Maximize an Almost-Factorable Expression

Algebra Level 5

Real numbers a , b , c , d a,b,c,d satisfy

a 2 + b 2 = c 2 + d 2 . a^2+b^2=c^2+d^2.

If the maximum of the expression

20 b c + 14 b d + 20 a d 14 a c a 2 + b 2 + c 2 + d 2 \dfrac{20bc+14bd+20ad-14ac}{a^2+b^2+c^2+d^2}

can be expressed as n \sqrt{n} for some positive integer n n , what is the value of n ? n?


The answer is 149.

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

Calvin Lin Staff
Apr 10, 2014

What does ( b c + a d ) (bc+ ad) and ( b d a c ) (bd - ac) remind you of? It has to be Fermat's Two Square Identity, which states that

( a 2 + b 2 ) ( c 2 + d 2 ) = ( b c + a d ) 2 + ( b d a c ) 2 ! (a^2 + b^2) ( c^2 + d^2) = ( bc + ad) ^2 + (bd - ac) ^2 !


For the solution, applying Cauchy Schwarz, we have

[ 20 ( b c + a d ) + 14 ( b d a c ) ] 2 [ ( b c + a d ) 2 + ( b d a c ) 2 ] [ 2 0 2 + 1 4 2 ] = ( a 2 + b 2 ) 2 × 596 \left[ 20 (bc + ad) + 14 (bd - ac) \right]^ 2 \leq \left[ ( bc+ ad)^2 + (bd-ac)^2 \right] \left[ 20^2 + 14^2 \right ] = ( a^2 + b^2) ^2 \times 596

This thus gives us

20 ( b c + a d ) + 14 ( b d a c ) a 2 + b 2 + c 2 + d 2 149 . \left| \frac{ 20 (bc+ad) + 14 (bd - ac) } { a^2 + b^2 + c^2 + d^2 } \right| \leq \sqrt{ 149 }.

It remains to verify that equality in Cauchy Schwarz can exist. This is done by setting a = 0 , b = 596 , c = 20 , d = 14 a = 0, b = \sqrt{596}, c = 20, d = 14 . Hence, the maximum is 149 \sqrt{149} .

@Daniel Liu @Finn Hulse Like this :) ?

You would notice a lot of similarities, especially if you are familiar with the trigonometric proof of Cauchy Schwarz.

Calvin Lin Staff - 7 years, 1 month ago

Log in to reply

Darn, I don't have the expertise in Ineqs to make this type of solution. Thanks for sharing though :)

Daniel Liu - 7 years, 1 month ago

Where can I found the trigonometric proof of Cauchy Schwarz? I searched but not success

Joel Antonio Vásquez - 6 years, 5 months ago

Hello @Calvin Lin , how could you know that * A=0, B= 596 \sqrt{596} , C=20 and D=14 *

Victor Zhang - 5 years, 3 months ago

Log in to reply

Review applying the cauchy schwarz inequality . What are the conditions for equality to occur?

Note further that we also have a 2 + b 2 = c 2 + d 2 a^2 + b^2 = c^2 + d^2 as given in the problem.

Hint: There are many equality cases, and we just want to find (at least) one. Setting a variable to be a specific value can simplify the work (and hopefully still allow for a solution). In this case, I set a = 0 a = 0 .

Calvin Lin Staff - 5 years, 3 months ago

Yes , you are right here.They are almost pretty same.

D K - 2 years, 10 months ago

Calvin, of course it's awesome, you rock. But I've been trying to boost my Algebra rating, and almost all Level 4 and Level 5 problems involve minimizing and maximizing weird expressions. I am TERRIBLE with inequalities and optimization, so any tips would be amazing. Also, how was IMO? How did you do?

Finn Hulse - 7 years, 1 month ago

I also solved it with Cauchy-Schwarz and Fermat's Two Square Identity. I like this problem and the solution.

Pieter-Jan Meuris - 7 years, 1 month ago

Exactly the same way.

