Minimum of absolute value of quadratic

Algebra Level 5

Consider all monic polynomials f ( x ) = x 2 + b x + c f(x) = x^2 + bx + c , where b b and c c are real numbers. What is the minimum value of N N , where

N = max x [ 10 , 10 ] f ( x ) ? N = \max_{x \in [-10,10]} \vert f(x) \vert?

Details and assumptions

The last equation states: "The maximum value of the absolute value of f ( x ) f(x) , as x x ranges from 10 -10 to 10 10 inclusive".


The answer is 50.

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.

9 solutions

Ang Yan Sheng
May 20, 2014

Note that f ( 10 ) = 100 + 10 b + c f ( 10 ) = 100 10 b + c f ( 0 ) = c \begin{aligned}f(10)&=100+10b+c\\f(-10)&=100-10b+c\\f(0)&=c\end{aligned} Hence, 200 = f ( 10 ) + f ( 10 ) 2 f ( 0 ) f ( 10 ) + f ( 10 ) + 2 f ( 0 ) \begin{aligned}200&=f(10)+f(-10)-2f(0)\\&\leq|f(10)|+|f(-10)|+2|f(0)|\end{aligned} Therefore, it is not possible for each of f ( 10 ) |f(10)| , f ( 10 ) |f(-10)| and f ( 0 ) |f(0)| to be less than 50. Hence N 50 N\geq50 .

On the other hand, by choosing f ( x ) = x 2 50 f(x)=x^2-50 it is clear that 50 f ( x ) 50 -50\leq f(x)\leq50 when x [ 10 , 10 ] x\in[-10,10] . Hence N 50 N\leq50 .

Therefore N = 50 N=50 , as desired.

This beautiful solution is based on the (generalized) triangle inequality. Another correct solution was to optimize c c for a fixed b b . Most common mistake was lack of rigor in justifying that x 2 50 x^2-50 is our best choice.

Calvin Lin Staff - 7 years ago

Nice problem

Venkatesh A - 1 year, 5 months ago

It is easy to see that max x [ 10 , 10 ] f ( x ) \max \limits_{x \in [-10, 10]} |f(x)| occurs at either x = 10 x=-10 , x = 10 x=10 , or x = 0 x=0 , so max x [ 10 , 10 ] f ( x ) { f ( 10 ) , f ( 10 ) , f ( 0 ) } \max \limits_{x \in [-10, 10]} |f(x)| \in \{f(10), f(-10), f(0) \} . Note that f ( 10 ) = 100 + 10 b + c f(10)= 100 + 10b + c f ( 10 ) = 100 10 b + c f(-10)= 100-10b+c f ( 0 ) = c f(0)= c Then, note that f ( 10 ) + f ( 10 ) 2 f ( 0 ) = 200 f(10) + f(-10) - 2f(0)= 200 From the triangle inequality, we have f ( 10 ) + f ( 10 ) 2 f ( 0 ) f ( 10 ) + f ( 10 ) + 2 f ( 0 ) f(10) + f(-10) - 2f(0) \leq |f(10)| + |f(-10)| + 2|f(0)| f ( 10 ) + f ( 10 ) + 2 f ( 0 ) 200 \implies |f(10)| + |f(-10)| + 2|f(0)| \geq 200 Let a = max ( f ( 10 ) , f ( 10 ) , f ( 0 ) a= \max (|f(10)|, |f(-10)|, |f(0)| . Then, f ( 10 ) + f ( 10 ) + 2 f ( 0 ) 4 a |f(10)| + |f(-10)| + 2|f(0)| \leq 4a 4 a 200 a 200 \implies 4a \geq 200 \implies a \geq 200 This shows that N 50 N \geq 50 . It can be checked that equality holds for f ( x ) = x 2 50 f(x)= x^2 - 50 . Thus N = 50 \boxed{N=50} is the smallest possible value of N N .

Why it occurs at x = 10 , 0 , 10 x= -10,0,10 . ??

