Sum With Four Squares

If a a , b b , c c , and d d are distinct pairwise co-prime positive integers such that a 2 + b 2 = c 2 + d 2 a^2 + b^2 = c^2 + d^2 , find the lowest possible value of a + b + c + d a + b + c + d .


The answer is 32.

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.

3 solutions

Chris Lewis
Apr 24, 2019

Some observations: since the integers a , b , c , d a,b,c,d are pairwise* coprime, at most one of them is even; so none of them is.

Similarly, assume exactly one of the integers is a multiple of 3 3 . Then the other three are not multiples of 3 3 , and reducing modulo 3 3 shows that we again reach a contradiction. So there are no multiples of 3 3 .

We can even do this modulo 5 5 : assume that a a is a multiple of 5 5 . Then b 2 c 2 + d 2 ( m o d 5 ) b^2 \equiv c^2+d^2 \pmod5 . But the quadratic residues modulo 5 5 are 0 0 , 1 1 and 4 4 , and none of b , c , d b,c,d can be a multiple of 5 5 ; so our initial assumption was once again wrong, and there are no multiples of 5 5 among the four integers.

Multiples of 2 2 , 3 3 and 5 5 aren't allowed. That means the integers must belong to the set { 1 , 7 , 11 , 13 , 17 , } \{1,7,11,13,17,\ldots\} , and we can stop searching here; it turns out the smallest possible set ( a , b , c , d ) = ( 1 , 13 , 7 , 11 ) (a,b,c,d)=(1,13,7,11) works, with total 32 \boxed{32} .

Incidentally, the next smallest set ( 1 , 17 , 11 , 13 ) (1,17,11,13) also satisfies the equation.

*the pairwise condition is needed - if the requirement is just that gcd ( a , b , c , d ) = 1 \gcd(a,b,c,d)=1 then the set ( a , b , c , d ) = ( 1 , 8 , 4 , 7 ) (a,b,c,d)=(1,8,4,7) works, with total 20 20 .

Chris Lewis - 2 years, 1 month ago

Great solution! Completely different than my approach, but I like yours better. (I'll post my solution later.)

David Vreken - 2 years, 1 month ago

Log in to reply

Thanks! Very interested to see a different solution, as ever. Mine would not have worked very well if the first valid set had totalled, say, 1000 1000 ...

(To be honest, my actual solution was to just code it and search, but I'm always keen to try and find a way without doing that. I was really surprised that eliminating the small prime factors left a set that worked straight away.)

Chris Lewis - 2 years, 1 month ago

Log in to reply

I posted my solution. Basically I rearranged the equation to be a difference of squares and factored it, and then used some clues and then trial and error to arrive at the correct answer.

David Vreken - 2 years, 1 month ago

I did this the same way, but at first forgot about 1 1 .

Taking 7 7 as the smallest prime, and looking at QRs (mod 7) (which are 0,1,2,4), we note that we must have 0 + 2 = 1 + 1 0 + 2 = 1 + 1 or 0 + 4 = 2 + 2 0 + 4 = 2 + 2

1 : 1 , , 11 : 2 , , 13 : 1 , , 17 : 2 , , 19 : 4 1: 1,, 11: 2,, 13: 1,, 17: 2,, 19: 4 First possibility is 7 2 + 1 1 2 = 1 2 + 1 3 2 7^2 + 11^2 = 1^2 + 13^2 . Total is 32. Second possibility is 7 2 + 1 9 2 = 1 1 2 + 1 7 2 7^2 + 19^2 = 11^2 + 17^2 . Total is 54.

With the smallest prime being 11 11 , QR of (1,11,13,17) (mod 11) are (1,0,4,3). And 1 1 2 + 1 3 2 = 1 2 + 1 7 2 11^2 + 13^2 = 1^2 + 17^2 . Total is 42.

Alex Burgess - 2 years, 1 month ago

I forgot that they had to be coprime and just put 7,24,15,25 that where the first ones y got in my mind

Pablo Sanchez - 2 years, 1 month ago
David Vreken
Apr 24, 2019

If one of the numbers is even, then the other numbers (as distinct pairwise co-prime integers) must be odd, but then the sum of squares with the even number is odd and the sum of the squares without the even number is even. Therefore, none of the numbers can be even, so all of the numbers must be odd.

a 2 + b 2 = c 2 + d 2 a^2 + b^2 = c^2 + d^2 can be rearranged to a 2 c 2 = d 2 b 2 a^2 - c^2 = d^2 - b^2 . Since the square of any odd number is 1 ( m o d 8 ) 1 \pmod {8} , a 2 c 2 a^2 - c^2 (and d 2 b 2 d^2 - b^2 ) is divisible by 8 8 . a 2 c 2 = d 2 b 2 a^2 - c^2 = d^2 - b^2 can further be rearranged to ( a c ) ( a + c ) = ( d b ) ( d + b ) (a - c)(a + c) = (d - b)(d + b) , and since all the numbers are odd, each factor a c a - c , a + c a + c , d b d - b , and d + b d + b must be even.

