Euclid's Riddle II

Geometry Level 5

Euclid was given order to measure the area of the quadrilateral A B C D ABCD , where he drew perpendicular lines ( B E BE & C F CF ) to the base A D AD , as shown above.

Euclid : This bizarre-looking shape has no integer side length except for its base ( A D AD ). But no despair. Fortunately, the new lines ( B E BE & C F CF ) here have the lengths of some integers. Now thou will measure and tell me their relationship.

Pupil : B E + C F = 2 A F BE + CF = 2AF , and A F AF is a multiple of F D FD , master. Still, how shall we calculate its area, my teacher?

Euclid then extended the new (red) lines to meet at point G G , as shown below:

Euclid : Now measure these new (red) lines. What can thou deduce from all these segments?

Pupil : A F + F C = C G AF + FC = CG , and they are all pairwise co-prime. Also, A D = D G AD = DG , master.

Euclid : Alas! At last, the solution stands before our eyes. What is the area of this quadrilateral A B C D ABCD ?


The answer is 72.

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.

2 solutions

Mark Hennings
Oct 4, 2017

Suppose that F D = x FD = x , A F = x y AF = xy , B E = b BE = b and C F = c CF= c for positive integers x , y , b , c x,y,b,c . The first set of conditions tell us that b + c = 2 x y b+c \;= \; 2xy Then we know that D G = x ( y + 1 ) DG = x(y+1) and C G = x y + c CG = xy+c , so that x y xy and c c are coprime and Pythagoras' Theorem applied to C F G CFG tells us that y c = 2 x ( y + 1 ) yc \; = \; 2x(y+1) Finally using similar triangles, we have b 2 x ( y + 1 ) A E = c x ( y + 2 ) \frac{b}{2x(y+1)-AE} \; = \; \frac{c}{x(y+2)} so that b ( y + 2 ) < 2 c ( y + 1 ) b(y+2) \; < \; 2c(y+1)

Since y y and y + 1 y+1 are coprime, y y must divide 2 x 2x . Since x x and c c are coprime, c c must divide 2 ( y + 1 ) 2(y+1) . There are two cases to consider:

  • If c = y + 1 c=y+1 , then y = 2 x y=2x and b = y 2 y 1 b=y^2-y-1 . The inequality reads ( y 2 y 1 ) ( y + 2 ) < 2 ( y + 1 ) 2 (y^2-y-1)(y+2) < 2(y+1)^2 , and the only positive integer solutions of this inequality are y = 1 , 2 , 3 y=1,2,3 . Since y = 2 x y=2x , we must have y = 2 y=2 . But this implies that b = 1 < 3 = c b = 1 < 3 = c , which is impossible.
  • If c = 2 ( y + 1 ) c=2(y+1) , then y = x y=x and b = 2 ( y 2 y 1 ) b=2(y^2-y-1) .The inequality again tells us that ( y 2 y 1 ) ( y + 2 ) < 2 ( y + 1 ) 2 (y^2-y-1)(y+2) < 2(y+1)^2 , and so y = 1 , 2 , 3 y=1,2,3 are the only possibilities. The case y = 1 y=1 gives b = 2 < 0 b=-2 < 0 . The case y = 2 y=2 gives b = 2 < 6 = c b = 2 < 6 = c . Thus we must have y = 3 y=3 , and so x = 3 x=3 , b = 10 b=10 , c = 8 c=8 .

Thus the area A B C D ABCD is 1 2 × b × 2 x ( y + 1 ) 1 2 × c × x ( y + 1 ) = 1 2 ( 2 b c ) x ( y + 1 ) = 72 \tfrac12 \times b \times 2x(y+1) - \tfrac12 \times c \times x(y+1) \; = \; \tfrac12(2b-c)x(y+1) \; = \; \boxed{72}

Considering the right triangle C F G CFG , it has all integer side lengths, making them Pythagorean triples. Then suppose C F = a CF = a ; F G = b FG = b ; and G C = c GC = c for some integers a , b , c a,b,c .

