Boxed in a Box

Geometry Level 3

There are 18 white squares of side length 1 packed inside a larger square.

The shaded area is equal to A A 2 B \dfrac{A\sqrt{A}}{2}-B , where A A is a prime number.

Find A + B A+B .


The answer is 11.

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.

5 solutions

Digvijay Singh
Mar 27, 2018

Let the width of the gap be x x , as shown.

Then Δ A B C \Delta ABC , is a right angled triangle, with A B = x + 1 AB=x+1 , B C = x BC=x and C A = 2 CA=2

Applying Pythagoras theorem, ( x + 1 ) 2 + x 2 = 2 2 (x+1)^2+x^2=2^2

x = 7 1 2 \implies x=\dfrac{\sqrt{7}-1}{2}

Side length of the larger square = 4 + x = 7 + 7 2 =4+x=\dfrac{\sqrt{7}+7}{2}

Shaded area = Area of the bigger square Total area of the smaller squares \text{Shaded area} = \text{Area of the bigger square} - \text{Total area of the smaller squares}

= ( 7 + 7 2 ) 2 18 =\left(\dfrac{7+\sqrt{7}}{2}\right)^2-18

= 7 7 2 4 =\boxed{\dfrac{7\sqrt{7}}{2}-4}

Moderator note:

This is incidentally thought to be an "optimal packing" of 18 unit squares in the smallest square possible. While this pattern was found in 1979 it has not yet been proven to be optimal.

A table of all such packings up to 89 is here. The smallest unproven packing as of this writing is for 11 unit squares.

haha, that was funny!

Jess McAllister Alicando - 3 years, 2 months ago

Log in to reply

What was funny?

Digvijay Singh - 3 years, 2 months ago

Log in to reply

What I mean is the hint, it is very subtle but it is very important to solve the problem

Jess McAllister Alicando - 3 years, 2 months ago

@Digvijay Singh Yup!! Exactly same!! And hey, thnx for the book!! It helped me a lot !!

Aaghaz Mahajan - 3 years, 2 months ago

You assume that AC is parallel to the edges of the smaller squares and therefore equal to 2. You have to prove it before you use it.

Kanenas Kanenos - 3 years, 2 months ago

Log in to reply

Yeah ofc you have to prove that. I am searching for a proof for a while now. Could someone help me??

Alkis Gaitanidis - 3 years, 2 months ago

May I ask why the width of vertical gap is the same as that of horizontal gap?

Ranian Liu - 3 years, 2 months ago

Log in to reply

The problems says the outer figure is "a larger square", and it has 4 tiles placed along each side. Therefore both gaps must be of equal size.

Fabio Gama - 3 years, 2 months ago

Log in to reply

That's right. I missed that description. Thanks!

Ranian Liu - 3 years, 2 months ago

I am pretty sure that there is no proof that AC is parraler to the edges of the smaller square with this clues that it gives...

Alkis Gaitanidis - 3 years, 2 months ago

I will agree with this if you can prove that AC and the edges of the smaller squares are parallel, until then I disagree

Noel Kahn - 3 years, 2 months ago

Log in to reply

see the new solution posted by John Frank.

Digvijay Singh - 3 years, 2 months ago

For those of you wondering why AC and the edges of the smaller squares are parallel, check out my attempt below John Frank's solution

Kevin Nguyen Nguyen - 3 years, 2 months ago

Need more proofs on that AC is parallel to the edges of the smaller squares.

Min Jiang - 3 years, 1 month ago

Look for and see Mike Jarvis' reply below for the best way to analyze this problem. Length A C AC is actually arbitrary ----unless we impose the condition that the area of the larger square has to be a minimum. Then A C AC has to be 2 2 . Check Peter Vanbroekhoven's work before reading Mike Jarvis' reply.

The statement of this problem needs to include this additional condition of minimum area of larger square.

Michael Mendrin - 3 years, 1 month ago

How do you get from (x+1)^2 + x^2 = 2^2 to x = sqrt(7) - 1 / 2?

David Schleter - 3 years, 1 month ago

Log in to reply

it's a quadratic equation. Which can be simplified to 2 x 2 + 2 x 3 = 0 2x^2+2x-3=0 .

Digvijay Singh - 3 years, 1 month ago

(Edited: changed the origin)

After some analysis, I figured out that any angle θ θ between 0 and 0.7254440201 will do (outside these bounds, the corner points A A through E E in the diagram below will no longer be on the edges or same edges of the rotated squares, see my reply below). For reference the diagram given in the question with some points and angles named.

The coordinates of the different points are (given the usual axes: x x to the right, y y up):

A = ( x , 2 + x ) B = ( 1 , 0 ) C = ( 1 + x , 1 + x ) D = ( 0 , 1 ) E = ( 1 + x , 1 ) F = ( 0 , 1 + x ) A=(x, 2+x)\\ B=(1, 0)\\ C=(1+x, 1+x)\\ D=(0, 1)\\ E=(1+x, 1)\\ F=(0, 1+x)

We then rotate these points by θ counterclockwise to bring them into the coordinate system of the 4 rotated squares. For this, we use rotation matrix R = [ c o s ( θ ) s i n ( θ ) s i n ( θ ) c o s ( θ ) ] R = \begin{bmatrix}cos(θ) & -sin(θ) \\ sin(θ) & cos(θ)\end{bmatrix} .

