How many ordered pairs of positive integers ( a , b ) are there such that both a 2 + 3 b and b 2 + 3 a are perfect squares?
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.
There are several ways to bound the values in the question. The idea of bounding a quality that's supposed to be a square between 2 known squares helps to greatly restrict the possibilities.
This question was used in the Live Challenge , in which many people missed the solution ( 1 , 1 ) . IN the comments discussion, most believed that there were only 1 solution and worked hard to prove it (unsuccessfully of course).
Mistakes
AWESOME AND BRILLIANT
Let a 2 + 3 b = ( a + x ) 2 , b 2 + 3 a = ( b + y ) 2 [ x , y ≥ 1 ; x , y ∈ N ] . Adding the two equations we get 3 a + 3 b = 2 a x + 2 b y + x 2 + y 2 . If both x , y ≥ 2 then 2 a x + 2 b y > 3 a + 3 b ⇒ 2 a x + 2 b y + x 2 + y 2 > 3 a + 3 b which is a contradiction. So at least one of x , y is 1 . WLOG we assume y = 1 .[Note: We are taking WLOG but we have to find ordered pair of solutions. So if ( p , q ) is a possible pair where p = q then ( q , p ) will be a solution too.] Now 3 a = 2 b + 1 ⇒ b = 2 3 a − 1 , 3 b = 2 a x + x 2 . Plugging the first value in the second equation we get 9 a − 4 a x = 2 x 2 + 3 . If x > 2 , then LHS becomes negative but RHS remains positive. So x = 1 , 2 . Solving for ( x , y ) = ( 1 , 1 ) , ( 2 , 1 ) yields ( a , b ) = ( 1 , 1 ) , ( 1 1 , 1 6 ) . Since 1 1 = 1 6 , ( a , b ) = ( 1 6 , 1 1 ) is another soluton. Hence we have three ordered pairs i.e. ( a , b ) = ( 1 , 1 ) , ( 1 1 , 1 6 ) , ( 1 6 , 1 1 ) .
Without loss of generality, assume 0 < a ≤ b . Hence, b 2 < b 2 + 3 a < b 2 + 4 b + 4 = ( b + 2 ) 2 so it must be that b 2 + 3 a = ( b + 1 ) 2 if it is a perfect square. Therefore, 3 a = 2 b + 1 ⇒ b = 2 3 a − 1 < 2 a so a 2 < a 2 + 3 b < a 2 + 6 a < ( a + 3 ) 2 ⇒ a 2 + 3 b = ( a + 1 ) 2 or ( a + 2 ) 2 .
Case 1: a 2 + 3 b = ( a + 1 ) 2 = a 2 + 2 a + 1 By simultaneously solving 3 b = 2 a + 1 and 3 a = 2 b + 1 from above, we have a = b = 1 .
Case 2: a 2 + 3 b = ( a + 2 ) 2 = a 2 + 4 a + 4 By simultaneously solving 3 b = 4 a + 4 and 3 a = 2 b + 1 from above, we have a = 1 1 , b = 1 6 .
Therefore, there are a total of 3 ordered pairs, ( a , b ) = ( 1 , 1 ) , ( 1 1 , 1 6 ) , ( 1 6 , 1 1 )
We can see that x 2 + 3 y < ( x + 2 ) 2 or y 2 + 3 x < ( y + 2 ) because if both of inequalities are not true, we have x 2 + 3 y ≥ ( x + 2 ) 2 and y 2 + 3 x ≥ ( y + 2 ) , then x 2 + y 2 + 3 x + 3 y ≥ x 2 + y 2 + 4 ( x + y ) + 8 leads to 0 ≥ x + y + 8 contradiction with x , y are positive integers.
Without loss of generality, we can assume that x 2 + 3 y < ( x + 2 ) 2 , but it is easy to see x 2 < x 2 + 3 y , hence, x 2 < x 2 + 3 y < ( x + 2 ) 2 and x 2 + 3 y is a perfect square. So x 2 + 3 y can only be ( x + 1 ) 2 or 3 y = 2 x + 1 . Solve this Diophantine equation, we get x = 3 k + 1 , y = 2 k + 1 with k ∈ N .
Continue, we have y 2 + 3 x = 4 k 2 + 1 3 k + 4 and with k > 5 , we have ( 2 k + 3 ) 2 < 4 k 2 + 1 3 k + 4 < ( 2 k + 4 ) 2 , which implies that y 2 + 3 x can not be a perfect square. Therefore, k ∈ { 0 , 1 , 2 , 3 , 4 , 5 } .
Consider each of six cases, we obtain k = 0 and k = 5 : with k = 0 , we have ( x , y ) = ( 1 , 1 ) and with k = 5 , we have ( x , y ) = ( 1 1 , 1 6 ) , ( 1 6 , 1 1 ) .