A Brilliant Member - 7 years, 7 months ago

Log in to reply

Consider the graph of f ( x ) |f(x)| from 10 -10 to 10 10 . It looks like a parabola, whose lower part is flipped above the x x axis. Now, max x [ 10 , 10 ] f ( x ) \max \limits_{x \in [-10, 10]} |f(x)| will occur either at the vertex of the parabola or at its two edges. In the latter two cases, it is obviously attained for either x = 10 x=-10 or x = 10 x=10 . If the maximum occurs at its vertex, note that the vertex must lie on the y y axis, otherwise either f ( 10 ) |f(10)| or f ( 10 ) |f(-10)| will be greater than it. Thus, the maximum occurs at either 10 -10 , 10 10 , or 0 0 .

Sreejato Bhattacharya - 7 years, 7 months ago

Log in to reply

If the maximum occurs at its vertex, note that the vertex must lie on the y y axis,

Actually, that isn't necessarily true. Consider b = 1 , c = 90 b=1, c=-90 .

Peter Byers - 7 years, 7 months ago

Log in to reply

@Peter Byers We are not talking about all quadratics, we are considering those quadratics f ( x ) f(x) for which the minimum of max x [ 10 , 10 ] f ( x ) \max \limits_{x \in [-10, 10]} |f(x)| is attained. In your example, the vertex of the parabola is ( 1 2 , 361 4 ) \left ( -\frac{1}{2}, \frac{361}{4} \right ) , but we can do better by shifting the vertex of the parabola to ( 1 2 , 50 ) \left ( -\frac{1}{2}, 50 \right ) , so now we have max x [ 10 , 10 ] f ( x ) = 60 \max \limits_{x \in [-10, 10]} |f(x)|= 60 , which is attained for x = 10 x=10 .

Sreejato Bhattacharya - 7 years, 7 months ago

Log in to reply

@Sreejato Bhattacharya Not necessarily true, you cannot ignore all "not better" cases first. Your argument was not true, since if we transform axes a little else where, possibly we can create a bound, where the maximum is not attained at the endpoints or the midpoint.

A Brilliant Member - 7 years, 7 months ago

Log in to reply

@A Brilliant Member Matt's polynomial is also an example in this range.

A Brilliant Member - 7 years, 7 months ago

@Sreejato Bhattacharya Yes and no. If we look at it that way, then the question becomes: Should the proof show that the optimal function is one whose vertex is on the y-axis?

I think you are selling your proof short. Consider what it says without the initial claim that the max is at x = 10 , 0 , x=-10,0, or 10 10 ... you would still have the conclusion that N a 50 N\ge a\ge 50 ! (That's not a factorial.) And then the explicit example still shows that N = 50 N=50 .

Peter Byers - 7 years, 7 months ago

Typo: a 50 a \geq 50

Sreejato Bhattacharya - 7 years, 7 months ago

Log in to reply

It should also read a = max { f ( 10 ) , f ( 10 ) , f ( 0 ) } a = \max \{|f(10)|,|f(-10)|,|f(0)|\} , but that's getting picky xD Good job :)

Sotiri Komissopoulos - 7 years, 7 months ago

Log in to reply

Oh, oops sorry. Silly typo :)

Sreejato Bhattacharya - 7 years, 7 months ago

Another typo: min ( max x [ 10 , 10 ] f ( x ) ) { f ( 10 ) , f ( 10 ) , f ( 0 ) } \min \left ( \max \limits_{x \in [-10, 10]} |f(x)| \right ) \in \{ |f(10)|, |f(-10)|, |f(0)| \}

Sreejato Bhattacharya - 7 years, 7 months ago

If f ( x ) = ( x 1 ) 2 200 f(x) = (x-1)^2 - 200 , then max [ 10 , 10 ] f ( x ) \max_{[-10, 10]} \lvert f(x) \lvert comes at x = 1 x = 1 .

