x , y , and z are real numbers such that x + y + z = 0 and x 2 + y 2 + z 2 = 1 . The largest possible value of x 2 y 2 z 2 can be expressed as b a , where a and b are positive coprime integers. What is the value of a + b ?
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.
A very clever solution!
Excellent method. I was only able to think of AM-GM for this.
I solved the same way as you. I think it could be slightly clearer if you had used Newton's Formula to get the equation (the one you called P ( t ) ).
x + y + z = 0 ⇒ z 2 = ( x + y ) 2
⇒ x 2 + y 2 + z 2 = 2 x 2 + 2 y 2 + 2 x y = 1
⇒ x 2 + y 2 + x y = 2 1 and ( x + y ) 2 = z 2 = 2 1 + x y
Using x 2 + y 2 ≥ 2 x y in above equation
⇒ x y ≤ 6 1
Now , x 2 y 2 z 2 = x 2 y 2 ( x + y ) 2 = ( x 2 y 2 ) ( 2 1 + x y )
Since , x y ≤ 6 1
⇒ x 2 y 2 z 2 ≤ 5 4 1 . ⇒ a + b = 5 5
Nice solution! However, Jimmy K. is right: you need to explicitly state that the inequality can be achieved. This is because "by default" the inequality may be true but not optimal. And even when you start with optimal inequalities, you may end up with a non-optimal one. For example, suppose a and b are positive real numbers with a b = 1 . Then a + b ≥ 2 and a + 4 b ≥ 4 (prove it!) Adding up, we get 2 a + 5 b ≥ 6 . This is true, but it does not imply that 6 is achievable. In fact, the optimal bound is 2 1 0 .
For completeness, you need to show that x 2 y 2 z 2 = 5 4 1 is achievable. Otherwise, the maximum value of x 2 y 2 z 2 could be smaller.
Log in to reply
I have already proved x y ≤ 6 1 , we just have to put that value.
ok i got it ,. thanks!!
Please also note that x, y should be of same sign. As x, y, z can be negative, AM-GM can't be put just like that.
Log in to reply
Good point! See Russel F.'s solution and master note on how to fix this.
where did i use AM- GM on x , y and z
We claim that the maximum is 5 4 1 . This could be achieved when x = y = 6 6 , z = 3 − 6 , any permutation or the result if you multiply each of x , y , z by − 1 .
We would disregard the case where any of x , y , z is 0 , because in that case, x 2 y 2 z 2 will be 0 , but we have shown that 5 4 1 , which is greater than 0 , is achievable.
By the pigeonhole principle, at least two of x , y , z are either both negative or nonnegative. Since we are disregarding the case where any of them is 0 , without loss of generality, we assume that x , y are positive. Then z = − x − y . Note that we are now maximizing x 2 y 2 z 2 = x 2 y 2 ( − x − y ) 2 .
Plugging z = − x − y in x 2 + y 2 + z 2 = 1 gives x 2 + y 2 + ( − x − y ) 2 = 1 , so 2 x 2 + 2 y 2 + 2 x y = 1 , so x 2 + y 2 + x y = 2 1 . By the AM-GM inequality, x 2 + y 2 ≥ 2 x y . Hence x y ≤ ( 2 1 ) ( 3 1 ) = 6 1 . Thus ( x + y ) 2 = x 2 + 2 x y + y 2 ≤ 2 1 + 6 1 = 3 2 , so x + y ≤ 3 2 = 3 6 .
Hence x y ( x + y ) ≤ ( 6 1 ) ( 3 6 ) = 1 8 6 . Thus x 2 y 2 z 2 = x 2 y 2 ( x + y ) 2 = ( x 2 y 2 ( − x − y ) ) 2 ≤ ( 1 8 6 ) 2 = 3 2 4 6 = 5 4 1 .
Since the largest possible value of x 2 y 2 z 2 is 5 4 1 , a = 1 and b = 5 4 , so a + b = 1 + 5 4 = 5 5 .
Nice solution! You should point out that there is an additional symmetry, so if both x and y are negative, one can flip all signs to make them positive.
Please scroll to the right to make sure you could read everything.
Log in to reply
On my screen, your solution doesn't have a scroll bar, but it is entirely within the margins.
