Polynomial phobia abandoned!

Algebra Level 4

What is the minimum value of p ( 2 ) p(2) if the following 4 conditions are followed?

  1. p ( x ) p(x) is a polynomial of degree 17 17 .

  2. All roots of p ( x ) p(x) are real.

  3. All coefficients are positive.

  4. The coefficient of x 17 x^{17} is 1.

  5. The product of roots of p ( x ) p(x) is -1.

Image Credit: Wikimedia Septic Graph


The answer is 129140163.

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

Aditya Raut
Mar 4, 2015

Here's a proof by generalization.

Let p ( x ) = x 17 + a 1 x 16 + . . . . + a 15 x 2 + a 16 x 1 p(x) = x^{17} +a_1x^{16}+ ....+ a_{15}x^2+a_{16}x-1 such that a i R + a_i \in \mathbb{R}^+

\bullet As all coefficients are positive, p ( x ) 1 p(x)\geq 1 whenever x 0 x \geq 0

\implies All the roots are Negative real numbers . Let the roots be α 1 , α 2 , . . . , α 17 -\alpha_1 , -\alpha_2 , ... , -\alpha_{17} .

\bullet Thus let p ( x ) = ( x + α 1 ) ( x + α 2 ) . . . ( x + α 17 ) p(x) = (x+\alpha_1 )(x+\alpha_2)...(x+\alpha_{17}) where α i R + \alpha_i \in \mathbb{R}^+

\implies For a real number t t , p ( t ) = ( t + α 1 ) ( t + α 2 ) . . . ( t + α 17 ) p(t) = (t+\alpha_1)(t+\alpha_2)...(t+\alpha_{17})


Now we apply the AM-GM inequality in a cute way whenever t t is integer ... 1 + 1 + . . + 1 ’t’ ones + α 1 t + 1 1 1 . . . . 1 ’t’ ones α 1 t + 1 \dfrac{\overbrace{1+1+..+1}^\text{'t' ones}+\alpha_1}{t+1} \geq \sqrt[t+1]{\underbrace{1\cdot1\cdot.... \cdot1}_\text{'t' ones} \cdot \alpha_1} 1 + 1 + . . + 1 ’t’ ones + α 2 t + 1 1 1 . . . . 1 ’t’ ones α 2 t + 1 \dfrac{\overbrace{1+1+..+1}^\text{'t' ones}+\alpha_2}{t+1} \geq \sqrt[t+1]{\underbrace{1\cdot1\cdot.... \cdot1}_\text{'t' ones} \cdot \alpha_2} . . . ... 1 + 1 + . . + 1 ’t’ ones + α 17 t + 1 1 1 . . . . 1 ’t’ ones α 17 t + 1 \dfrac{\overbrace{1+1+..+1}^\text{'t' ones}+\alpha_{17}}{t+1} \geq \sqrt[t+1]{\underbrace{1\cdot1\cdot.... \cdot1}_\text{'t' ones} \cdot \alpha_{17}}

Taking the product of all, ( t + α 1 ) ( t + α 2 ) . . . ( t + α 17 ) ( t + 1 ) 17 α 1 α 2 . . . α 17 t + 1 \dfrac{(t+\alpha_1)(t+\alpha_2)...(t+\alpha_{17})}{(t+1)^{17}} \geq \sqrt[t+1]{\alpha_1\cdot\alpha_2\cdot...\cdot \alpha_{17}}


But the product of all the roots is 1 1 as given in question.

( t + α 1 ) ( t + α 2 ) . . . ( t + α 17 ) ( t + 1 ) 17 1 \therefore \dfrac{(t+\alpha_1)(t+\alpha_2)...(t+\alpha_{17})}{(t+1)^{17}} \geq 1

( t + α 1 ) ( t + α 2 ) . . . ( t + α 17 ) ( t + 1 ) 17 \therefore (t+\alpha_1)(t+\alpha_2)...(t+\alpha_{17}) \geq (t+1)^{17}

p ( t ) ( t + 1 ) 17 \therefore p(t) \geq (t+1)^{17}

That gives us p ( 2 ) 3 17 p(2) \geq 3^{17} , minimum value of p ( 2 ) = 3 17 = 129140163 p(2) = 3^{17} = \boxed{129140163}


Note - For a polynomial of degree n n with specified conditions, p ( t ) ( t + 1 ) n p(t) \geq (t+1)^n

Nice question !!

Took me about 4-5 mins before I stumbled upon the "cute" way ! Did it strike you in the beginning itself to proceed in that way ?

A Former Brilliant Member - 6 years, 3 months ago

I feel that the solution to this question is incorrect. Here are some of the flaws:

1. 1. For applying the AM-GM \text{AM-GM} inequality a l l \mathbf{all} the numbers must be positive. However according to the given solution all the α i ’s \alpha_{i}\text{'s} are negative.

2. 2. t t is a real number and not necessarily an integer.

3. 3. The product of the roots is 1 -1 and n o t \mathbf{not} 1 1 as mentioned in the solution.

Please look into the mater and correct me if I am wrong. ¨ \ddot\smile

Karthik Kannan - 6 years, 3 months ago

Log in to reply

Concerning flaw 1 1 : he is minimizing each term of the polynomial separately, t + α i t+\alpha_i , which is positive. With points 2 2 and 3 3 however, I think you're correct. @Aditya Raut , care to clarify?

Ryan Tamburrino - 6 years, 3 months ago

Log in to reply

Well yeah, Only the third point of your argument is an issue (Because we wanted p ( 2 ) p(2) so we actually did take t t as integer. About the 3rd point, I've clarified the question and made change in the answer. Thank you very much for pointing that out :)

Aditya Raut - 6 years, 3 months ago

Log in to reply

@Aditya Raut Awesome! Great problem and clever use of AM-GM!