Akshay Sharma - 5 years, 4 months ago

I had a slightly different approach. I considered the sequences {a, b, c, d} and {-14c, 14d, 20b, 20a} and tried to use the cauchy shwarz inequality. I got:

(a² + b² + c² + d²)(196c² + 196d² + 400b² + 400a²) ≥ (-14ac + 14bd + 20bc + 20ad)²

=> (a² + b² + c² + d²)(196(c² + d²) + 400(b² + a²)) ≥ (-14ac + 14bd + 20bc + 20ad)²

or, (a² + b² + c² + d²)(596(a²+b²))≥(-14ac + 14bd + 20bc + 20ad)²

Which is, 298(a² + b² + c² + d²)² ≥(-14ac + 14bd + 20bc + 20ad)² Hence,

√298 ≥(-14ac + 14bd + 20bc + 20ad)/(a² + b² + c² + d²)

I am not sure why. Any help would be appreciated.

Rishabh Garg - 1 year, 10 months ago

Log in to reply

Remember that to show something is a maximum, we have to show that
1. It is an upper bound.
2. It is the least upper bound (IE that it can be achieved).

You have done 1, but not 2 as yet. Saying that 298 \sqrt{ 298} is an upper bound does not contradict the claim that 149 \sqrt{149} is the least upper bound.

Did you check the equality condition for your Cauchy Schwarz? When does equality hold?

Calvin Lin Staff - 1 year, 10 months ago

Log in to reply

I tried to check for an equality condition but I don't think it exists. I am beginning to see why my approach was not working.

So does this mean if I take any two sets of terms and apply the Cauchy Shwarz inequality, the equality condition is not guaranteed to exist?

Also, are there any further implications of such a case, where the equality condition does not exist in the Cauchy-Shwarz inequality?

Rishabh Garg - 1 year, 10 months ago

Log in to reply

@Rishabh Garg Right, your approach produced an upper bound but doesn't yet show that it is the least upper bound (and you can't because it is not the least upper bound).

Yes, and this is true more generally. As a simpler example to consider, suppose you wanted to minimize f ( x ) = g ( x ) + h ( x ) f(x) = g(x) + h(x) , where g ( x ) = x 2 g(x) = x^2 and h ( x ) = ( x 2 + 4 x + 4 ) h(x) = (x^2 + 4x + 4 ) . We know that 0 is the minimum of g ( x ) g(x) and 0 is the minimum of h ( x ) h(x) .
1. Is it true that 0 is a lower bound of f ( x ) f(x) ?
2. is it true that 0 is the minimum of f ( x ) f(x) ?
3. What is the minimum of f ( x ) f(x) ?

You raised a good question. If the equality condition of g ( x ) g(x) and h ( x ) h(x) have a non-empty intersection, what can we say about the min/max of g ( x ) + h ( x ) g(x) + h(x) ?
E.g. We can say that m a x ( g ( x ) + h ( x ) ) m a x g ( x ) + m a x h ( x ) max ( g(x) + h(x) ) \leq max g(x) + max h(x) .

Calvin Lin Staff - 1 year, 10 months ago
Daniel Liu
Apr 10, 2014

Off the bat, seeing a 2 + b 2 = c 2 + d 2 a^2+b^2=c^2+d^2 makes us think of the identity sin 2 θ + cos 2 θ = 1 \sin^2\theta+\cos^2\theta=1 . But will substituting simple variables as complicated trigonometric functions help? Let's find out.

First, to make the substitution, let's define a variable x x such that a 2 + b 2 = c 2 + d 2 = x 2 a^2+b^2=c^2+d^2=x^2 . Thus, we can substitute a = x sin α a=x\sin\alpha , b = x cos α b=x\cos\alpha , c = x sin β c=x\sin\beta , and d = x cos β d=x\cos\beta . Cool, we reduced it to only three variables to juggle around!