So in summary, there are 3 ordered pairs of positive integers satisfy all of the given conditions.
WLOG, let a ≤ b (we will account for the ordered pairs after). Then note that b 2 < b 2 + 3 a < b 2 + 4 b + 4 = ( b + 2 ) 2 , so for b 2 + 3 a to be a perfect square, b 2 + 3 a = ( b + 1 ) 2 = b 2 + 2 b + 1 ⟹ 3 a = 2 b + 1 . Taking this equation modulo 3 produces b ≡ 1 ( m o d 3 ) , so let b = 3 c + 1 . Then 3 a = 6 c + 3 ⟹ a = 2 c + 1 .
Therefore, a 2 + 3 b = ( 2 c + 1 ) 2 + 3 ( 3 c + 1 ) = 4 c 2 + 1 3 c + 4 . But note that 4 c 2 + 1 3 c + 4 < 4 c 2 + 1 6 c + 1 6 = ( 2 c + 4 ) 2 , and ( 2 c + 3 ) 2 = 4 c 2 + 1 2 c + 9 < 4 c 2 + 1 3 c + 4 if c > 5 . Hence, for c > 5 , ( 2 c + 3 ) 2 < a 2 + 3 b < ( 2 c + 4 ) 2 , so a 2 + 3 b can't be a perfect square.
If c = 5 , a = 1 1 , b = 1 6 , and 1 2 1 + 4 8 = 1 6 9 and 2 5 6 + 3 3 = 2 8 9 , so a = 1 1 , b = 1 6 works.
If c = 4 , a = 9 , b = 1 3 , but 8 1 + 3 9 = 1 2 0 , which isn't a perfect square.
If c = 3 , a = 7 , b = 1 0 , but 4 9 + 3 0 = 7 9 , which isn't a perfect square.
If c = 2 , a = 5 , b = 7 , but 2 5 + 2 1 = 4 6 , which isn't a perfect square.
If c = 1 , a = 3 , b = 4 , but 9 + 1 2 = 2 1 , which isn't a perfect square.
If c = 0 , a = 1 , b = 1 , and 1 + 3 = 4 , so a = b = 1 works.
Consequently, the only ordered pairs are ( 1 , 1 ) , ( 1 1 , 1 6 ) , ( 1 6 , 1 1 ) , for a total of 3 ordered pairs.
since a^2 +3b is a square of some integer k and a,b are positive therefore
a^2+3b=k^2(where k is an integer >a)
therefore k=a+x (x is a positive integer).
or, k^2=(a+x)^2
or, k^2=a^2+2ax+x^2
or, 2ax+x^2=k^2-a^2
or, 2ax+x^2=3b (since a^2+3b=k^2 therefore k^2-a^2=3b)
or, 2ax-3b+x^2=0 ...............(1)
again since b^2+3a is a square of some integer and a,b are positive therefore b^2+3a=m^2 (where m >b)
therefore m=b+y(y is a positive integer).
or, m^2=b^2+2by+y^2
or, 2by+y^2=m^2-b^2
or, 2by+y^2=3a(since b^2+3a=m^2 therefore 3a=m^2-b^2)
or, -3a+2by+y^2=0 ..................(2)
since a^2+3b and b^2+3a are both square number therefore equation (1)
and (2) holds.
therefore, 2ax-3b+x^2=0
-3a+2by+y^2=0
this implies a/(-3y^2-2yx^2)=b/(-3x^2-2xy^2)=1/(4xy-9)
i.e, a=(3y^2+2x^2y)/(9-4xy) and b=(2xy^2+3x^2)/(9-4xy)
now a,b are positive integer and (3y^2+2x^2y),(3x^2+2xy^2) are positive
because x and y are positive.
therefore 9-4xy>0
therefore (x,y) can be (1,1) or (1,2), or (2,1) .
when x=1,y=1 then a=1,b=1.
when x=1,y=2 then a=11, b=16.
when x=2,y=1 then a=16 and b=11.
therefore (1,1);(11,16);(16,11) are the only three solutions of (a,b).
Set a 2 + 3 b = ( a + c ) 2 and b 2 + 3 a = ( b + d ) 2 , where c and d are positive integers. The equations are equivalent to 3 b = 2 a c + c 2 and 3 a = 2 b d + d 2 . Treating this as a system of linear equations in a and b , we get the solutions a = 9 − 4 c d 2 d c 2 + 3 d 2 and b = 9 − 4 c d 2 c d 2 + 3 c 2 . Since c and d are positive, the numerators are positive. Thus, in order for a and b to be positive, we must have 9 − 4 c d > 0 . As such, c d = 1 or c d = 2 .
