One more of those

Algebra Level 4

1330 x y + 3192 y z 1330xy+3192yz

Find the maximal value of the expression above if x 2 + y 2 + z 2 = 1 x^2+y^2+z^2=1 , where x , y x,y and z z are real numbers.

Bonus question : Find the maximal value of a x y + b y z axy+byz if x 2 + y 2 + z 2 = 1 x^2+y^2+z^2=1 .


The answer is 1729.

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

Let x = cos ( θ ) sin ( ϕ ) , y = sin ( θ ) x = \cos(\theta)\sin(\phi), y = \sin(\theta) and z = cos ( θ ) cos ( ϕ ) . z = \cos(\theta)\cos(\phi). Then x 2 + y 2 + z 2 = 1 x^{2} + y^{2} + z^{2} = 1 and

1330 x y + 3192 y z = sin ( θ ) cos ( θ ) ( 1330 sin ( ϕ ) + 3192 cos ( ϕ ) ) = 1330xy + 3192yz = \sin(\theta)\cos(\theta)(1330\sin(\phi) + 3192\cos(\phi)) =

sin ( 2 θ ) 2 133 0 2 + 319 2 2 ( 1330 133 0 2 + 319 2 2 sin ( ϕ ) + 3192 133 0 2 + 319 2 2 cos ( ϕ ) ) = \dfrac{\sin(2\theta)}{2}*\sqrt{1330^{2} + 3192^{2}}*\left(\dfrac{1330}{\sqrt{1330^{2} + 3192^{2}}}\sin(\phi) + \dfrac{3192}{\sqrt{1330^{2} + 3192^{2}}}\cos(\phi)\right) =

sin ( 2 θ ) 2 3458 sin ( α + ϕ ) = 1729 sin ( 2 θ ) sin ( α + ϕ ) , \dfrac{\sin(2\theta)}{2}*3458*\sin(\alpha + \phi) = 1729*\sin(2\theta)\sin(\alpha + \phi), where α = cos 1 ( 1330 3458 ) . \alpha = \cos^{-1}\left(\dfrac{1330}{3458}\right).

This expression has a maximum value of 1729 \boxed{1729} when θ = 4 5 \theta = 45^{\circ} and ϕ = 9 0 α . \phi = 90^{\circ} - \alpha.

As noted by Harsh, the maximum in general for a x y + b y z axy + byz given that x 2 + y 2 + z 2 = 1 x^{2} + y^{2} + z^{2} = 1 is

1 2 a 2 + b 2 . \dfrac{1}{2}\sqrt{a^{2} + b^{2}}.

Great use of spherical coordinates, Brian! Thanks! (+1)

Otto Bretscher - 5 years, 7 months ago

Log in to reply

sir i have posted a solution for your problem using Lagrange,s method of multipliers , xtrmly sorry for not writing in Latex, a kind request to you is to check my solution , i got correct answer ,but i do not know whether my approach is correct or not.

manish kumar singh - 5 years, 7 months ago

@Otto Bretscher Any thoughts on a general formula for the maximum of a x y + b y z + c x z axy + byz + cxz given x 2 + y 2 + z 2 = 1 x^{2} + y^{2} + z^{2} = 1 ? It doesn't appear to be 1 2 a 2 + b 2 + c 2 . \frac{1}{2}\sqrt{a^{2} + b^{2} + c^{2}}.

Brian Charlesworth - 5 years, 7 months ago

Log in to reply

Maybe I will ask for that one next ;) ("Inspired by Brian", of course)

Otto Bretscher - 5 years, 7 months ago

Log in to reply

Haha O.k., in anticipation, I'll now start trying to figure that one out. :)

Brian Charlesworth - 5 years, 7 months ago

Log in to reply

@Brian Charlesworth I have submitted a problem like that before.... now let's generalize.

Otto Bretscher - 5 years, 7 months ago

Log in to reply

@Otto Bretscher sir actually if we substitute x=y=z=(1/3) surely the value of the expression will give a value of 1507.333

PSN murthy - 5 years, 7 months ago

Log in to reply

@Psn Murthy Actually, if we substitute x = y = z = 1 3 x=y=z=\frac{1}{3} , we are not satisfying the constraint.

