Is parametrization necessary?

I want to construct a rectangle such that

  • it has integer dimensions a × b a \times b that differ by exactly 1 ( i.e. a b = 1 ) ; (\text{i.e. }|a-b|=1);
  • its diagonal also has an integer length.

The diagram to the right shows such a rectangle.

Is there any other way to construct a rectangle satisfying the above properties?

Yes, infinitely many ways Yes, but only finitely many ways No, there are no other ways

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.

7 solutions

Arjen Vreugdenhil
Dec 18, 2017

We wish to construct a Pythagorean triple a 2 + b 2 = c 2 a^2 + b^2 = c^2 with b = a ± 1 b = a \pm 1 .

It is well-known that Pythagorean triples can be generated by choosing integers p > q > 0 p > q > 0 and setting a = p 2 q 2 , b = 2 p q , c = p 2 + q 2 . a = p^2 - q^2,\ b = 2pq,\ c = p^2 + q^2. With ( p , q ) = ( 2 , 1 ) (p,q) = (2,1) we get the given triple ( 3 , 4 , 5 ) (3,4,5) . Also, set r = p q r = p - q ; then r 2 = c b r^2 = c - b .

In order for b = a ± 1 b = a \pm 1 , we must have ( p q ) 2 = 2 q 2 ± 1 (p - q)^2 = 2q^2 \pm 1 . Rewrite this as r 2 2 q 2 = ± 1 r^2 - 2q^2 = \pm 1 . This is Pell's equation.

It is know that the positive Pell's equation has infinitely many solutions ( q n , r n ) (q_n,r_n) . The integers r n / q n r_n/q_n form a sequence with limit 2 \sqrt 2 . The first few terms are r n / q n a 2 + b 2 = c 2 3 / 2 2 1 2 + 2 0 2 = 2 9 2 17 / 12 69 5 2 + 69 6 2 = 98 5 2 99 / 70 2366 1 2 + 2366 0 2 = 3346 1 2 \begin{array}{ll} r_n/q_n & a^2 + b^2 = c^2 \\ \hline 3/2 & 21^2 + 20^2 = 29^2 \\ 17/12 & 695^2 + 696^2 = 985^2 \\ 99/70 & 23661^2 + 23660^2 = 33461^2 \end{array}

Each next term may be generated recursively as ( r , q ) = ( 3 r + 4 q , 2 r + 3 q ) (r',q') = (3r + 4q, 2r + 3q) . Since this process can be continued indefinitely, there are infinitely many solutions.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
r = 1
q = 0
repeat
   new_r = 3*r + 4*q
   new_q = 2*r + 3*q
   r = new_r
   q = new_q

   p = r + q
   a = p*p - q*q
   b = 2*p*q
   c = p*p + q*q

   print "sides " + a + ", " + b + "; diagonal: " + c
end repeat

Note: You could have jumped to Pell's equation directly without using the Pythagorean triple substitution.

We have ( 2 a + 1 ) 2 2 c 2 = 1 (2a+1)^2 - 2c^2= -1 , which has infinitely many solutions. 7 2 2 × 5 2 = 1 7^2 - 2 \times 5^2 = -1 gives us the ( 3 , 4 , 5 ) (3,4,5) solution.

Calvin Lin Staff - 3 years, 5 months ago

Log in to reply

Interesting! The title of the problem gave a hint, I now realize.

Arjen Vreugdenhil - 3 years, 5 months ago

The second triple is wrong, the correct one is 696^2+697^2=985^2

Gaetano Di Caprio - 3 years, 5 months ago
Leonel Castillo
Dec 28, 2017

I will solve by reducing the problem to solving Pell's equation. Lets consider the polynomial f ( n ) = n 2 + ( n + 1 ) 2 f(n) = n^2 + (n+1)^2 . This polynomial happens to have an interesting property that we will exploit. We wish to solve the equation f ( n ) = b 2 f(n) = b^2 in order to find the side lengths of a square with the properties given in the problem. If we solve for positive n n we get that n = 2 b 2 1 1 2 n = \frac{\sqrt{2b^2 - 1} - 1}{2} .

Let's for a moment suppose that 2 b 2 1 2b^2 - 1 is a perfect square. 2 b 2 1 2b^2 - 1 is an odd number, and therefore its square root will also be odd and if we subtract 1 to this odd number we get an even number. Therefore if 2 b 2 1 2b^2 - 1 is a perfect square then n n is an integer, and as long as b > 1 b>1 , n will be larger than 0. This means we have no difficulties in studying the discriminant instead of the polynomial. Therefore we now want to solve the equation 2 b 2 1 = m 2 2b^2 - 1 = m^2 to find when this discriminant is a perfect square. Arranging this properly we get m 2 2 b 2 = 1 m^2 - 2b^2 = -1 . This is Pell's equation, for which we may obtain infinitely many solutions.