This is a very nice solution. Shouldn't you point out when both x and y are negative?
Log in to reply
I wrote:By the pigeonhole principle... Please read that part again.
@ Challenge Master: Yes, I should have pointed out that it is okay to disregard the x and y are negative case and just assume that at least two are positive and state the cyclic symmetry and so on...
Notice z = − ( x + y ) . Via substitution, x and y must satisfy x 2 + y 2 + ( x + y ) 2 = 1 . By symmetry, the maximum of x 2 y 2 z 2 = x 2 y 2 ( x + y ) 2 occurs when x = y . Thus x = y = 6 1 , hence z = 6 2 . The value of x 2 y 2 z 2 is therefore 2 1 6 4 = 5 4 1
Not all symmetric functions achieve maximum when the variables are the same. As a simple example, for x + y = 1 the function x 2 + y 2 has minimum when x = y . More complicated behavior can also occur for more complicated symmetric function. For example, you may have a local maximum, but not a global maximum there.
Just realized: the value I give for z should be negated. My bad!
For another "simple" example, try minimizing/maximizing x 3 + y 3 constrained to x 2 + y 2 = 1 .
Log in to reply
Yeah, I had a feeling I got lucky.
However, constrained optimization problems aside, I can't help but feel there is a more elegant line of reasoning as to why the function f ( x ) = x 2 y 2 ( x + y ) 2 attains a maximum when x = y than the solutions being put forth (I hope that doesn't offend anyone). When I reached that equation, my brain immediately said "set them equal to each other!" However, I won't rule out the possibility that it was sheer dumb luck.
Log in to reply
This has to do with some kind of concavity, which cannot be immediately seen. Let's say it was not a dumb luck, but an educated guess :)
This is quite inelegant and involves calculus, but I could not seem to make the geometry angle work for me.
Treat x , y , z as the roots of a cubic equation with leading coefficient 1. The given information implies that the sum of the roots is 1.
We also have ( x + y + z ) 2 = x 2 + y 2 + z 2 + 2 ( x y + y z + z x ) . So we must have x y + y z + z x = − 1 / 2 We also write c = x y z
Putting these bits of information together, we know the equation must be of the form u 3 − 2 1 u − c = 0
Rearranging the equation we have u 3 = 2 1 u + c . So we are looking for intersections between a line of slope 2 1 and the graph of u 3 . The question says that x,y,z are distinct real numbers, so we need the cubic to have 3 real roots (not necessarily distinct).
Clearly changing c only changes the intercept of the line, and we want the intercept to be as large as possible. Sketching the graphs, it is clear to see that this is the case when the line with slope 2 1 is tangent to the cubic.
Differentiating and setting the slope to 2 1 we get 3 u 2 = 2 1 and u = ± 6 1 .
At this value of u both the line and the curve have the same y coordinate, so − 6 6 1 = 2 1 × 6 6 1 + c
∴ c = 3 6 1 and c 2 = 5 4 1
This is actually a pretty nice solution, very similar to the one by Jimmy K.
If it doesn't matter to involve calculus, the Lagrange multiplier method is another good approach.
We already know x + y + z = 0 and x y + y z + z x = − 1 / 2 . Now let u = x y z and consider the cubic equation
t 3 − 2 1 t − u = 0
Recall that for the cubic equation t 3 + p t + q = 0 we have the discriminant
Δ = 4 q 2 + 2 7 p 3
, where the equation has real roots iff Δ ≤ 0 . So for our equation, we have
Δ = − 2 1 6 1 + 4 u 2 ≤ 0 ⟹ x 2 y 2 z 2 = u 2 ≤ 5 4 1
Take x as rcos(t) and y as rsin(t) where r=sqrt(1-z^2) . This is got from the second equation . Now substitute this in the first equation where we get sin(t) + cos(t) as a function of z. Now square the equation and get sin(t)xcos(t) . Now x and y as a function of z only . Now substitute into the third equation which becomes a function of z only . Now differentiate once to get the corresponding value of z ( z^2 is 1/6) and substitute this into the third equation to get maximum value .
Since x+y+z=0 either x=y=z=0, 2 of the variables are positive and 1 negative, or vice versa. The former case doesn't hold for the second condition, and the second and third cases are synonymous. WLOG let x+y=z.
