A hemispherical bubble is placed on top of a spherical bubble of radius 1 .
A smaller, second hemispherical bubble is then placed on the first one. This process is continued until n hemispheres are placed.
Find the maximum possible height of such a tower with 8 0 hemispheres on top of the sphere.
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.
I would love to understand this explanation better; how do you sum the final height so easily to 81sqrt(d)+1? This is the step I do not understand. Love to understand this better. Thank you.
Log in to reply
from equations that all partial derivatives must be equal to zero, we get that 1 − r 1 2 = r 1 2 − r 2 2 and r 1 2 − r 2 2 = r 2 2 − r 3 2 . . . . or this can be written as 1 − r 1 2 = r 1 2 − r 2 2 = r 2 2 − r 3 2 = r 3 2 − r 4 2 ... and so on... this is exactly what we have under square roots in height expression, so all square roots are equal. Also because distance between each consecutive r squares are equal, and it is equal to distance from r 1 2 to 1 and to distance from r 8 0 2 to zero. it means that points r 1 2 , r 2 2 , . . . r 8 0 2 divide line segment [0..1] into 81 equal pieces. and length of each piece is 1/81. so all the expressions under each square root are equal to 1/81. i.e 1 − r 1 2 = r 1 2 − r 2 2 = r 2 2 − r 3 2 = . . . = 1 / 8 1
This visualization shows how height depends on bubble radius. (you can change radius by dragging slide bars on the left part of screen).
Log in to reply
That's an excellent visual tool. I like that it helps you easily address whether maximizing each bubble's height maximizes the tower's height.
In general for n bubbles h=sqrt(n+1)+1.
I think this brings out an interesting aspect. Maximizing the height at each level, if you were stacking one by one, does not maximize the height overall. One must give up height at a previous level to allow for more height at the next level. The link shows an example of this with 3 bubbles: Desmos Graph
Log in to reply
I considered the optimal radius for every sphere, and that is exactly how i got this wrong :).
I tried to do it using trigonometry and the angle for max height was π/4. The total height became the sum of a GP and answer comes out as 2 + (1-(1/2)^40)(√2)
Log in to reply
I also got the same answer as you friend... In solution above I couldn't get how he took 0 as the first term of GP.. That means he is saying that after 80 hemispheres no other hemisphere could be put over them
As I mentioned in another comment. The interesting thing for me about this problem is that one cannot just optimize at each step. One has to lower the height of intermediate hemisphere (by overshooting the radius) in order to provide a great height for the next hemisphere.
Could u pls explain why is it 81 sqrt(d)!!!thanks
Log in to reply
because r80^2 - 0 = r79^2 - r80^2 = r78^2 - r79^2 = ... = 1 - r1^2 = 1/81 (= d) there are 81 square roots and each of them is equal to sqrt(d). so sum equals to 81 * sqrt(d)
The height is the not the one that you denoted, but the one that adds on top.It isn't clear where you got the final answer.
The height of bubbles is given by for the nth bubble H n = 1 + 1 − r 1 2 + r 1 2 − r 2 2 + … + r n − 1 2 − r n 2 + r n . Using QM-AM maximum height is: 8 1 1 − r 1 2 + r 1 2 − r 2 2 + … + r 7 9 2 − r 8 0 2 + r 8 0 2 ≥ 8 1 1 − r 1 2 + r 1 2 − r 2 2 + … + r 7 9 2 − r 8 0 2 + r 8 0
8 1 × 9 1 ≥ H 8 0 − 1
1 0 ≥ H 8 0
Thus maximum height possible is 10. equality occurs when 1 − r 1 2 = r 1 2 − r 2 2 = … = r 7 9 2 − r 8 0 2 = r 8 0
Certainly more elegant than partial derivatives.
I took a derivative for the base case and inducted to find that the total height for n hemispheres would be 1 + n + 1 , but this is a fantastically simpler method. Well done!
Suppose that the maximum height of a bubble tower of n bubbles, where the largest bubble has radius 1 is 1 + n . This is certainly true if n = 1 . If we now consider a tower of n + 1 bubbles, then then second to ( n + 1 ) st bubbles form a tower of n bubbles (apart from the first hemisphere), where the bottom bubble (of this subtower) has radius 0 < x < 1 (and the height of this subtower will be as great as possible). Thus the height of the part of the tower from the centre of the bottom bubble (of the subtower) to the top will be x n , and so the height of the whole tower will be H ( x ) = x n + 1 + 1 − x 2 and so we need to choose x to maximize H ( x ) . Since H ′ ( x ) = n − 1 − x 2 x , we see that H ( x ) is maximized when x 2 = n + 1 n . This makes the maximum height of the whole tower 1 + n + 1 .
Thus we deduce, by induction, that the maximum height of a tower of n bubbles is 1 + n . This makes the answer 1 + 8 1 = 1 0 .
Is there a method that doesn't involve pre-assuming and induction?
Log in to reply
What have you got against induction? This question is one that responds to an inductive proof nicely.