If ( c , d ) = ( 1 , 1 ) , we get ( a , b ) = ( 1 , 1 ) . If ( c , d ) = ( 1 , 2 ) , we get ( a , b ) = ( 1 6 , 1 1 ) . If ( c , d ) = ( 2 , 1 ) , we get ( a , b ) = ( 1 1 , 1 6 ) . Hence, there are 3 solutions in all.
let a^2+3b=n^2,and b^2+3a=z^2 add both eqn,(a^2)+(b^2)+3(a+b)=(n^2)+(z^2),as a,b>=1 the minimum value of (n^2)+(z^2)=8. by am-gm inequality,(n^2+z^2)/2>=sqrt(n^2 z^2) and equality holds good only when (n=z) or (n^2=z^2) thus maximum value of (n^2) (z^2)=16. thus from the above conditions we get (1,1) as the only solution.
Comments and replies:
Calvin:
I do not see how you're getting a contradiction anywhere. All that you are saying, is that n 2 + z 2 ≥ 8 , which occurs when n = z = 2 , and I agree with that. That doesn't mean that we cannot have n = 5 0 1 , z = 1 2 3 .
Suresh:
minimum value of n^2+z^2=8,since a,b>=1 also by am-gm inequality i got it as the minimum value of (n^2)+(z^2)=2 n z. thus n*z=4. thus n=2,z=2 is the only set that satisfies all the above constraints thus n^2=4 and z^2=4 which is satisfied only by (a,b)=(1,1)
Calvin:
As I said, your argument doesn't make any sense. You're merely showing that the minimum value is 8, and that doesn't mean that the value can only be 8. Why can't it be 25? or 100?
For example, the minimum value of a prime numbers is 2. That doesn't mean that the only prime number is 2. However, if you can show, that the minimum value of an even prime number is 2, and the maximum value of an even prime number is 2, then you can conclude that the only possible value of an even prime number is 2 (and then check if it is an even prime).
Since,a^2+3
b is a perfect square
then, a^2+3
b>=0
a^2>=-3b ........(1)
and b^2+3
a>=0
b^2>=-3
a
a>=b^2/-3 ..........(2)
Substituting (2) in (1), we get (-b^2/3)^2>=-3*b b^3>= -27 b>= -3 .......,(3)
Substituting (3) in (2), we get a>=-3
when putting b=3 in (1), we get a^2>=-9, which is impossible
therefore, -3<=b<3 similarly, -3<=a<3
Hence, (1,1) is the only set which satisfy the condition that both a^2+3 b and b^2+3 a are perfect squares.
Comments and replies:
Calvin:
I disagree with you substituting (2) into (1). Inequalities do not work this way. As a clear example, if (1) was 3 > 1 and (2) was 3 > 0 , then when you are 'substituting (2) into (1)', you claim to get 0>1 . If we then 'substitute (1) into (2)', you claim to get 1 > 0 . Clearly one of these equations must be wrong.
Please review rules of inequalities. What you need is (1) 3 > 1 and (2) 5>3 in order to 'substitute (2) into (1)' to get 5 > 3 > 1 .
(1,1) is a solution which satisfies this equation. To satisfy the equation,both 'a' and 'b' should be odd numbers as difference between two square numbers which is a multiple of 5 should be odd. By using this fact a second pair can be obtained whose inverse is the third pair.
Let us call a=x and b=y
Since x and y are positive, we may write
x^2 + 3y = (x + a)^2, and y^2 + 3x = (y + b)^2
where a, b are positive integers.
Expanding, we find that the squared terms cancel, leaving the linear simultaneous equations (Simultaneous equations are a set of equations which have more than one value which has to be found).
3y = 2ax + a^2
3x = 2by + b^2
Solving, we obtain
x = (2a^2b + 3b^2)/(9 − 4ab)
y = (2b^2a + 3a^2)/(9 − 4ab)
Since a and b are positive, the numerator in each fraction will be positive. For the denominator to be positive, we must have ab = 1 or 2.
If (a,b) = (1,1), (1,2), (2,1), then, respectively, (x,y) = (1,1), (16,11), (11,16). Hence these are the 3 only solutions.
Now as a and b are positive integers, (a+1)^2 > or = a^2+3b, which implies 2a+1 > or = 3b. Also, (b+1)^2 > or = b^2+3a will imply 2b+1 > or = 3a. Combining these inequalities, a > or = 1 and b > or = 1. So the only possible positive integer solution is (1,1).
Comments and replies:
Calvin:
I disagree with your statement that ( a + 1 ) 2 ≥ a 2 + 3 b . Why must this be true?
I'm glad you pointed out that ( 1 , 1 ) is a solution. Many people claimed that there were no solutions.
P.S. To type LaTex on the blog, put your code as c o d e , but remove the space between the first $ and the word latex.
Lee Kee Wei
Since a^2+3b must be a perfect square and that a^2 is a perfect square already and with b>0, then a^2+3b must at least or larger than the perfect square which is next of a^2, which is (a+1)^2.