Ryan Tamburrino - 6 years, 3 months ago

Thank you, for bringing that to my notice, I've edited things now.

Aditya Raut - 6 years, 3 months ago

Log in to reply

I still feel that flaw 1. 1. is the biggest flaw in the solution.. Also if the product of the roots is made 1 1 you cannot ensure that all the coefficients are positive. Please clarify @Aditya Raut .

Karthik Kannan - 6 years, 3 months ago

Nice exercise. I tried to solve it, 1 year and 3 months later. The condition (4) should state: the product of all roots is -1. Interestingly with +1 and asking for positive coefficients for all but the constant, the minimum does not exist. The infimum is (2+2^(4/17))^16*(2-2^(-64/17)) = 207658223.5..., for which the coefficient of x is vanishing.

Jochem König - 5 years ago

Just brilliant!!

Adarsh Kumar - 6 years, 3 months ago

How can you be sure that all the roots are negative real numbers??

Raghav Vaidyanathan - 6 years, 3 months ago

Log in to reply

See , the least value the function attains is p ( 0 ) = 1 p(0) = 1 and after that x > 0 \forall x > 0 the function will increase above 1 , hence there will be no roots which are positive . So , if at all roots exist for p ( x ) p(x) , it has to be in the negative x axis .

I hope I made myself clear :)

A Former Brilliant Member - 6 years, 3 months ago

Log in to reply

Thanks Roopesh! Don't know what I was thinking! And I have posted an alternative solution(which I used) to solve this question. Please see if it is correct(if you have time that is).

Raghav Vaidyanathan - 6 years, 3 months ago

@Aditya Raut

Raghav Vaidyanathan - 6 years, 3 months ago

Log in to reply

Yes Raghav, Azhaghu Roopesh M is correct. For a root, value of polynomial is 0 0 . But now for positive coefficients, the value can't be zero for any positive real number, not even for zero. So if at all there are real roots, they'll be negative and given that all roots are real, so all are negative :)

Aditya Raut - 6 years, 3 months ago

Log in to reply

@Aditya Raut Thank you. There is an alternative way to solve this question. Refer to my solution.

Raghav Vaidyanathan - 6 years, 3 months ago

How can the last term be 1 if product of roots is 1. The last term should be -1 :p

parv mor - 5 years, 10 months ago

Sorry, too many dark points ....

José Ramón de Diego Luis - 3 years, 2 months ago

We can also do this problem by repeatedly using Vieta's formula and then using AM-GM on each coefficient.

For example:

q ( x ) = x n + a n 1 x n 1 + . . . + a 2 x 2 + a 1 x + 1 q(x) =x^n+a_{n-1}x^{n-1} +...+a_2x^2+a_1x+1

q ( x ) = ( x + α 1 ) ( x + α 2 ) . . . ( x + α 17 ) q(x) = (x+\alpha_1 )(x+\alpha_2)...(x+\alpha_{17}) where α i R + \alpha_i \in \mathbb{R}^+

Since all the roots are of same sign(-ve), we can do this for all r r (please reason this out yourself).

sum of (products of α taken ’r’ at a time) ( n r ) product of (products of α taken ’r’ at a time) ( n r ) \frac { \sum {\text{sum of (products of }\alpha \text{ taken 'r' at a time)} } }{ \left( \begin{matrix} n \\ r \end{matrix} \right) } \ge \sqrt [ \left( \begin{matrix} n \\ r \end{matrix} \right)]{\text{ product of (products of } \alpha \text{ taken 'r' at a time)} }

a r ( n r ) 1 a r ( n r ) \Rightarrow \frac { { a }_{ r } }{ \left( \begin{matrix} n \\ r \end{matrix} \right) }\ge 1 \\ \Rightarrow { a }_{ r }\ge \left( \begin{matrix} n \\ r \end{matrix} \right)

The funny part is that the minimum occurs for each and every a r a_r for the same condition, that is, in this case: all roots= 1 -1

Hence, the required polynomial is ( x + 1 ) 17 (x+1)^{17}

Which yields 3 17 \boxed{3^{17}} at x = 2 x=2 .

Log in to reply

This is totally cool, I had thought of that but I was totally stumped in proving the thing. So I thought that a r ( n r ) a_r \geq \binom{n}{r} is just a derivation from the proof I wrote, but now you've given it! Cool :D

Aditya Raut - 6 years, 3 months ago

Nice method , Raghav. :)

A Former Brilliant Member - 6 years, 3 months ago

Lovely. Hats off to you.

Shashwat Shukla - 6 years, 3 months ago

But for some sum of roots taking k at a time are negative. So how can we apply AM GM .

Aaron Jerry Ninan - 4 years, 7 months ago

I wanted to try (2+1) to the 17th power when I first saw the problem, but then I saw the Applying AM-GM at the bottom of the screen so I decided to do it this way, which took me like 5 minutes to compute all the combinations...lol

Yashas Ravi - 2 years, 1 month ago
Anand O R
Jan 1, 2017

This is my solution Please tell me where i went wrong

Remember that a number is a minimum if it satisfies the following conditions:
1. It is a lower bound.
2. It can be achieved.

You have only shown that we have a lower bound. You have not shown that it can be achieved.

Calvin Lin Staff - 4 years, 5 months ago

Log in to reply

Thanks! @Calvin Lin

Anand O R - 4 years, 4 months ago

Calvin lin

Topper Forever - 4 years, 3 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...