I have not assumed anything (except the assumptions inbuilt in an inductive proof). If a tower of n + 1 bubbles is maximal, it will consist of a first bubble, plus a tower of n bubbles, and this tower of n will be as big as it can be, given the radius of its initial sphere. Induction then tells us how tall the tower of n bubbles can be. We are then just choosing the radius of the second sphere to maximize the overall height.
how did you get 1 + √n to assume?
Log in to reply
Suppose the maximum over n bubbles was 1 + F ( n ) . This would give the function H ( x ) = 1 + F ( n ) x + 1 − x 2 , which yields the maximum 1 + 1 + F ( n ) 2 for n + 1 bubbles. From here getting F ( n ) = n is easy.
32年前一个外国男孩给了马云两百,如今马云老了32岁
I believe the third sentence should read "...where the second bubble has radius 0 < x < 1 ..." as opposed to "...where the smallest bubble has radius 0 < x < 1 ..." Nice proof otherwise!
Log in to reply
Actually, it should be the largest or bottom hemisphere of the subtower of the 2 nd to the ( n + 1 ) st hemispheres. I have changed smallest to bottom .
Really good proof! I think you left out a short (but important) step though:
Proof that subtower has max height of x*sqrt(n): (Sorry for my lack of LATEX)
We've assumed the max height of a tower with a bottom sphere of radius 1 is 1 + sqrt(n). Thus the max height of a tower of hemispheres such that the first has radius 1 is sqrt(n). Let h be the height of the tower (sqrt(n) is the maximum of h by definition). For constant x the maximum of xh is clearly xsqrt(n).
Log in to reply
You yourself say that the height of the sub tower clearly equals x n . It was so clear I left out the details. We are just using the idea of similar shapes, after all.
Log in to reply
I wrote it at the end of my proof, after I'd... proved it. I should hope it's clear by then, although 'clearly' is a poor choice of words. :P
IMO it's a nontrivial step, but to each their own! Your proof is definitely my favorite.
I got it currently I'm only 14
I though that if is 3 dimension an 80 bubbles, (80)/(2^3)=10
While I never was able to prove it using induction, the solution to the problem can actually be derived using Lagrange’s multiplier method. In particular, let’s say that the entire height is composed of multiple segments h[1], h[2], h[3] … h[n] with associated radii r[1], r[2], r[3] … r[n]. Then clearly h[n] = sqrt(r[n]^2 – r[n+1]^2).
Now let’s square this to get h[n]^2 = r[n]^2 – r[n+1]^2 (this is an expression [a])
The radius of last hemisphere is h[n] = r[n].
From [a], it follows that r[n]^2 = h[n]^2 + r[n+1]^2 = h[n]^2 + h[n+1]^2 + … h[n]^2.
So when r[1] = 1 then
h[1]^2 + h[2]^2 + … h[n]^2 = 1.0
So we’ve established a constraint on the sum of h[n]. The tower height H = F(h[i]) = 1.0 + sum (h[i], 1 <= i <= n)
Lagrange’s multiplier method says that
gF = lamdba x gG (where g is nabla and lambda is a scalar.) So
gF = [1 1 1 … 1]
and gG = [1/2 lamdba h[1], 1/2 lamdba h[2] … 1/2 lambda h[n])
So h[i] = n / (4 lambda^2), so lambda = sqrt(n) / 2. Subsituting lambda in individual h[i] expression we get:
h[i] = 1/2 * 2/sqrt(n) = 1/sqrt(n).
Finally substituting into original equation we get the sought-after result:
H = 1 + sum(h[i], 1 <= i <= n) = 1 + n/sqrt(n) = 1 + sqrt(n)
H=1+81/9 H=1+9 H= *10 *
Problem Loading...
Note Loading...
Set Loading...
Lets denote radius of hemispherical bubbles as r 1 , r 2 , ... r 8 0 (from largest to smallest)
distance between centers of two consecutive bubbles i and i + 1 will be h i , i + 1 = r i 2 − r i + 1 2 ( by Pythagorean theorem, r i 2 = h i , i + 1 2 + r i + 1 2 )
so total height of a bubble tower will be h = r 8 0 + r 7 9 2 − r 8 0 2 + r 7 8 2 − r 7 9 2 + . . . + r 1 2 − r 2 2 + 1 − r 1 2 + 1
then take partial derivatives and compare them to zero (necessary conditions for optima)
d r 1 d h = r 1 2 − r 2 2 r 1 − 1 − r 1 2 r 1
...
d r i d h = r i 2 − r i + 1 2 r i − r i − 1 2 − r i 2 r i
...
d r 8 0 d h = 1 − r 7 9 2 − r 8 0 2 r 8 0
they will be equal to zero, when
1 − r 1 2 = r 1 2 − r 2 2
...
r i 2 − r i + 1 2 = r i − 1 2 − r i 2
...
r 8 0 2 = r 7 9 2 − r 8 0 2
so, numbers 0 , r 8 0 2 , r 7 9 2 , . . . r 2 2 , r 1 2 , 1 forms arithmetic progression with difference d = 8 1 1
and total height of a bubble tower can be written as
h = r 8 0 2 − 0 + r 7 9 2 − r 8 0 2 + . . . + r 1 2 − r 2 2 + 1 − r 1 2 + 1 = 8 1 d + 1 = 9 8 1 + 1 = 1 0