Otto Bretscher - 5 years, 7 months ago

The maximum is given by a cubic equation in this case, so, writing the general formula in terms of a , b a,b and c c would be messy. I'm having some fun with this problem here

Otto Bretscher - 5 years, 7 months ago

Log in to reply

Messy indeed. I'll try and write a solution to your new question in the morning, (unless someone else beats me to it.)

Brian Charlesworth - 5 years, 7 months ago

Log in to reply

@Brian Charlesworth I'm sorry for the messiness... you asked for it, Brian! ;)

Otto Bretscher - 5 years, 7 months ago

Another possible generalization is a x y + b y z + c z w axy+byz+czw for a fourth variable w w ... you are giving me so many good ideas. I'm playing with this one here

Otto Bretscher - 5 years, 7 months ago

sir , i have solved it using Lagrange's method of undetermined multipliers. i am writing solution let me know whether i am correct or not. define a function F = axy+byz+L(x^2+ y^2+z^2 - 1), for some constant L. Differentiating F partially w .r t. x, y, z and putting them equal zero, 1st differentiating partially w.r.t x we have ; ay+2Lx=0 ----eqn(1). differentiating partially w.r.t y ; ax+bz +2Ly=0-------eqn(2). differentiating partially w.r.t z; by+2Lz=0-------eqn(3).

from eqn 1 ,x=-(ay/2L). from eqn 3, z=-(by/2L).

PUT THESE values of x and z in eqn 2, we have ; (y/2L)[{4L^2} - (a^2 + b^2)]=0:

here y can not be zero, because it leads to x=0, y=0, that is the origin of the sphere.

hence 2L= sqrt(a^2+b^2);

also ,we have x^2+y^2+z^2=1; putting the values of x and z in terms of y in this equation we have;

(y^2)[(a^2/4L^2) + (b^2/4L^2) +1]=1;

(y^2)[(a^2+b^2)/(4L^2 ) + 1]=1;

Putting 4L^2 = [a^2+b^2];

we have y = +1/(sqrt 2) or y = -1/(sqrt 2); we can take any one of these because x,y,z are lying on the surface of unit sphere ,so for every point there must be a symmetric point.

putting value of y and L, WE HAVE ;

X= (1/sqrt 2)[a/sqrt(a^2+b^2)] , Z = (1/sqrt 2)[ b/sqrt( a^2+b^2)] ;

so maximum value of axy+ byz= (1/2) sqrt(a^2+b^2); put a=1330. b= 3192. we get ans= 1729.

manish kumar singh - 5 years, 7 months ago

Log in to reply

Yes, this is correct and nicely done! The equations you find in the first paragraph are linear in x , y x,y and z z ; that's why this is relatively easy to solve.

The coefficient matrix of the system is A = [ 2 L a 0 a 2 L b 0 b 2 L ] A=\begin{bmatrix} 2L&a&0\\a&2L&b\\0&b&2L\end{bmatrix}

To produce nonzero solutions, this matrix must be singular (non-invertible), which means that det ( A ) = 8 L 3 2 ( a 2 + b 2 ) L = 0 \det(A)=8L^3-2(a^2+b^2)L=0

The largest solution, L = a 2 + b 2 2 L=\frac{\sqrt{a^2+b^2}}{2} is the maximum we seek.

Otto Bretscher - 5 years, 7 months ago

Log in to reply

sir i have a question, please help. In an arithmetic progression Nth term is given by An=Ao+Nd; where Ao is 1st term and 'd' is common difference. Ao and 'd' are integers. Then show that if any one term of the progression is the cube of an integer,then so are infinitely many others.

manish kumar singh - 5 years, 4 months ago

Log in to reply

@Manish Kumar Singh I'll assume that the progression is (strictly) ascending. Denote the cubed term a 3 a^{3} . Then the integers

a 3 , a 3 + d , a 3 + 2 d , . . . . , a 3 + k d a^{3}, a^{3} + d, a^{3} + 2d, .... , a^{3} + kd

are all terms of the AP for all k N k \in \mathbb{N} . Then

( a + d ) 3 = a 3 + 3 a 2 d + 3 a d 2 + d 3 = a 3 + ( 3 a 2 + 3 a d + d 2 ) d (a + d)^{3} = a^{3} + 3a^{2}d + 3ad^{2} + d^{3} = a^{3} + (3a^{2} + 3ad + d^{2})d

is in the progression, where k = 3 a 2 + 3 a d + d 2 k = 3a^{2} + 3ad + d^{2} .