From the first part, we know that A F = C G F C = c a AF = CG - FC = c-a .

Let then B E = h BE = h , and h + a = 2 ( c a ) h + a = 2(c-a) . h = 2 c 3 a h = 2c - 3a .

We also know that h > a h > a , so 2 c 3 a > a 2c - 3a > a . Hence, c > 2 a c > 2a .

The upper bound of h h happens when we create the triangle A H G AHG to C F G CFG with H A G \angle HAG as a right angle.

Then with similarity ratio, A H = h m a x = a b ( A H ) = a ( c a + b ) b > h AH = h_{max} = \dfrac{a}{b}(AH) = \dfrac{a(c-a+b)}{b} > h .

Thus, a ( c a + b ) b > 2 c 3 a \dfrac{a(c-a+b)}{b} > 2c - 3a .

a c a 2 > 2 c b 4 a b ac - a^2 > 2cb - 4ab

4 a b a 2 > c ( 2 b a ) 4ab - a^2 > c(2b-a)

4 a b a 2 2 b a > c \dfrac{4ab - a^2}{2b-a} > c

2 a + a 2 2 b a = a ( 2 + a 2 b a ) = a ( 2 + 1 2 b a 1 ) > c 2a + \dfrac{a^2}{2b-a} = a(2 + \dfrac{a}{2b-a}) = a(2 + \dfrac{1}{2\frac{b}{a}-1}) > c

Since b > a b > a , then b a > 1 \dfrac{b}{a} >1 .

2 b a 1 > 1 \dfrac{2b}{a} - 1 >1 .

Thus, 1 2 b a 1 < 1 \dfrac{1}{2\frac{b}{a}-1} < 1 , and a ( 2 + 1 2 b a 1 ) < 3 a a(2 + \dfrac{1}{2\frac{b}{a}-1}) < 3a .

Therefore, 3 a > a ( 2 + 1 2 b a 1 ) > c > 2 a 3a > a(2 + \dfrac{1}{2\frac{b}{a}-1}) > c > 2a . In other words, 3 > c a > 2 3 > \dfrac{c}{a} > 2 .

Now from F D A F FD | AF , we can rewrite as: a + b c 2 c a \dfrac{a+b-c}{2} | c-a . According to Euclid's formula, we can parametize c = m 2 + n 2 c= m^2 + n^2 ; a = 2 m n a = 2mn or b = m 2 n 2 b = m^2 - n^2 or vice versa, depending on which one is even for some co-prime integers m > n m>n .

Then m 2 n 2 + 2 m n ( m 2 + n 2 ) 2 c a \dfrac{m^2-n^2+2mn - (m^2+n^2)}{2} | c-a .

m n n 2 c a mn - n^2 | c-a .

n ( m n ) c a n(m-n) | c-a .

For odd a = m 2 n 2 a = m^2 - n^2 , c a = 2 n 2 c-a = 2n^2 . Then m n 2 n m-n | 2n . Since m n m-n is odd, then m n n m-n| n .

Hence, k ( m n ) = n k(m-n) = n for some integer k k . k m = ( k + 1 ) n km = (k+1)n . However, g c d ( k , k + 1 ) = 1 gcd(k, k+1) = 1 for any k k , so k n k|n . And since g c d ( m , n ) = 1 gcd(m, n) = 1 , then n k n|k .

That means n = k n = k and m = k + 1 = n + 1 m = k+1 = n+1 .

Returning to the inequality 3 > c a > 2 3 > \dfrac{c}{a} > 2 , we can substitute in terms of m , n m,n :

3 > m 2 + n 2 m 2 n 2 > 2 3 > \dfrac{m^2 + n^2}{m^2 - n^2} > 2 . Then 3 > ( n + 1 ) 2 + n 2 ( n + 1 ) 2 n 2 > 2 3 > \dfrac{(n+1)^2 + n^2}{(n+1)^2 - n^2} > 2 .

3 > 2 n 2 + 2 n + 1 2 n + 1 > 2 3 > \dfrac{2n^2 + 2n +1}{2n+1} > 2 .