A = R A = ( x c o s ( θ ) ( 2 + x ) s i n ( θ ) , x s i n ( θ ) + ( 2 + x ) c o s ( θ ) ) B = R B = ( c o s ( θ ) , s i n ( θ ) ) C = R C = ( ( 1 + x ) c o s ( θ ) ( 1 + x ) s i n ( θ ) , ( 1 + x ) s i n ( θ ) + ( 1 + x ) c o s ( θ ) ) D = R D = ( s i n ( θ ) , c o s ( θ ) ) E = R E = ( ( 1 + x ) c o s ( θ ) s i n ( θ ) , ( 1 + x ) s i n ( θ ) + c o s ( θ ) ) F = R F = ( ( 1 + x ) s i n ( θ ) , ( 1 + x ) c o s ( θ ) ) A' = R \cdot A = (x \cdot cos(θ) - (2+x) \cdot sin(θ), x \cdot sin(θ) + (2 + x) \cdot cos(θ))\\ B' = R \cdot B = (cos(θ), sin(θ))\\ C' = R \cdot C = ((1+x) \cdot cos(θ) - (1+x) \cdot sin(θ), (1+x) \cdot sin(θ) + (1+x) \cdot cos(θ))\\ D' = R \cdot D = (-sin(θ), cos(θ))\\ E' = R \cdot E = ((1+x) \cdot cos(θ) - sin(θ), (1+x) \cdot sin(θ) + cos(θ))\\ F' = R \cdot F = (-(1+x) \cdot sin(θ), (1+x) \cdot cos(θ))

We then calculate the vertical distances between the pairs of points that bound the 4 rotated suqares:

vertical distance A to D = x s i n ( θ ) + ( 2 + x ) c o s ( θ ) ( c o s ( θ ) ) = x s i n ( θ ) + ( 1 + x ) c o s ( θ ) vertical distance C to B = ( 1 + x ) s i n ( θ ) + ( 1 + x ) c o s ( θ ) ( s i n ( θ ) ) = x s i n ( θ ) + ( 1 + x ) c o s ( θ ) horizontal distance E to F = ( 1 + x ) c o s ( θ ) s i n ( θ ) ( ( 1 + x ) s i n ( θ ) ) = ( 1 + x ) c o s ( θ ) + x s i n ( θ ) \text{vertical distance } A' \text{ to } D' = x \cdot sin(θ) + (2 + x) \cdot cos(θ) - (cos(θ)) = x \cdot sin(θ) + (1+x) \cdot cos(θ)\\ \text{vertical distance } C' \text{ to } B' = (1+x) \cdot sin(θ) + (1+x) \cdot cos(θ) - (sin(θ)) = x \cdot sin(θ) + (1+x) \cdot cos(θ)\\ \text{horizontal distance } E' \text{ to } F' =(1+x) \cdot cos(θ) - sin(θ) - (-(1+x) \cdot sin(θ)) = (1+x) \cdot cos(θ) + x \cdot sin(θ)

Each of these distances is actually equal, and should equal 2:

x s i n ( θ ) + ( 1 + x ) c o s ( θ ) = 2 x \cdot sin(θ) + (1+x) \cdot cos(θ) = 2

Solving for x x gives x = ( 2 c o s ( θ ) ) / ( s i n ( θ ) + c o s ( θ ) ) x = (2 - cos(θ)) / (sin(θ) + cos(θ)) .

For every θ θ between 0 and 0.7254440201, we can find a corresponding gap x x . For θ = 0 θ = 0 , we find a x = 1 x = 1 , and a blue area of 7 (this is easily verified visually). Finding the θ θ for which x x is minimal, leads to the solution presented by other people (I haven't included this to keep the length down, but I did check). But it is clear that the question needs to specify something to fix the angle, otherwise it is ambiguous.

Actually, π / 4 is the wrong upper bound. It's some number larger than the angle in the image, but I haven't calculated it yet.

Peter Vanbroekhoven - 3 years, 2 months ago

Log in to reply

The maximum for θ is where the x coordinate of A' = -1, meaning that A is on the top left corner of the rotated square. This corresponds to θ = 0.7254440201. The value of θ that corresponds to the configuration with FE parallel is 0.4240310394 and is in the range where the squares can fit.

Peter Vanbroekhoven - 3 years, 1 month ago

The easiest way to do the minimization is to realize that the smallest FE can be is 2, since it has to cross 2 unit squares. So if it's not going perpendicular to DA, then it will be longer than 2. So FE = 2 and triangle FDE is the (x, 1+x, 2) right triangle.

Mike Jarvis - 3 years, 2 months ago

Log in to reply

You are right, but see how deceiving the diagram is. I should have taken the effort to draw it at another angle. You probably meant to say "perpendicular to the left side of the rotated squares".