Next, we observe that

( a + 2 d ) 3 = a 3 + 6 a 2 d + 12 a d 2 + 8 d 3 = a 3 + ( 6 a 2 + 12 a d + 8 d 2 ) d (a + 2d)^{3} = a^{3} + 6a^{2}d + 12ad^{2} + 8d^{3} = a^{3} + (6a^{2} + 12ad + 8d^{2})d

is also a term in the progression, with k = 6 a 2 + 12 a d + 8 d 2 k = 6a^{2} + 12ad + 8d^{2} .

Similarly, we can show that ( a + n d ) 3 (a + nd)^{3} for any n N n \in \mathbb{N} can be written in the form a 3 + m d a^{3} + md for some positive integer m m , and thus all such perfect cubes are terms in the progression, i.e., given that one term in an AP (as specified) is a perfect cube, an infinite number of terms in the AP will be perfect cubes.

Brian Charlesworth - 5 years, 4 months ago

sir i have a question, please help. In an arithmetic progression Nth term is given by An=Ao+Nd; where Ao is 1st term and 'd' is common difference. Ao and 'd' are integers. Then show that if any one term of the progression is the cube of an integer,then so are infinitely many others.

manish kumar singh - 5 years, 4 months ago
Department 8
Nov 18, 2015

Probably the Harsh Shrivastava and Brian Charlesworth posted awesome solution but I will stick to the basic AM-GM.

We have

( x 2 + 1330 2 y 2 1330 2 + 3192 2 ) + ( 3192 2 y 2 1330 2 + 3192 2 + z 2 ) 2 x 2 × 1330 2 y 2 1330 2 + 3192 2 + 2 3192 2 y 2 1330 2 + 3192 2 × z 2 1 2 1330 2 + 3192 2 ( 1330 x y + 3192 y z ) 1729 ( 1330 x y + 3192 y z ) \Large{\left( { x }^{ 2 }+\frac { { 1330 }^{ 2 }{ y }^{ 2 } }{ { 1330 }^{ 2 }+{ 3192 }^{ 2 } } \right) +\left( \frac { { 3192 }^{ 2 }{ y }^{ 2 } }{ { 1330 }^{ 2 }+{ 3192 }^{ 2 } } +{ z }^{ 2 } \right) \\ \ge 2\sqrt { { x }^{ 2 }\times \frac { { 1330 }^{ 2 }{ y }^{ 2 } }{ { 1330 }^{ 2 }+{ 3192 }^{ 2 } } } +2\sqrt { \frac { { 3192 }^{ 2 }{ y }^{ 2 } }{ { 1330 }^{ 2 }+{ 3192 }^{ 2 } } \times { z }^{ 2 } } \\ \\ 1\ge \frac { 2 }{ \sqrt { { 1330 }^{ 2 }+{ 3192 }^{ 2 } } } \left( 1330xy+3192yz \right) \\ \\ 1729\ge \left( 1330xy+3192yz \right) }

Now for the general expression we have

( x 2 + a 2 y 2 a 2 + b 2 ) + ( b 2 y 2 a 2 + b 2 + z 2 ) 2 x 2 × a 2 y 2 a 2 + b 2 + 2 b 2 y 2 a 2 + b 2 × z 2 a 2 + b 2 2 a x y + b y z \Large{\left( { x }^{ 2 }+\frac { { a }^{ 2 }{ y }^{ 2 } }{ { { a }^{ 2 }+{ b }^{ 2 } } } \right) +\left( \frac { { b }^{ 2 }{ y }^{ 2 } }{ { { a }^{ 2 }+{ b }^{ 2 } } } +{ z }^{ 2 } \right) \\ \ge 2\sqrt { { x }^{ 2 }\times \frac { { a }^{ 2 }{ y }^{ 2 } }{ { { a }^{ 2 }+{ b }^{ 2 } } } } +2\sqrt { \frac { { b }^{ 2 }{ y }^{ 2 } }{ { { a }^{ 2 }+{ b }^{ 2 } } } \times { z }^{ 2 } } \\ \\ \frac { \sqrt { { a }^{ 2 }+{ b }^{ 2 } } }{ 2 } \ge axy+byz}

Thanks for an alternative solution! Can you show us that the maximum is actually attained?

You guys really love your "AM-GM." It seems so complicated to me compared to other methods, particularly matrices and eigenvalues.