Matt McNabb - 7 years, 7 months ago

Log in to reply

See my comment here . I will be posting a better polynomial than ( x 1 ) 2 200 (x-1)^2-200 soon. :)

Sreejato Bhattacharya - 7 years, 7 months ago

I thought someone might have have posted solution using calculus. However, I am posting my solution here on my friend's solution thread.

f ( x ) = x 2 + b x + c f ( x ) = 2 x + b f(x) = x^2 + bx + c \Rightarrow f'(x)= 2x+b .

If b > 0 b>0 , then f ( x ) > 2 x > 0 f'(x)> 2x > 0 , so in [ 0 , 10 ] [0,10] , the function has a higher rate of increase. So, given any value if f ( 0 ) f(0) , f ( 10 ) f(10) is a much higher value than it would have been in case b 0 b \leq 0 .

If b < 0 b<0 , then f ( x ) < 2 x < 0 f'(x) < 2x < 0 , so in [ 10 , 0 ] [-10,0] , the function has a higher rate of decrease (from 0 0 to 10 -10 ). So, given any value if f ( 0 ) f(0) , f ( 10 ) f(-10) is a much lower value (or f ( 10 ) |f(-10)| is a much higher value) than it would have been in case b 0 b \geq 0 .

We conclude b = 0 b=0 . Now f ( x ) = x 2 + c f(x) = x^2 + c . This is just a vertical transformation of y = x 2 y=x^2 , at c c units from the origin. Now we can say f ( x ) m a x |f(x)|_{max} is attained at x = 0 , ± 10 x= 0,\pm 10 , (& that the function is symmetric about y y -axis). We adjust the graph along y y -axis to observe c = 50 c=-50 is best. This can also be rigorously shown using case work.

A Brilliant Member - 7 years, 7 months ago

I like your solution even though it had some flaws in it. As pointed out, the initial claim is not necessarily true of all polynomials. It can get tricky to fully explain why it is true "for those quadratics for which the minimum of the expression is attained."

Can you make a quick patch to your solution?
Hint: Delete the first sentence entirely. That idea is distracting.

Calvin Lin Staff - 7 years, 7 months ago

The leading coefficient being 1, we know that the smallest height a parabola can have in a 20 unit interval is 100, so putting half of it above and below gives is 50 as the farthest distance from the x axis.

Jerry McKenzie - 3 years, 6 months ago
Anony Moose
May 20, 2014

Note that if b is fixed, then the minimum value is obtained when

c = - \frac 12 (min {x \in [-10,10]} |x^2 + bx| + max {x \in [-10,10]} |x^2 + bx|)

= - \frac 12 (max_{x \in [-10,10]} |x^2 + bx|), since |x^2 + bx| = 0 when x = 0

The resulting minimum value will then be

\frac 12 (max_{x \in [-10,10]} |x^2 + bx|)

The problem is therefore equivalent to minimising (max_{x \in [-10,10]} |x^2 + bx|).

Furthermore, |x^2 + bx| is maximum when x = 10 if b \geq 0, or x = -10 if b \leq 0

WLOG let x, b \geq 0

Then

max_{x \in [0, 10]} |x^2 + bx| = |10^2 + 10b|

Since b \geq 0, the minimum possible value is when b = 0.

Then c = - 50

And so the minimum value of max_{x \in [-10,10]} |f(x)| is obtained when f(x) = x^2 - 50, and the value is 50.

Interesting solution

Calvin Lin Staff - 7 years ago
Jonathon Capps
Nov 3, 2013

A monic polynomial is just a translation of x 2 x^2 which will have it's smallest range between -10 and 10 when the vertex is on the the x-axis. The max would be 100 but we can move the vertex to (0, -50) and then the max is 50.

But what of we move it further down???

Biswadeep Sen - 7 years, 7 months ago

Log in to reply

We are taking the absolute value, so if you move it down further than it will actually go higher when you flip it over the x-axis.

Jonathon Capps - 7 years, 7 months ago

Okay then the valur at x = 0 becomes >50 right???So it attains minimum value at x = 50.......

Biswadeep Sen - 7 years, 7 months ago

Can you explain what you mean in the second sentence?