David Vreken
Dec 27, 2017

One subset of rectangles with the given properties has positive integer side lengths of a a and a + 1 a + 1 and a positive integer diagonal of c c such that it satisfies Pythagorean’s Theorem a 2 + ( a + 1 ) 2 = c 2 a^2 + (a + 1)^2 = c^2 . This can be rearranged to ( 2 a + 1 ) 2 + 1 = 2 c 2 (2a + 1)^2 + 1 = 2c^2 , which is in the form of x 2 + 1 = 2 y 2 x^2 + 1 = 2y^2 for positive integers x = 2 a + 1 x = 2a + 1 and y = c y = c .

We can generate Pythagorean triples ( a , b , c ) (a’, b’, c’) by a Pythagorean generator for which a = p 2 q 2 a’ = p^2 - q^2 , b = 2 p q b’ = 2pq , and c = p 2 + q 2 c’ = p^2 + q^2 for positive integers p p and q q where p > q > 0 p > q > 0 , but in this case we will also place the given restriction that b = a + 1 b’ = a’ + 1 , so that 2 p q = p 2 q 2 + 1 2pq = p^2 - q^2 + 1 which can be rearranged to ( p q ) 2 + 1 = 2 q 2 (p - q)^2 + 1 = 2q^2 , which is also in the form of x 2 + 1 = 2 y 2 x^2 + 1 = 2y^2 but for positive integers x = p q x = p - q and y = q y = q .

Using these similar forms we can set y = q = c y = q = c and x = p q = 2 a + 1 x = p - q = 2a + 1 in an attempt to create a recursive equation for a a , b b , and c c . Then q = c q = c and p = 2 a + 1 + q = 2 a + 1 + c p = 2a + 1 + q = 2a + 1 + c . Substituting these into a a’ , b b’ , and c c’ and simplifying (and using the above equation ( 2 a + 1 ) 2 + 1 = 2 c 2 (2a + 1)^2 + 1 = 2c^2 ) gives:

a = 2 c 2 + 4 a c + 2 c 1 a’ = 2c^2 + 4ac + 2c - 1

b = 2 c 2 + 4 a c + 2 c b’ = 2c^2 + 4ac + 2c

c = 3 c 2 + 4 a c + 2 c 1 c’ = 3c^2 + 4ac + 2c - 1

First of all, a a’ , b b’ , and c c’ must be integer sides of a right triangle because they were generated by a Pythagorean generator. Secondly, the above equations for a a’ and b b’ show that b = a + 1 b’ = a’ + 1 . Finally, if a a , b b , and c c are positive integers, then a > a a’ > a , b > b b’ > b , and c > c c’ > c , which ensures that each triplet generated is bigger than the last.

Therefore, the above equations for a a’ , b b’ , and c c’ are recursive equations that can be continued indefinitely. (Using the given example ( a , b , c ) = ( 3 , 4 , 5 ) (a, b, c) = (3, 4, 5) as the seed, we can obtain the triplets ( 119 , 120 , 169 ) (119, 120, 169) , ( 137903 , 137904 , 195025 ) (137903, 137904, 195025) , and so on.) There are therefore infinitely many ways to construct a rectangle satisfying the given properties.

Since x 2 + 1 = 2 y 2 x^2 + 1 = 2y^2 is a Pell-type equation it can be written recursively as x = 3 x + 4 y x' = 3x + 4y and y = 2 x + 3 y y' = 2x + 3y . Since x = 2 a + 1 x = 2a + 1 and y = c y = c , then ( 2 a + 1 ) = 3 ( 2 a + 1 ) + 4 c (2a + 1)' = 3(2a + 1) + 4c and c = 2 ( 2 a + 1 ) + 3 c c' = 2(2a + 1) + 3c . Also, b = a + 1 b' = a' + 1 . These all simplify to

a = 3 a + 2 c + 1 a' = 3a + 2c + 1

b = 3 a + 2 c + 2 b' = 3a + 2c + 2

c = 4 a + 3 c + 2 c' = 4a + 3c + 2

which are alternative recursive equations that can also be continued indefinitely.

David Vreken - 3 years, 5 months ago

Yes, infinitely many ways. If we denote n,n+1 the dimensions, then we have: *cos(x)=(n+1)/k *sin(x)=n/k where k is the hypotenuse of the n,n+1 right triangle , we note that cos(x)-sin(x)=1/k. So let f(x)=cos(x)-sin(x), f is a continue function, and reaches the interval [0,1], so there existe x in R such that f(x)=1/k.

SoufYane Mit - 3 years, 4 months ago

Log in to reply

Unless I am missing something, this doesn't prove that n is an integer.

David Vreken - 3 years, 4 months ago
F R
Dec 25, 2017

Although there is no limiting factor and we can keep generating numbers infinitely fulfilling the given conditions.

But there are only 8 sets in a million and 12 up to a billion numbers. I am not sure whether we should call it "infinitely many ways" OR "finitely many ways". Let us hear from gurus.