2 > 2 n 2 2 n + 1 > 1 2 > \dfrac{2n^2}{2n+1} >1

Solving for each side, 2 n 2 > 2 n + 1 2n^2 > 2n+1 . 2 n 2 2 n 1 > 0 2n^2 -2n-1>0 . Solving the quadratic formula, we will obtain n > 1 + 3 2 1.37 n > \dfrac{1+\sqrt{3}}{2} \approx 1.37 or n 2 n \geq 2 .

On the other hand, 2 ( 2 n + 1 ) > 2 n 2 2(2n+1) > 2n^2 . 0 > n 2 2 n 1 0 > n^2 - 2n -1 . Solving the quadratic formula, we will obtain n < 1 + 2 2.41 n < 1+\sqrt{2} \approx 2.41 or n 2 n \leq 2 .

The only possible value is n = 2 n= 2 and, thus, m = 3 m = 3 , and Pythagorean triple of ( 5 , 12 , 13 ) (5, 12,13) . However, with this, h = 2 c 3 a = 26 15 = 11 h = 2c- 3a = 26 - 15 = 11 , but 11 > h m a x = a ( c a + b ) b = 5 ( 13 5 + 12 ) 12 8.33 11 > h_{max} = \dfrac{a(c-a+b)}{b} = \dfrac{5(13-5+12)}{12} \approx 8.33 , which is not applicable.

That means a = 2 m n a = 2mn is even, and so n ( m n ) c a = ( m 2 + n 2 ) 2 m n = ( m n ) 2 n(m-n)| c-a = (m^2+n^2) - 2mn = (m-n)^2 . Hence, n m n n| m-n and n m n|m .

The only applicable value is n = 1 n=1 .

Then similarly for the inequality 3 > c a > 2 3 > \dfrac{c}{a} > 2 , we can substitute in terms of m , n m,n :

3 > m 2 + 1 2 m > 2 3 > \dfrac{m^2 + 1}{2m} > 2 .

Solving for each side, m 2 + 1 > 4 m m^2 +1 > 4m . m 2 4 m + 1 > 0 m^2 -4m+1>0 . Solving the quadratic formula, we will obtain m > 2 + 3 3.73 m > 2+\sqrt{3} \approx 3.73 or m 4 m \geq 4 .

On the other hand, m 2 + 1 < 6 m m^2 +1 < 6m . m 2 6 m + 1 < 0 m^2 -6m+1 < 0 . Solving the quadratic formula, we will obtain m < 3 + 2 2 5.83 m < 3+ 2\sqrt{2} \approx 5.83 or m 5 m \leq 5 .

Only m = 4 , 5 m = 4,5 works, but since n = 1 n=1 is odd, then m m must be even and equals to 4 4 .

That leads to the Pythagorean triple of ( 8 , 15 , 17 ) (8, 15,17) . h = 2 c 3 a = 34 24 = 10 h = 2c- 3a = 34 - 24 = 10 .

A F = 9 AF = 9 . F D = 9 + 15 2 9 = 3 FD = \dfrac{9+15}{2} - 9 = 3 , and 3 9 3|9 . A G = 24 AG = 24 , and D G = 12 DG =12 .

Finally, the area of the quadrilateral A B C D = 10 2 ( 24 ) 8 2 ( 12 ) = 120 48 = 72 ABCD = \dfrac{10}{2}(24) - \dfrac{8}{2}(12) = 120 - 48 = \boxed{72} .

Let me do say that this problem has been a particularly awkward one. Like trying to untangle an horrific knot of Xmas lights.

Michael Mendrin - 3 years, 8 months ago

Log in to reply

Oh...is it too hard to understand?

Worranat Pakornrat - 3 years, 8 months ago

Log in to reply

There's nothing wrong with the way you've worded your problem. Or your solution for that matter. It's just very tricky.

Michael Mendrin - 3 years, 8 months ago

Log in to reply

@Michael Mendrin Yes, I agree and think it's special to see the unique ratio for this kind of shape.

Worranat Pakornrat - 3 years, 8 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...