Therefore, our search is narrowed down to testing multiples of 8 8 that can be written as a product of two even numbers in at least two different ways. For example, testing 24 24 we can have 2 12 2 \cdot 12 and 4 6 4 \cdot 6 , for a = 7 a = 7 , c = 5 c = 5 , d = 5 d = 5 and b = 1 b = 1 , but unfortunately in this case c c and d d are not distinct. The table below summarizes the first few possibilities.

The first number that gives us a correct result is 48 = 2 24 = 6 8 48 = 2 \cdot 24 = 6 \cdot 8 , for a = 13 a = 13 , c = 11 c = 11 , d = 7 d = 7 , and b = 1 b = 1 , so that 1 3 2 + 1 2 = 1 1 2 + 7 2 = 170 13^2 + 1^2 = 11^2 + 7^2 = 170 , which results in the lowest value for a + b + c + d a + b + c + d of 13 + 1 + 11 + 7 = 32 13 + 1 + 11 + 7 = \boxed{32} .

Very nice! I tried the differences of two squares idea but didn't get anywhere with it. Nice to see it can lead to a systematic approach. I wonder if there's any way that doesn't involve case analysis/trial and error. It seems there might be a way to construct quadruples that satisfy the equation in a way similar to Euclid's method for constructing primitive Pythagorean triples, but I don't quite know how.

Chris Lewis - 2 years, 1 month ago

Chris Lewis' solution is correct.

Union [ a + b + c + d /. Solve [ ( a b c d ) Z > 0 gcd ( a , b ) = 1 gcd ( a , c ) = 1 gcd ( a , d ) = 1 gcd ( b , c ) = 1 gcd ( b , d ) = 1 gcd ( c , d ) = 1 a 2 + b 2 = c 2 + d 2 a < 15 b < 15 c < 15 d < 15 a b a c a d b c b d c d ] ] [ [ 1 ] ] 32 \text{Union}\left[a+b+c+d\text{/.}\, \text{Solve}\left[(a|b|c|d)\in \mathbb{Z}_{>\, 0}\land \gcd (a,b)=1\land \gcd (a,c)=1\land \gcd (a,d)=1\land \gcd (b,c)=1\land \gcd (b,d)=1\land \gcd (c,d)=1\land \\ a^2+b^2=c^2+d^2\land a<15\land b<15\land c<15\land d<15\land a\neq b\land a\neq c\land a\neq d\land b\neq c\land b\neq d\land c\neq d\right]\right][[1]] \Rightarrow 32

Solutions with value less than or equal to 24:

( 1 13 7 11 1 13 11 7 1 17 11 13 1 17 13 11 1 23 13 19 1 23 19 13 7 11 1 13 7 11 13 1 7 19 11 17 7 19 17 11 11 7 1 13 11 7 13 1 11 13 1 17 11 13 17 1 11 17 7 19 11 17 19 7 11 23 17 19 11 23 19 17 13 1 7 11 13 1 11 7 13 11 1 17 13 11 17 1 13 19 1 23 13 19 23 1 17 1 11 13 17 1 13 11 17 11 7 19 17 11 19 7 17 19 11 23 17 19 23 11 19 7 11 17 19 7 17 11 19 13 1 23 19 13 23 1 19 17 11 23 19 17 23 11 23 1 13 19 23 1 19 13 23 11 17 19 23 11 19 17 ) \left( \begin{array}{cccc} 1 & 13 & 7 & 11 \\ 1 & 13 & 11 & 7 \\ 1 & 17 & 11 & 13 \\ 1 & 17 & 13 & 11 \\ 1 & 23 & 13 & 19 \\ 1 & 23 & 19 & 13 \\ 7 & 11 & 1 & 13 \\ 7 & 11 & 13 & 1 \\ 7 & 19 & 11 & 17 \\ 7 & 19 & 17 & 11 \\ 11 & 7 & 1 & 13 \\ 11 & 7 & 13 & 1 \\ 11 & 13 & 1 & 17 \\ 11 & 13 & 17 & 1 \\ 11 & 17 & 7 & 19 \\ 11 & 17 & 19 & 7 \\ 11 & 23 & 17 & 19 \\ 11 & 23 & 19 & 17 \\ 13 & 1 & 7 & 11 \\ 13 & 1 & 11 & 7 \\ 13 & 11 & 1 & 17 \\ 13 & 11 & 17 & 1 \\ 13 & 19 & 1 & 23 \\ 13 & 19 & 23 & 1 \\ 17 & 1 & 11 & 13 \\ 17 & 1 & 13 & 11 \\ 17 & 11 & 7 & 19 \\ 17 & 11 & 19 & 7 \\ 17 & 19 & 11 & 23 \\ 17 & 19 & 23 & 11 \\ 19 & 7 & 11 & 17 \\ 19 & 7 & 17 & 11 \\ 19 & 13 & 1 & 23 \\ 19 & 13 & 23 & 1 \\ 19 & 17 & 11 & 23 \\ 19 & 17 & 23 & 11 \\ 23 & 1 & 13 & 19 \\ 23 & 1 & 19 & 13 \\ 23 & 11 & 17 & 19 \\ 23 & 11 & 19 & 17 \\ \end{array} \right)

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...