Thanks

Like there are infinitely many perfect numbers, sometimes patterns like this ARE infinite.

Shaun Strudwick - 3 years, 5 months ago

Log in to reply

Note: It is not known if there are infinitely many perfect numbers.

Calvin Lin Staff - 3 years, 5 months ago

The triple in the last row is wrong. The correct triple is 927538920, 927538921, 1311738121

Gaetano Di Caprio - 3 years, 5 months ago
Gaetano Di Caprio
Dec 30, 2017

We are looking for integers x , z x,z such that:

x 2 + ( x + 1 ) 2 = z 2 x^2+(x+1)^2=z^2

which is equivalent to

2 x 2 + 2 x + ( 1 z 2 ) = 0 2x^2+2x+(1-z^2 )=0

Solving for x x , we get the positive solution:

x = 2 z 2 1 1 2 \displaystyle x=\frac{\sqrt{2z^2-1}-1}{2}

In order to obtain an integer solution, the discriminant of the equation must be a perfect square, i.e. there must exist t Z t \in \mathbb{Z} such that:

2 z 2 1 = t 2 2z^2-1=t^2

The latter can be rewritten as

t 2 2 z 2 = 1 t^2-2z^2=-1

which is Pell's equation. All the solutions ( t n , z n ) (t_n,z_n) to this equation can be calculated using the relation:

( 1 + 2 ) 2 n + 1 = t n + z n 2 (1+\sqrt{2})^{2n+1}=t_n+z_n\sqrt{2}

for n = 1 , 2 , 3 , . . . n=1,2,3,...

As a consequence, all the (infinitely many) Pythagorean triples ( x n , y n , z n ) (x_n,y_n,z_n) , with y n = x n + 1 y_n=x_n+1 , can be generated by the formulas ( n = 1 , 2 , 3 , n=1,2,3,… ):

x n = ( 1 + 2 ) 2 n + 1 + ( 1 2 ) 2 n + 1 4 1 2 \displaystyle x_n=\frac{(1+\sqrt{2})^{2n+1}+(1-\sqrt{2})^{2n+1}}{4}-\frac{1}{2}

y n = x n + 1 y_n=x_n+1

z n = ( 1 + 2 ) 2 n + 1 ( 1 2 ) 2 n + 1 2 2 \displaystyle z_n=\frac{(1+\sqrt{2})^{2n+1}-(1-\sqrt{2})^{2n+1}}{2\sqrt{2}}

For n = 1 , 2 , , 20 n=1,2,…, 20 we obtain the following:

\( \begin{array}{} x_n & y_n & z_n \\ 3 & 4 & 5 \\ 20 & 21 & 29 \\ 119 & 120 & 169 \\ 696 & 697 & 985 \\ 4059 & 4060 & 5741 \\ 23660 & 23661 & 33461 \\ 137903 & 137904 & 195025 \\ 803760 & 803761 & 1136689 \\ 4684659 & 4684660 & 6625109 \\ 27304196 & 27304197 & 38613965 \\ 159140519 & 159140520 & 225058681 \\ 927538920 & 927538921 & 1311738121 \\ 5406093003 & 5406093004 & 7645370045 \\ 31509019100 & 31509019101 & 44560482149 \\ 183648021599 & 183648021600 & 259717522849 \\ 1070379110496 & 1070379110497 & 1513744654945 \\ 6238626641379 & 6238626641380 & 8822750406821 \\ 36361380737780 & 36361380737781 & 51422757785981 \\ 211929657785303 & 211929657785304 & 299713796309064 \\ \end{array} \)

I got it totally wrong. It is clear that there are infinitely many triangles with that property. I thought constructing means creating a triangle using ruler and compass. Suppose we have a triangle with side lengths 3,4and 5.we can draw a line AB of length 5.with centre A and radius 3units draw an arc. Then with centre B and radius 4.draw an arc intersecting the first arc. You get a triangle. There are other ways of constructing the triangles.

Srikanth Tupurani - 3 years, 2 months ago
_Abledozen _
Dec 26, 2017

x² + x² + 1 = y² where all values are integers 2x² + 1 = y² therefore y = sqrt(2x² + 1) for every x there must be some corresponding y value

That doesn't assure y is an integer.

Benjamin Keilty - 3 years, 5 months ago

Also, I don't think it would be x^2 + 1, but (x+1)^2.

David Ortiz - 3 years, 5 months ago

a^2 + b^2 = c^2 => Every three sets of integers(a,b,c) following this condition are called Pythagorean triplets. I.e. 3^2 +4^2 =5^2 => (3,4,5) .......... We can generate numbers infinitely which satisfy the theorem. So the answer is, "yes, infinitely many ways"

Note that you need to satisfy the condition " a and b differ by exactly 1".

Calvin Lin Staff - 3 years, 5 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...