Calvin:
Indeed, that is true. What you are saying now is that a 2 + 3 b ≥ ( a + 1 ) 2 , whereas you initially said that ( a + 1 ) 2 ≥ a 2 + 3 b .
For example, if a = 1 , b = 5 , then a 2 + 3 b = 1 6 which is a perfect square that is at least as large as ( a + 1 ) 2 = 4 .
Lee Kee Wei:
Opps, i see, that was a bad careless mistake i made.
Now as a and b are positive integers, since a^2+3b must be a perfect square and that a^2 is a perfect square already and with b>0, then a^2+3b must at least or larger than the perfect square which is next of a^2, which is (a+1)^2, so a^2+3b > or =(a+1)^2 , which implies 3b > or = 2a+1. Also, b^2+3a > or = (b+1)^2 will imply 3a > or = 2b+1. Combining these inequalities, a > or = 1 and b > or = 1. So the only possible positive integer solution is (1,1).
Calvin:
You have shown that (beginning with the assumption that a and b are positive integers) then a and b must be at least 1. I can draw this conclusion without any of the conditions in the question.
I do not see how you can conclude that the only possible positive integer solution is ( 1 , 1 ) .
I have been able to show that the g.c.d of a and b is either 1 or 3.
Let a 2 + 3 b = x 2 b 2 + 3 a = y 2
Let (a, b)= d, so that a= ds and b= dr for some coprime integers and s.
So, d 2 ∗ s 2 + 3 d r = x 2 Or, d ( d s + 3 r ) = x 2
So, we can conclude that d divides x^2. However since x^2 is a perfect square, d must be present at least twice [or an even number of times] in the factorization of x^2.
But d is already present once in the factorization of x^2.
For d to be present twice in the factorization, d must divide ds + 3r
Or, d must divide 3r
Now, r= b/d.
So, d must divide 3b/d
Or, d^2 must divide 3b.
Now d^2 cannot divide b, the proof is given below.
On the contrary, assume that d^2 divides b, so that b = p d 2 for some integer p.
It has been shown that d divides x. So, x= dq, for some integer q.
So, a = d 2 ( q − 3 p )
Therefore, d^2 divides a, and the g.c.d of a & b is d^2, a contradiction [it can be possible only when d=1. Then if d not equal to 1, d=3, the proof of which is shown below].
Since d 2 cannot divide b, the gcd of d and b is d.
So, d must divide 3.
Thus d=1 or d=3.
Comments and replies:
Calvin:
I disagree with the statement that "So, we can conclude that d divides x^2. However since x^2 is a perfect square, d must be present at least twice [or an even number of times] in the factorization of x^2." What happens when d = 4 ? Must 16 divide x 2 ?
As such, we may not conclude that "For d to be present twice in the factorization, d must divide ds + 3r".
Even assuming that you are right (lie in cases where d is a prime), your statement that a = d 2 ( q − 3 p ) should actually be a 2 = d 2 ( q − 3 p ) , which then gives that d divides a . For example, a = 5 , b = 2 5 gives 5 2 + 3 × 2 5 = 1 0 2 .
Note, though you defined y , you never used it in the rest of the proof. Hence my example of a = 5 , b = 2 5 is a valid dispute of your statements, even thought 2 5 2 + 3 × 5 is not a perfect square.
Problem Loading...
Note Loading...
Set Loading...
It is clear that a 2 + 3 b > a 2 and b 2 + 3 a > b 2 . Since both of them are perfect squares, then there exist positive integers k and l for which a 2 + 3 b = ( a + k ) 2 and b 2 + 3 a = ( b + l ) 2 .
Case 1. k ≥ 2 and l ≥ 2 . We have a 2 + 3 b ≥ a 2 + 4 a + 4 and b 2 + 3 a ≥ b 2 + 4 b + 4 . Combining the two inequalities yields 0 ≥ a + b + 8 , which is impossible. So, there is no solution.
Case 2. One of k or l equals 1, W.L.O.G. k = 1 . So, a 2 + 3 b = ( a + k ) 2 = a 2 + 2 a + 1 ⇔ 3 b = 2 a + 1 . Substituting this into the second equation, b 2 + 3 a = ( b + l ) 2 ⇔ 3 ( 2 3 b − 1 ) = 2 b l + l 2 ⇔ b ( 9 − 4 l ) = 3 + 2 l 2 . Since the right hand side of the last equality is always positive, then 9 > 4 l which implies that l = 1 or 2 . For the case l = 1 , we have b = 1 and a = 1 as well. For the case l = 2 , we have b = 1 1 and a = 1 6 . Hence, we have 3 solutions: ( a , b ) = ( 1 , 1 ) , ( 1 1 , 1 6 ) , ( 1 6 , 1 1 ) .
So, there are 3 ordered pairs ( a , b ) satisfying the problem.