You see, FE is always perpendicular to DA. They are both the hypotenuse of a (x, 1+x, 2) right triangle. And because of that, the vertical distance between FE will always equal the horizontal distance between DA. Which explains intuitively the result I got above that all distances A'D', C'B' and E'F' are the same.

Peter Vanbroekhoven - 3 years, 1 month ago

Log in to reply

You probably meant to say "perpendicular to the left side of the rotated squares".

Yes. That is what I had meant to say. Of course it is always perpendicular to DA. But it has its minimum when it is also perpendicular to sides of the inner unit squares. Sorry for the error.

Mike Jarvis - 3 years, 1 month ago

FE is always going to be perpendicular to both DA and CB, which can be checked by extending some lines the sides of the blue areas. But you've hit on the key and correct point---first we can see that for any arbitrary angle the inner squares are set, all 6 corners of the outer squares can be made to touch the edges of the inner squares. Then for minimum area of the large square, FE has to be 2. This is the simplest way of analyzing this problem, which leads to the solution as already given.

Michael Mendrin - 3 years, 1 month ago

why is the first coordinate of A -(1-x)/2 and not 0?

Ido Hadanny - 3 years, 1 month ago

Log in to reply

Because of where I put the origin. It is halfway between A and B, and as A is to the left of B (unless x = 1), A must be to the left of the origin.

In retrospect, I should have put the origin in the bottom left corner or so as that would get rid of all the divisions by 2, but initially I tried a different approach for which the current position of the origin was handier and I never changed it. Changing the origin doesn't make a difference.

I'll see if I can redo the calculation with a different origin. That should make things easier to understand. Thanks.

Peter Vanbroekhoven - 3 years, 1 month ago

Thanks for demonstrating an important fact: Length A C AC is arbitrary , unless we impose the condition that the area of the larger square is the minimum.

Michael Mendrin - 3 years, 1 month ago

Log in to reply

Do you mean length AD? Length AC is sqrt(2).

Richard Desper - 3 years, 1 month ago

Log in to reply

Use Digvijay Singh's diagram to see where A C AC goes. It's the length of 2 unit squares. Yes, in Peter Vanbroekhaven's diagram, it's A D AD .

Michael Mendrin - 3 years, 1 month ago

This problem, was very poorly drawn. Although the problem, if one did not look at the figure, might have been interesting, with the figure it was misleading, and appeared impossible. The trouble with it was that the width of the open areas appears different. Only when one looks at the solution, does it appear that they are supposed to be identical.... So, I think it would help, if some of these problems from people who have repeated errors should be checked for accuracy. Sorry, but the figure was, to me, misleading. I should be grateful to hear what Brilliant's reaction is to this problem which was more of problem than it was uplifting.

kenneth schatten - 3 years, 1 month ago

Log in to reply

Because the large square is a square, the gaps touching all four sides have to be identical. Then to see how the arrangement of the boxes touching each other, first cut off the right most column of squares from the large square. Then you'll see a resulting rectangle arrangement of squares that is pefectly symmetrical. The problem with this problem is that for an unique answer, it requires the condition that the area of the large square be minimum.

Michael Mendrin - 3 years, 1 month ago

Peter, isnt the angle fixed by the problem specifying that CB is the lenght of 2? It appears to me that you have assumed the red line is giving the side of the two stacked squares (which we already know is 1+1 ) rather than the distance between C and B. Is there any reason we can't look at the original figure and take CB as 2 as a given? Is there something I'm missing?

Gustaf Carstam - 3 years, 1 month ago

Log in to reply

The question was edited in response to my analysis. The length of CB was not given initially.

Peter Vanbroekhoven - 3 years, 1 month ago

Log in to reply

ah, thank you :)

Gustaf Carstam - 3 years, 1 month ago

I had same thinking that the length 2 in the drawing was referring to the stacked squares and not the distance between the non-rotating corners. Thus, there was an infinite amount of solutions with area between 5.26 and 7.

Pierre Carrette - 3 years, 1 month ago
Weidong You
Mar 27, 2018

Dario Goedtkindt
Apr 15, 2018

I don't know if my way has any value because it is based on a very wild guess:

I approximated the blue area. I counted it was approximately 7. Then I guessed that B was the area of the 4 squares in the middle: B = 4.

So I had this equation:

A ( 3 2 ) 2 B = 7 \frac{A^{(\frac{3}{2})}}{2}-B = 7

When I solve for A : A = 7.8 A=7.8 . Thus, the closest prime number is 7 and thus the answer is 7 + 4 = 11 7+4 = 11

Michael Mendrin
Apr 13, 2018

Look for and see Mike Jarvis' reply to Peter Vanbroekhoven's solution for the best way to analyze this problem. Length A C AC is actually arbitrary ----unless we impose the condition that the area of the larger square has to be a minimum. Then A C AC has to be 2 2 . Check Peter Vanbroekhoven's work to see why A C AC is arbitrary.

The statement of this problem needs to include this additional condition of minimum area of larger square.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...