Now plugging that into the second condition and simplifying we obtain x^2+y^2+xy=1/2. We want to maximize x^2y^2(x+y)^2.
Now, consider the quadratic a^2+ba+c with real roots x and y. The sum of the roots (x+y) is -b, and the product (xy) is c. Therefore we can rewrite as having b^2-c=1/2 and wanting to maximize b^2c^2.
From AM-GM, note that 3 b 2 − c / 2 − c / 2 ≥ 3 4 b 2 c 2 . This boils down to ( 3 1 / 2 ) 3 ≥ b 2 c 2 / 4 so the maximum value of b^2c^2 is 1/54 and the answer is 55.
Once you have got a bound, you must show that it is achievable. Only then can answer be taken as complete.
From the given conditions, we have z = − ( x + y ) so that x 2 + y 2 + ( x + y ) 2 = 1 . Expanding and collecting terms, we have x 2 + y 2 + x y = 2 1 . By AM-GM inequality, we obtain x 2 + y 2 ≥ 2 x y with equality if and only if x = y . From the previous equation, we then have 2 x y + x y ≤ 2 1 or x y ≤ 6 1 . Since x 2 + y 2 = ( x + y ) 2 − 2 x y , from the previous equation, we also have ( x + y ) 2 − x y = 2 1 or ( x + y ) 2 = 2 1 + x y . This implies 2 1 + x y ≥ 0 or x y ≥ − 2 1 and x 2 y 2 z 2 = ( x y ) 2 ( x + y ) 2 = ( x y ) 2 ( 2 1 + x y ) . Let t = x y . We want to maximize the function f ( t ) = t 2 ( 2 1 + t ) , − 2 1 ≤ t ≤ 6 1 . With a littile help from calculus, we find f ′ ( t ) = t ( 3 t + 1 ) so that relative extreme values occur at t = 0 or t = − 3 1 . Since f is continuous on [ − 2 1 , 6 1 ] , the absolute extreme values have to occur at one of the following points: − 2 1 , − 3 1 , 0 , 6 1 . We compare the values of f ( t ) at these points and find that f ( − 3 1 ) = f ( 6 1 ) = 5 4 1 is the maximum value. Hence the maximum of x 2 y 2 z 2 is 5 4 1 , attained when, for instance, x = y = 6 1 and z = − 6 2 .
I like your solution, Aram T. I did something similar, but with one difference which I will describe for the sake of variety.
Instead of determining the range of possible values of x y , I assumed WLOG that x y ≤ 0 , which is a permissible assumption because of symmetry and the fact that x y , x z , y z cannot all be positive.
Therefore, the set of values of x 2 y 2 z 2 = f ( x y ) is a subset of the values of f ( t ) over t ≤ 0 . But f ( t ) is increasing on t < − 1 / 3 and decreasing on − 1 / 3 < t < 0 . And, as you pointed out, x 2 y 2 z 2 = f ( − 1 / 3 ) = 1 / 5 4 is attained when e.g. x = z = 1 / 6 and y = − 2 / 6 , and is therefore the maximum.
To simplify the problem, we will eliminate x by writing x as ( − y − z ) . The problem now becomes: given that ( − y − z ) 2 + y 2 + z 2 = 1 , find the maximum value of ( − y − z ) 2 y 2 z 2
The constraint is equivalent to 2 y 2 + 2 z 2 + 2 y z = 1 , or y 2 + z 2 + y z = 2 1 .
We have to find the maximum value of ( − y − z ) 2 y 2 z 2 = ( y 2 + 2 y z + 2 z 2 ) ( y z ) 2 . From the constraint we have y 2 + z 2 + y z = 2 1 , so y 2 + z 2 + 2 y z = 2 1 + y z . Therefore we need to maximize ( 2 1 + y z ) ( y z ) 2 .
Now, since y 2 + z 2 + y z = 2 1 , and we know that y 2 + z 2 ≥ 2 y z , 2 1 ≥ 2 y z + y z = 3 y z . Thus, y z ≤ 6 1
Hence the maximum value we need to find is ( 2 1 + y z ) ( y z ) 2 ≤ ( 2 1 + 6 1 ) ( 6 1 ) 2 = 5 4 1 . The equality sign occurs when y = z = 6 1 , x = 6 − 2 .
