Double maxed out

Algebra Level 5

For a positive real number r r , let M ( r ) M(r) be the largest possible value of F ( x , y , z ) = x 2 + y z 3 F(x,y,z) = x^2 +yz^3 over all ordered triples of non-negative real numbers ( x , y , z ) , (x,y,z), such that x 2 + y 2 + z 2 = r . x^2+y^2+z^2=\sqrt{r}.

There is a value of r , r, denoted by r r^* , such that F ( x , y , z ) = M ( r ) F(x,y,z)=M(r^*) for two different triples ( x , y , z ) (x,y,z) which satisfy x 2 + y 2 + z 2 = r . x^2+y^2+z^2=\sqrt{r^*}. If r = a b r^* = \frac{a}{b} where a a and b b are coprime positive integers, what is a + b a + b ?


The answer is 283.

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.

4 solutions

Mark Hennings
Nov 10, 2013

Using Lagrange multipliers, we need to solve the equation ( 2 x z 3 3 y z 2 ) λ ( 2 x 2 y 2 z ) = ( 0 0 0 ) \left(\begin{array}{c} 2x \\ z^3 \\ 3yz^2 \end{array}\right) - \lambda\left(\begin{array}{c} 2x \\ 2y \\ 2z \end{array} \right) \; = \; \left(\begin{array}{c}0\\0\\0 \end{array}\right) and so 2 x ( 1 λ ) = z 3 2 λ y = z ( 3 y z 2 λ ) = 0 2x(1-\lambda) \; = \; z^3 - 2\lambda y \; = \; z(3yz - 2\lambda) \; = \; 0

  1. If z = 0 z=0 , then λ y = x ( 1 λ ) = 0 \lambda y = x(1-\lambda) = 0 . If λ = 0 \lambda = 0 then x = 0 x=0 and y = r 1 4 y = r^{\frac14} , and the matching extremal value of F F is 0 0 . If λ 0 \lambda \neq 0 then y = 0 y=0 , so that x = r 1 4 x = r^{\frac14} and λ = 1 \lambda=1 , and the matching extremal value of F F is r 1 2 r^{\frac12} .

  2. If z 0 z \neq 0 then λ = 3 2 y z \lambda = \tfrac32yz and x ( 2 3 y z ) = z ( z 2 3 y 2 ) = 0 x(2 - 3yz) = z(z^2 - 3y^2) = 0 . Thus z 2 = 3 y 2 z^2 = 3y^2 , so z = y 3 z = y\sqrt{3} , so x ( 2 3 3 y 2 ) = 0 x(2 - 3\sqrt{3}y^2) = 0 . If x = 0 x=0 then y = 1 2 r 1 4 y = \tfrac12r^{\frac14} and z = 3 2 r 1 4 z = \tfrac{\sqrt{3}}{2}r^{\frac14} , and the matching extremal value of F F is 3 3 16 r \tfrac{3\sqrt{3}}{16}r . If x 0 x \neq 0 then y = 2 3 3 z = 2 3 y \; = \; \sqrt{\frac{2}{3\sqrt{3}}} \qquad z \; = \; \sqrt{\frac{2}{\sqrt{3}}} and hence x 2 = r 1 2 8 3 3 x^2 \; = \; r^{\frac12} - \frac{8}{3\sqrt{3}} The matching extremal value of F F is r 1 2 4 3 3 r^{\frac12} - \frac{4}{3\sqrt{3}} which is smaller than r 1 2 r^{\frac12} .

Thus r r^\star is identified by the requirement that ( r ) 1 2 = 3 3 16 r (r^\star)^{\frac12} \,=\, \frac{3\sqrt{3}}{16}r^\star , which implies that r = 256 27 r^\star \,=\, \frac{256}{27} . The required answer is 256 + 27 = 283 256+27 = 283 .

Woah--What are lagrange multipliers?

Tanishq Aggarwal - 7 years, 7 months ago

Log in to reply

Is there any simpler way to do the problem?

Tanishq Aggarwal - 7 years, 7 months ago

Log in to reply

Lagrange multipliers are a method for extremizing a function F ( x , y , . . . ) F(x,y,...) of several variables subject to a constraint G ( x , y , . . . ) = 0 G(x,y,...)=0 . You obtain the extrema by solving F λ G = 0 G = 0 \nabla F - \lambda \nabla G \; = \; 0 \qquad G = 0 for some constant λ \lambda . You could do this another way by "parametrizing the constraint". In this case you could write x = r 1 4 cos θ y = r 1 4 sin θ cos ϕ z = r 1 4 sin θ sin ϕ \begin{array}{rcl} x & = & r^{\frac14}\cos\theta \\ y & = & r^{\frac14}\sin\theta\cos\phi \\ z & = & r^{\frac14}\sin\theta\sin\phi \end{array} in which case θ \theta can vary freely between 0 0 and π \pi , and ϕ \phi can vary freely between 0 0 and 2 π 2\pi . Then F F is equal to F = r 1 2 cos 2 θ + r sin 4 θ cos ϕ sin 3 ϕ F \; = \; r^{\frac12}\cos^2\theta + r\sin^4\theta\cos\phi\sin^3\phi and we could successively maximize this with respect to ϕ \phi and θ \theta . There won't be a massive saving of labour, though.

