Upper and lower bounds for polynomial zeros using synthetic division

4 replies · 5K views · last reply
Saturnine Zero
Messages
7
Reaction score
0

Homework Statement



Upper Bound[/B]
If all of the numbers in the final line of the synthetic division tableau are non-positive, prove for ##f(b)<0##, no real number ##b > c## can be a zero of ##f##

Lower Bound
To prove the lower bound part of the theorem, note that a lower bound for the negative real zeros of ##f(x)## is an upper bound for the positive real zeros of ##f(−x)##. Applying the upper bound portion to ##f(−x)## gives the result.

Do you see where the alternating signs come in?

Homework Equations



##p(x)=d(x)q(x)+r(x)##
##f(x)=(x − c)q(x)+r##

The Attempt at a Solution



I've been struggling for three weeks with this topic trying to properly understand it and to formulate a sensible and reasonably short question. I'm hoping someone can tolerate this long post and let me know if I'm heading in the right direction.

Upper Bound
My book shows a proof for the upper bound for ##f(b)>0## which makes sense to me, but just states there is a simlilar proof for ##f(b)<0##. So I've attempted to follow the proof in the book and adapt it as follows:

Suppose ##c>0## is divided into f and the resulting last line in the division tableau contains all non-positive numbers.

This means ##f(x)=(x-c)q(x)-r##, where the coefficients of the quotient polynomial and the remainder are non-positive.

If ##b>c##, then ##f(b)=(b-c)q(b)-r##, where ##(b-c)## is positive and ##q(b)## is negative and ##r≤0##.

Hence ##f(b)<0## which shows b cannot be a zero of ##f##. Thus no real number ##b>c## that can be a zero of ##f## as required.
​
Does this show ##c## is an upper bound? That is to have ##f(b)=0##, ##b## would need to be less than it is so that ##b=c## and ##b-c## sum to zero, so ##c## is an upper bound?

Lower Bound
For the lower bound part. I've struggled more. I really didn't see where the alternating signs came in. But I figured this morning I would default to using a numerical example, and I think I'm closer now.

I think the alternating sings of the coefficients come into it because the odd/even powers of the polynomial change the signs in ##f(-x)## and this results in a quotient with coefficients of all the same sign.

So my current understanding is, if the quotient polynomial of ##f(x)## alternates signs, when ##f(-x)## is divided by the negative of the divisor, all the coefficients of quotient polynomial of ##f(-x)## will always have all the same sign, and the upper bound proof can be used.

Choosing a more negative divisor for ##f(x)## would make each coefficient in the quotient either more positive or more negative alternatively, so not approaching a zero. And then dividing by the negative of that divisor for ##f(-x)## would make the coefficients in the quotient either more positive or more negative, so not approaching a zero. Therefore that divisor for ##f(x)## is a lower bound.

I'd like to know if my upper bound proof is correct and if I am on the right track for understanding the lower bound part?

Sorry for the long post, I'm not sure how to cut it down any further.
 
Saturnine Zero said:

Homework Statement



Upper Bound[/B]
If all of the numbers in the final line of the synthetic division tableau are non-positive, prove for ##f(b)<0##, no real number ##b > c## can be a zero of ##f##

Lower Bound
To prove the lower bound part of the theorem, note that a lower bound for the negative real zeros of ##f(x)## is an upper bound for the positive real zeros of ##f(−x)##. Applying the upper bound portion to ##f(−x)## gives the result.

Do you see where the alternating signs come in?

Homework Equations



##p(x)=d(x)q(x)+r(x)##
##f(x)=(x − c)q(x)+r##

The Attempt at a Solution



I've been struggling for three weeks with this topic trying to properly understand it and to formulate a sensible and reasonably short question. I'm hoping someone can tolerate this long post and let me know if I'm heading in the right direction.

Upper Bound
My book shows a proof for the upper bound for ##f(b)>0## which makes sense to me, but just states there is a simlilar proof for ##f(b)<0##. So I've attempted to follow the proof in the book and adapt it as follows:

Suppose ##c>0## is divided into f and the resulting last line in the division tableau contains all non-positive numbers.

This means ##f(x)=(x-c)q(x)-r##, where the coefficients of the quotient polynomial and the remainder are non-positive.

If ##b>c##, then ##f(b)=(b-c)q(b)-r##, where ##(b-c)## is positive and ##q(b)## is negative and ##r≤0##.

Hence ##f(b)<0## which shows b cannot be a zero of ##f##. Thus no real number ##b>c## that can be a zero of ##f## as required.
​
Does this show ##c## is an upper bound? That is to have ##f(b)=0##, ##b## would need to be less than it is so that ##b=c## and ##b-c## sum to zero, so ##c## is an upper bound?
...
Please state the entire problem.

What is ƒ ? What are b and c ? What is being divided by what ?
 
  • Like
Likes   Reactions: StoneTemplePython
SammyS said:
Please state the entire problem.

What is ƒ ? What are b and c ? What is being divided by what ?
OK, thanks for looking, I will try to be a bit clearer:

  • Suppose ##f## is a polynomial of degree ##n ≥ 1##

  • If ##c > 0## is synthetically divided into ##f## and all of the numbers in the final line of the division tableau have the same signs, then ##c## is an upper bound for the real zeros of ##f## . That is, there are no real zeros greater than ##c##.

So ##f## is the polynomial and ##c## is a possible real zero and is the divisor in the synthetic division, so ##f(x)## is being divided by ##(x-c)##, and ##b## would any real number greater than ##c##.

I hope that's better.
 
Saturnine Zero said:
OK, thanks for looking, I will try to be a bit clearer:

  • Suppose ##f## is a polynomial of degree ##n ≥ 1##

  • If ##c > 0## is synthetically divided into ##f## and all of the numbers in the final line of the division tableau have the same signs, then ##c## is an upper bound for the real zeros of ##f## . That is, there are no real zeros greater than ##c##.

So ##f## is the polynomial and ##c## is a possible real zero and is the divisor in the synthetic division, so ##f(x)## is being divided by ##(x-c)##, and ##b## would any real number greater than ##c##.

I hope that's better.
Are you saying that
$$f(x) = (x-c)(a_{n-1} x^{n-1} + a_{n-2} x^{n-2} + \cdots + a_1 x + a_0) + r,$$
where ##a_0, a_1, \ldots, a_{n-2}, a_{n-1}## and ##r## all have the same sign?
 
Last edited:
Ray Vickson said:
Are you saying that
$$f(x) = (x-c)(a_{n-1} x^{n-1} + a_{n-2} x^{n-2} + \cdots + a_1 x + a_0),$$
where ##a_0, a_1, \ldots, a_{n-2}, a_{n-1}## all have the same sign?

Yes, exactly that.

I was just using ##q(x)## as the quotient polynomial ##(a_{n-1} x^{n-1} + a_{n-2} x^{n-2} + \cdots + a_1 x + a_0)##, and ##r## as the remainder as the book does.

I actually found a preview online of the exact two pages from the book which would make it clearer: Theorem 3.11.