Substituting these values in the expression we want to maximize, along with substituting a 2 + b 2 = x 2 a^2+b^2=x^2 and c 2 + d 2 = x 2 c^2+d^2=x^2 , we obtain that the expression is equivalent to

20 x 2 cos α sin β + 14 x 2 cos α cos β + 20 x 2 sin α cos β 14 x 2 sin α sin β 2 x 2 \dfrac{20x^2\cos\alpha\sin\beta + 14x^2\cos\alpha\cos\beta + 20x^2\sin\alpha\cos\beta - 14x^2 \sin\alpha\sin\beta}{2x^2}

Hey look, the 2 x 2 2x^2 cancel out! We know know that we are definitely heading the right direction, with only two variables to handle now. We are left with the expression

10 cos α sin β + 7 cos α cos β + 10 sin α cos β 7 sin α sin β 10\cos\alpha\sin\beta + 7\cos\alpha\cos\beta + 10\sin\alpha\cos\beta - 7 \sin\alpha\sin\beta

Seeing the coefficients, we should group the pairs together. Grouping gives

10 ( sin α cos β + sin β cos α ) + 7 ( cos α cos β sin α sin β ) 10(\sin\alpha\cos\beta + \sin\beta\cos\alpha)+7(\cos\alpha\cos\beta-\sin\alpha\sin\beta)

Incredible! We recognize the expressions inside the parentheses as two of the four angle sum/difference identities! Applying the two applicable identities in reverse, we obtain

10 sin ( α + β ) + 7 cos ( α + β ) 10\sin(\alpha+\beta)+7\cos(\alpha+\beta)

We can now substitute even further by letting θ = α + β \theta=\alpha+\beta , leaving us with a one-variable expression to maximize. We want to maximize

10 sin θ + 7 cos θ 10\sin\theta+7\cos\theta

However, it isn't immediate how to maximize this. We can't use sin 2 θ + cos 2 θ = 1 \sin^2\theta+\cos^2\theta=1 because the coefficients are different (even if they are the same, we actually still cannot obtain the maximum anyways), nor can we use sin θ = cos ( 90 θ ) \sin\theta=\cos(90-\theta) because that would make the angle value of the two expression be different...

Well, what if 10 = cos φ 10=\cos\varphi and 7 = sin φ 7=\sin\varphi for some variable φ \varphi ? We could use the sine angle sum identity one more time to reduce it to one trig function! However, it's clear that there isn't any real φ \varphi that satisfies either equation.

However, if we scale the entire expression down, maybe there will exist a φ \varphi ! So let's scale the expression down by a factor of S S to obtain

S ( 1 S 10 sin θ + 1 S 7 cos θ ) S\left(\dfrac{1}{S}\cdot 10\sin\theta+\dfrac{1}{S}\cdot 7\cos\theta\right)

Now, let' set cos φ = 1 S 10 \cos\varphi = \dfrac{1}{S}\cdot 10 and sin φ = 1 S 7 \sin\varphi = \dfrac{1}{S}\cdot 7 .

Now we can use sin 2 φ + cos 2 φ = 1 \sin^2\varphi + \cos^2\varphi=1 to find the value of S S : ( 1 S 10 ) 2 + ( 1 S 7 ) 2 = 1 100 S 2 + 49 S 2 = 1 149 = S 2 S = 149 \begin{aligned}\left(\dfrac{1}{S}\cdot 10\right)^2+\left(\dfrac{1}{S}\cdot 7\right)^2&=1\\ \dfrac{100}{S^2}+\dfrac{49}{S^2}&=1\\ 149&=S^2\\ S&=\sqrt{149} \end{aligned}

Thus, our expression turns into 149 ( sin θ cos φ + sin φ cos θ ) \sqrt{149}(\sin\theta\cos\varphi+\sin\varphi\cos\theta) which is simply 149 sin ( θ + φ ) \sqrt{149}\sin(\theta+\varphi)

The maximum value of this expression is clearly 149 \sqrt{149} , as sin ( θ + φ ) 1 \sin(\theta+\varphi) \le 1 . Thus, the maximum of our original expression is also 149 \sqrt{149} and our final answer is 149 \boxed{149} . \Box


Problem Maker's Note: Using this method, we can also very easily find the minimum value of our expression. Because sin ( θ + φ ) 1 \sin(\theta+\varphi)\ge -1 , the minimum value is simply 149 -\sqrt{149} !