Mark Hennings - 7 years, 7 months ago

There are non-calculus approaches.

HInt: If you fix x x , what is the maximum value of F ( x , y , z ) F(x,y,z) (in terms of x x )?

Calvin Lin Staff - 7 years, 7 months ago

Log in to reply

@Calvin Lin This is just the same as doing the ϕ \phi maximization. The only difference is that you get your answer in terms of x x , not θ \theta .

Mark Hennings - 7 years, 7 months ago

Why is the above requirement?

Nihhaar Chandra Routhu - 7 years, 7 months ago

Log in to reply

We have three positive extrema for general r r , each obtainable in one way. These are r 1 2 r^{\frac12} , 3 3 16 r \tfrac{3\sqrt{3}}{16}r and r 1 2 4 3 3 r^{\frac12} - \tfrac{4}{3\sqrt{3}} . If the global maximum is to be achieved in two different ways, two of these extrema must be the same. Hence the requirement that the first two must be equal for r r^\star , since the third extremum is clearly smaller than the first.

Mark Hennings - 7 years, 7 months ago
Peter Byers
Nov 15, 2013

First let R = y 2 + z 2 = r x 2 R=y^2+z^2=\sqrt r -x^2 , which ranges from 0 0 to r \sqrt r . We'll find the largest value of F ( x , y , z ) = x 2 + y z 3 = r R + y z 3 F(x,y,z)=x^2+yz^3 =\sqrt r -R +yz^3 for a fixed R > 0 R>0 . The key here is to maximize the function g ( θ ) = cos θ sin 3 θ g(\theta) = \cos \theta \sin^3 \theta with domain [ 0 , π / 2 ] [0,\pi/2] . We note that g ( θ ) g'( \theta) is

3 cos 2 θ sin 2 θ sin 4 θ = sin 2 θ ( 3 cos 2 θ sin 2 θ ) = sin 2 θ ( 4 cos 2 θ 1 ) 3 \cos^2 \theta \sin ^2 \theta - \sin^4 \theta =\sin ^2 \theta (3 \cos^2 \theta - \sin^2 \theta) =\sin ^2 \theta (4 \cos^2 \theta - 1) .

Also note that at the endpoints, g ( 0 ) = 0 = g ( π / 2 ) g(0)=0=g(\pi/2) , and the only other critical point is θ = π / 3 \theta=\pi/3 . Hence g g has a unique maximum, g ( π / 3 ) = ( 1 2 ) ( 3 2 ) 3 = 3 3 16 g(\pi/3)= \left ( \frac 1 2 \right ) \left ( \frac {\sqrt{3}} 2 \right ) ^{3} =\frac {3\sqrt{3}} {16}

From the above result it follows that, for a fixed R 0 R\ge0 , the largest value of F ( x , y , z ) = x 2 + y z 3 = r R + y z 3 F(x,y,z)=x^2+yz^3 =\sqrt r -R +yz^3 is:

m ( R ) = r R + 3 R 2 3 16 m(R)=\sqrt r -R + \frac {3R^2\sqrt{3}} {16} .

(Note that this is trivial for R = 0 R=0 .) Since m ( R ) m(R) is an upward-concave parabola (as a function of R R ) it achieves its maximum at one or both of the endpoints R = 0 R=0 , R = r R= \sqrt r . Therefore r = r r=r^ * if and only if m ( 0 ) = m ( r ) m(0)=m(\sqrt {r}) . Squaring both sides, this equation becomes r = ( 3 r 3 16 ) 2 r= \left (\frac {3r\sqrt{3}} {16}\right) ^2 and since r 0 r\ne 0 this means that r = 256 27 r^*= \frac {256} {27} .

How can we substitute y and z with the trigo functions? What it this methid called?

Led Tasso - 7 years, 6 months ago

Log in to reply

@Abhimanyu Swami,

I didn't say so explicitly, but yes that is what's happening in the first part: substituting R cos θ \sqrt R \cos \theta for y y and R sin θ \sqrt R \sin \theta for z z .

Peter Byers - 7 years, 6 months ago
Rajeh Alghamdi
Mar 4, 2014

283

283

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...