Otto Bretscher - 5 years, 6 months ago

Log in to reply

Can you show me how to show when the equality is attained.

Department 8 - 5 years, 6 months ago

Same way!!

Dev Sharma - 5 years, 6 months ago

As you know, equality holds in AM-GM iff the two terms are equal, so , you want x 2 = 133 0 2 y 2 133 0 2 + 319 2 2 x^2=\frac{1330^2y^2}{1330^2+3192^2} (Equ I) and, analogously, z 2 = . . . z^2=... , Equ II. If you add up equations I and II and remember that x 2 + y 2 + z 2 = 1 x^2+y^2+z^2=1 , you find that y 2 = 1 y^2=1 . Now equations I and II give you x x and z z .

Otto Bretscher - 5 years, 6 months ago

Nice solution. Did the same way

Jun Arro Estrella - 5 years, 5 months ago
Harsh Shrivastava
Nov 14, 2015

Answer to BONUS QUESTION : 1 2 a 2 + b 2 \dfrac{1}{2} \sqrt{ a^{2} + b^{2}}

First of all let K K = Maximum value of a x y + b y z axy+byz subject to constraints x 2 + y 2 + z 2 = 1 x^2+y^2+z^2=1

K = y ( a x + b z ) K = y(ax+bz)

By Cauchy Schwarz inequality, a x + b z ( a 2 + b 2 ) ( x 2 + z 2 ) = ( a 2 + b 2 ) ( 1 y 2 ) ax+bz \leq \sqrt{(a^{2} + b^{2})(x^{2}+ z^{2})} = \sqrt{(a^{2}+b^{2})(1 - y^{2})}

K = y ( ( a 2 + b 2 ) ( 1 y 2 ) ) \implies K = y(\sqrt{(a^{2}+b^{2})(1 - y^{2})})

K = ( ( a 2 + b 2 ) ( ( y 4 y 2 + ( 0.5 ) 2 ) + ( 0.5 ) 2 ) \implies K =(\sqrt{(a^{2}+b^{2})(-(y^{4} - y^{2} + (0.5)^{2}) + (0.5)^{2}})

K = ( a 2 + b 2 ) ( ( 1 2 ) 2 ( y 2 1 2 ) 2 ) \implies K = \sqrt{(a^{2}+b^{2})((\frac{1}{2}) ^{2}-(y^{2} - \frac{1}{2})^{2} })

For maximum, y should be equal to 1 2 \sqrt{\frac{1}{2}} .

K = ( a 2 + b 2 ) ( 1 2 ) 2 ( 0 ) 2 ) \implies K =\sqrt{(a^{2}+b^{2})(\frac{1}{2})^{2} - (0)^{2} })

K = 1 2 a 2 + b 2 \implies K = \dfrac{1}{2} \sqrt{ a^{2} + b^{2}}

Exactly! We are looking forward to your proof.

Otto Bretscher - 5 years, 7 months ago

Log in to reply

Posted! @Otto Bretscher sir

Harsh Shrivastava - 5 years, 7 months ago

Log in to reply

Can you show us that your upper bound 1 2 a 2 + b 2 \frac{1}{2}\sqrt{a^2+b^2} is actually attained?

Otto Bretscher - 5 years, 7 months ago

For maximal value of axy+byz we can simplify it to y*(ax+bz).

Now, if sum of squares of x,y,z is 1 then (x^2 + z^2 = 1- y^2).....................(i)

Considering only (ax+bz), we can apply Cauchy-Schwarz Inequality.

Which is (ax+bz)^2 <= (a^2 + b^2)*(x^2 + z^2)

after taking the square root on both sides we obtain,

(ax+bz) <= sqrt [ (a^2 + b^2)*(x^2 + z^2) ]

Then by multiplying y on both sides we have.

y (ax+bz) <= y * {sqrt [ (a^2 + b^2) (x^2 + z^2) ]}

after taking y on RHS inside the square root and using (i) we obtain,

y (ax+bz) <= sqrt[ (a^2 + b^2) (y^2 - y^4)]

Now by applying calculus we can find the maximum value of the equation (y^2 - y^4) , which turns out to be 1/4.

Hence, we have max(axy+byz) <= (1/2) * sqrt(a^2 + b^2).

So for this question we have a=1330 , b=3192,

hence after evaluating we have the answer to be 1729.

This is my approach, please guide me if I am wrong somewhere. Also please know that I am not a good LATEX user.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...