Well I used complex numbers approach to yield a somewhat similar answer. Nevertheless, good solution Daniel. +1 for it.

Sudeep Salgia - 7 years, 2 months ago

Well, I just immediately thought of Cauchy-Schwarz. This is really a great and different solution! One more thing to say is, the maximum value of 10 sin θ + 7 cos θ 10\sin \theta+7\cos\theta can also be easily get by the R-Method , it will be 1 0 2 + 7 2 = 149 \sqrt{10^2+7^2}=\sqrt{149}

Christopher Boo - 7 years, 1 month ago

Exactly what I did!! I dnt think it was reqd to write anything after 10 s i n θ + 7 c o s θ 10sin\theta+7cos\theta Anyways, awesome!

Pranjal Jain - 6 years, 7 months ago

In a same way I solved the problem as you did for getting answer very quickly

Taisanul Haque - 2 years, 7 months ago

Hmm... I wonder if there's a pattern. Seeing as the answer was 1 0 2 + 7 2 10^2+7^2 and those were both two very critical parts of the expression...

Finn Hulse - 7 years, 2 months ago

Log in to reply

Hint: Cauchy Schwarz.

This is also the right place to reiterate one of my favorite algebraic identities, namely Fermat's Two square identity.

Calvin Lin Staff - 7 years, 2 months ago

Log in to reply

I see... kind of. :P

Finn Hulse - 7 years, 2 months ago

Yep, there indeed is a pattern.

Let's try to maximize the expression a sin θ + b cos θ a\sin\theta+b\cos\theta We first multiply by the scale factor, like in the solution. We obtain the expression S ( a S sin θ + b S cos θ ) S\left(\dfrac{a}{S}\sin\theta+\dfrac{b}{S}\cos\theta\right)

Now we let cos φ = a S \cos\varphi = \dfrac{a}{S} and sin φ = b S \sin\varphi = \dfrac{b}{S} . We use sin 2 φ + cos 2 φ = 1 \sin^2\varphi+\cos^2\varphi = 1 to get a 2 + b 2 S 2 = 1 \dfrac{a^2+b^2}{S^2}=1 or S = a 2 + b 2 S=\sqrt{a^2+b^2} . Thus, we have proved that the maximum of a sin θ + b cos θ a\sin\theta+b\cos\theta or any real numbers a , b a,b is a 2 + b 2 \sqrt{a^2+b^2} .

EDIT: sniped by Calvin. However, my solution is different than both of his hints.

Daniel Liu - 7 years, 2 months ago

Log in to reply

My comment yields a 3 line solution to your problem. You are nearly there, you just got sidetracked by doing trignometric substitutions.

Have fun figuring out the solution. If you don't get it by Sunday, give me a shout and I'd post mine.

Calvin Lin Staff - 7 years, 2 months ago

Log in to reply

@Calvin Lin Darn, I was hoping there wouldn't be any easier solutions, because my solution was the one I intended. I'm pretty shaky at C-S, so I don't see the solution right now... I'll see what I can think up.

Daniel Liu - 7 years, 2 months ago

This is also the maximum for a sin k θ + b cos k θ . a\sin k\theta+b\cos k\theta.

Trevor B. - 7 years, 2 months ago

i've used a=b=c=d=x, but it gave me wrong answer........why so?

he man - 7 years, 1 month ago

Log in to reply

The equality case for optimization problems does not have to be where the variables are all equal! There are many inequalities where the inequality case is not where all the variables are the same. Quoting from @Calvin Lin :

This is done by setting a = 0 , b = 596 , c = 20 , d = 14 a=0,b=\sqrt{596},c=20,d=14 . Hence, the maximum is 149 \sqrt{149} .

As you can see, the equality case is rather wonky and unforeseeable.

Daniel Liu - 7 years, 1 month ago

I used complex numbers and trigonometry to get here

Ashtik Mahapatra - 7 years, 1 month ago
Sarith Imaduwage
Feb 12, 2016

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...