Therefore, a + b = 1 + 5 4 = 5 5
there's a typo in the 3rd paragraph, equation should say ( − 6 − z ) 2 y 2 z 2 = ( y 2 + 2 y z + z 2 ) ( y z ) 2
We have: 2 y z = ( y + z ) 2 − ( y 2 + z 2 ) So 2 y z = 2 x 2 − 1 Rewrite x 2 y 2 z 2 = x 6 − x 4 + 4 1 x 2 . Applying : ( y + z ) 2 ≥ 4 y z We get: 0 ≤ x 2 ≤ 3 2 . Let x 2 = t ;We see f ( t ) = t 3 − t 2 + 4 1 t in the interval 0 ≤ t ≤ 3 2 . f ( 0 ) = f ( 2 1 ) = 0 ; f ( 6 1 ) = f ( 3 2 ) = 5 4 1 So max of f ( t ) = 5 4 1 = n m So m + n = 5 5 . Not different .
The condition x 2 + y 2 + z 2 = 1 reminded me the sphere in R 3 , so I used spherical coordinates, i.e. x = sin θ sin ϕ , y = sin θ cos ϕ , z = cos θ . The condition x + y + z = 0 then implies tan θ ( cos ϕ + sin ϕ ) = − 1 , where I already ruled out the possibility of cos θ = 0 , i.e. z = 0 , since then x 2 y 2 z 2 = 0 . After some manipulation we can write x 2 y 2 z 2 = 4 [ 2 + sin ( 2 ϕ ) ] 3 1 + sin ( 2 ϕ ) sin 2 ( 2 ϕ ) . We now consider − 1 ≤ sin ( 2 ϕ ) ≤ 1 as our new variable. The above function of sin ( 2 ϕ ) has positive derivative for − 1 ≤ sin ( 2 ϕ ) ≤ 1 , therefore the maximum value is attained for sin ( 2 ϕ ) = 1 . Thus, we have: x 2 y 2 z 2 = 5 4 1 = b a , and finally a + b = 5 5 .
Great! The geometric visualization helps us see that we have a plane intersecting a sphere, which gives us a circle. It makes sense that we could have a one-variable parametrization, and you found a possible approach.
Log in to reply
@Calvin Lin Sir, won't the intersection be an ellipse instead of a circle????
Log in to reply
The intersection of any sphere with any plane is always a circle.
To visualize that, rotate the plane till it's perpendicular to the z-axis.
Note: All circles are ellipses.
Problem Loading...
Note Loading...
Set Loading...
Let x , y , z be the real roots of a monic, cubic polynomial.
Then, P ( t ) = ( t − x ) ( t − y ) ( t − z ) = t 3 + A t 2 + B t + C .
By expanding the left side, and equating coefficients, we get:
A = − ( x + y + z ) = 0 ,
B = x y + y z + z x = 2 1 ( ( x + y + z ) 2 − ( x 2 + y 2 + z 2 ) ) = − 2 1 ,
C = − x y z .
We want the maximum value of x 2 y 2 z 2 = C 2 such that the cubic equation P ( t ) = t 3 − 2 1 t + C = 0 has three real roots.
Since P ′ ( t ) = 3 t 2 − 2 1 = 0 ⇝ t = ± 6 1 , and P ( t ) is a cubic with positive leading coefficient, we know that P ( t ) has a local minimum/maximum at ( ± 6 1 , P ( ± 6 1 ) ) = ( ± 6 1 , ∓ 3 6 1 + C ) .
Note that a cubic polynomial has three real roots iff the local maximum value is non-negative and the local minimum value is non-positive. Therefore, we need 3 6 1 + C ≥ 0 and − 3 6 1 + C ≤ 0 .
So, − 3 6 1 ≤ C ≤ 3 6 1 , and thus, x 2 y 2 z 2 = C 2 ≤ 5 4 1 . Equality is achievable for ( x , y , z ) = ( ∓ 6 2 , ± 6 1 , ± 6 1 ) and permutations.
Thus, the maximum value of C 2 = x 2 y 2 z 2 is 5 4 1 = b a ⇝ a + b = 5 5 .