Calvin Lin Staff - 7 years, 7 months ago

My solution was similar: the monic polynomial can be written as ( x p ) 2 + q (x-p)^2 + q so it is a translation of y = x 2 y=x^2 . Instead of thinking of translating y = x 2 y=x^2 to the best place, we can think of placing a "window" of width 20 20 over the graph of y = x 2 y=x^2 , and then try to find a placement of the window such that the graph reaches the top and the bottom of the window and the window has the minimum possible height. This must be at the bottom of the graph because the gradient of the graph is lower there.

Matt McNabb - 7 years, 7 months ago

Log in to reply

Thank you, I can imagine this in my head easily but trying to put words to it is hard.

Jonathon Capps - 7 years, 7 months ago
Ryan Broder
May 20, 2014

We begin by considering b b . Since x x ranges from -10 to 10, for all x x in the range, there is also x -x . Thus the equation can be written as

max x [ 0 , 10 ] ( x 2 + b x + c , x 2 b x + c ) \displaystyle \max_{x \in [0,10]}{(|x^2+bx+c|, |x^2-bx+c|)} .

As b |b| increases, either x 2 + b x + c |x^2+bx+c| or x 2 b x + c |x^2-bx+c| increases. Thus to decrease max x [ 0 , 10 ] ( x 2 + b x + c , x 2 b x + c ) \displaystyle \max_{x \in [0,10]}{(|x^2+bx+c|, |x^2-bx+c|)} , we must decrease b |b| , and hence b = 0 b=0 .

Now we consider c c . Since b = 0 b=0 , f ( x ) = x 2 + c f(x)=x^2+c is symmetric about the y-axis and since x x ranges from -10 to 10, f ( x ) |f(x)| has critical values at f ( 0 ) |f(0)| and f ( ± 10 ) |f(\pm10)| . The original equation can thus be written as max ( 100 + c , c ) \displaystyle \max{(|100+c|, |c|)} . This is minimized at c = 50 c=-50 . Thus f ( x ) = x 2 50 f(x)=x^2-50 and max x [ 10 , 10 ] f ( x ) = 50 \displaystyle \max_{x \in [-10,10]}{|f(x)|} = 50 .

"As b |b| increases, either x 2 + b x + c |x^2+bx+c| or x 2 b x + c |x^2-bx+c| increases. Thus to decrease max x [ 0 , 10 ] ( x 2 + b x + c , x 2 b x + c ) \displaystyle \max_{x \in [0,10]}{(|x^2+bx+c|, |x^2-bx+c|)} , we must decrease b |b| , and hence b = 0 b=0 ." This does not directly imply that the maximum increases, unless one knows that there is one "worst-case scenario", and it is symmetric.

Calvin Lin Staff - 7 years ago
Halil Arslan
May 20, 2014

This polinamial represents a parabola. If the polinomial is desired to have minimum values in interval -10 < x < 10, it is obvious that the bottom of the curve represented by the polinomial must be at x=0 (mid point of the interval).

df/dx = 2x + b

df/dx|(x=0) = 0 --> b = 0

Therefore;

f(x) = x^2+c

Let's assume |f(10)| = |f(0)| = max|f(x)| and c < 0 since |f(x)| have minimum values in given interval.

100+c = -c --> c = -50 --> f(x) = x^2 - 50

Finally, we can get;

max|f(x)| = |f(10)| = |f(0)| = 50

No real proof is given: "it is obvious that the bottom of the curve represented by the polinomial must be at x=0 (mid point of the interval)." , "Let's assume |f(10)| = |f(0)| = max|f(x)| and c < 0 since |f(x)| have minimum values in given interval."

Calvin Lin Staff - 7 years ago
Trevor B.
Nov 4, 2013

The horizontal shift of f ( x ) f(x) is not relevant for this problem, so we can say that f 0 ( x ) = x 2 + c f_0(x)=x^2+c . In the range [ 10 , 10 ] [-10,10] , f 0 ( x ) f_0(x) has a maximum value at ( ± 10 , f 0 ( ± 10 ) ) (±10,f_0(±10)) . The minimum maximum of the function occurs when the minimum and maximum of f 0 ( x ) f_0(x) have equal absolute values. Thus, c = 50 c=-50 , and the minimum maximum N |N| is 50 \boxed{50} .

