 360; i++) {
      float x2 = r * cos(radians(i));
      float y2 = r * sin(radians(i));

      //ellipse(x+x2, y+y2, 3,3);
      points[currentPoint][0] = x + x2;
      points[currentPoint][1] = y + y2;
      currentPoint++;
  }
}

void draw() {
currentPoint = 0;
background(0);
translate(width/2, height/2);

ang += inc;
if (ang <=60) { inc=1;}
if ( ang>=180) {inc=-1;}


// 180 degreees total, subtract the angle we have set, then divide by two to find
// the side angles 
float angSide = -1 * (180-ang)/2;

// Circle 1
//ellipse(0,0, r*2,r*2);
drawCircle(0,0,r, 45,135);

// Circle 2
float c2x = 2 * r * cos(radians(angSide));
float c2y = 2 * r * sin(radians(angSide));

//println( c2x,c2y);
//ellipse(c2x, c2y, r*2,r*2);
drawCircle(c2x,c2y,r,270,360);
line(0,0, c2x,c2y);

// Circle 3
float c3x = 2 * r * cos( radians( 180-angSide) );
float c3y = 2 * r * sin( radians( 180-angSide) );
//ellipse(c3x, c3y, r*2,r*2);
drawCircle(c3x,c3y,r, 180,270);
line(0,0, c3x,c3y);

Hull myHull = new Hull( points );

MPolygon myRegion = myHull.getRegion();
fill(255,0,0);
myRegion.draw(this);

int total=0;
for (int x=0; x< width; x++) {
  for (int y=0; y< height; y++) {
    //println(red(get(x,y)));
    if (  red(get(x,y)) == 255 ) {
       total = total + 1;
    }
  }
}

//println( ang, total);

if (total > tmax) { 
  tmax=total; 
  rmax=ang; 
  println("new max:", total, ang);
}
}

My congratulations to the developer of the convex hull calculator. "instead of using" might better read ":Instead of solving the problem."

Even allowing that pixel counting (probably) gives the correct answer, and that checking only the given answers in a multiple choice quiz is sometimes a good practical test-taking tactic, but it gives little insight into this problem. For example, you get no assurance that any of the given answers is the true maximum. (Nor do you need that if the only objective is to get past the question.)

This approach is appropriate for a problem for which approximation methods are the only alternative. But even then there must be some solid theory behind the approximation method that insures that the computation is sound. (I wonder if those stories about the rocket going the wrong way because someone assumed a divergent series converged are true or merely urban legend?)

I prefer infinitesimals to pixels.

Robert DeLisle - 3 years ago
Zeeshan Ali
May 11, 2018

Intuitively, I guessed that the area would be max when all the three sides are the same which would be when angle between the known vectors is 12 0 0 120^0 .

Omar Valeinis
May 9, 2018

Huh. I just figured that if the lowest areas correspond to 60 degrees and 180 degrees, then the largest area would be halfway between the two at 120.

Not even hand waving.

Robert DeLisle - 3 years ago
Takahiro Waki
May 24, 2018

Observe when more than 120 degree, area is decreasing.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...