Actually the horizontal shift matters. The maximum value of this function can be increased if we move the vertex of the parabola away from y y -axis. Your mere assumption at y-axis,doesn't prove it. Secondly, "the minimum maximum of the function occurs when the minimum and maximum of f 0 ( x ) f_0(x) have equal absolute values" is not necessarily true. You must prove that this indeed is possible, & leads to an extremum.

A Brilliant Member - 7 years, 7 months ago

Log in to reply

What I meant was that f ( x ) f(x) does not have a horizontal shift. The minimum maximum of N |N| for any x [ k , k ] x\in[-k,k] occurs when there is no horizontal shift; the function can be represented by f ( x ) = x 2 + c f(x)=x^2+c . The minimum maximum is achieved when c = k 2 2 c=\frac{-k^2}{2} . k = ± 10 k=±10 , so the answer to this problem is 50 \boxed{50} .

f ( x ) f(x) would have a horizontal shift of s s when the range in which the minimum maximum is being looked for is [ s C , s + C ] [s-C,s+C] where C C is a constant.

When does the smallest maximum absolute value of f 0 ( x ) f_0(x) not occur when the absolute values of the minimum and maximum values of f 0 ( x ) f_0(x) over x [ 10 , 10 ] x\in[-10,10] are equal?

Trevor B. - 7 years, 7 months ago

Can you explain your thinking in not considering at absolute values in the first half?

Can you explain the step which led to "Thus c = 50 c = - 50 "? I don't see how equation f ( 10 ) = f ( 10 ) |f(-10)| = |f(10) | gives us c = 50 c = - 50 .

Calvin Lin Staff - 7 years, 7 months ago

Log in to reply

I'm afraid I don't really understand your first question. Can you quote what you are referencing?

For f 0 ( x ) f_0(x) , there are 3 3 important values: x = 0 x=0 and x = ± 10 x=±10 . Over the range x [ 10 , 10 ] x\in[-10,10] , f 0 ( x ) f_0(x) has its absolute minimum at ( 0 , c ) (0,c) and absolute maxima at ( ± 10 , 100 + c ) (±10,100+c) . For the problem, N = max { c , 100 + c } |N|=\max\{|c|, |100+c|\} . The minimum value of this expression is achieved when c = c + 100 c = 50 |c|=|c+100| \Rightarrow c=-50 . We can say f 0 ( x ) = x 2 50 f_0(x)=x^2-50 . For that function, f ( 0 ) = 50 f(0)=-50 and f ( ± 10 ) = 50 f(±10)=50 . N = max { 50 , 50 } = max { 50 , 50 } = 50 |N|=\max\{|-50|, |50|\}=\max\{50, 50\}=\boxed{50} .

Trevor B. - 7 years, 7 months ago
Suresh Yadav
Jul 26, 2017

We can easily prove by logical thinking that for minimum, coefficint of x must be zero for any value of c.now to find c we have to apply a little graphical analysis and we can easily prove that f(0)=-f(10) for minimum. Interesting question sir!

Srinivas Gunti
Nov 8, 2013

Use symmetry!!

Any other way would simply get one side, f(-10) or f(10), greater than the other.

We know we have an extremum at x = 0.

( d f ( x ) d x ) x = 0 = 0 \Rightarrow (\frac {d f(x)}{dx})_{x=0} = 0

2 x + b = 0 \Rightarrow 2x +b = 0

b = 2 x = 0 \Rightarrow b = -2x = 0

Now we have f ( x ) = x 2 + c f(x) = x^2 + c

f ( 10 ) = f ( 10 ) = 100 + c ; f ( 0 ) = c \Rightarrow f(10) = f(-10) = 100+c ; f(0) = c

Equating absolutes of these, | f(10) | = | f(0) | , we get the minimum as 50 \boxed{50} .

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...