1

Find all the rational roots of the polynomial $p(x)=2x^4-5x^3+7x^2-25x-15$. I only found $x=3, -\frac{1}{2}$. I am not sure whether there is any other rational roots.

Is there a way to tell whether these are the ONLY rational roots other than doing the long polynomial division?

Thanks.

user71346
  • 4,171
  • @ClaudeLeibovici. That's what I was asking. Can we tell without doing the division? – user71346 May 05 '16 at 12:12
  • 1
    But doing the division is easy. It should take you about 10 seconds! – almagest May 05 '16 at 12:13
  • 2
    By the rational root theorem, for any rational root p/q in lowest terms p must divide 15 and q must divide 2. This leaves you with a small list of possibilities. – Ori May 05 '16 at 12:13
  • 1
    And as you've already found $2$ roots, there shouldn't be a problem solving the resulting quadratic. – Nikunj May 05 '16 at 12:13
  • 1
    Just to be clear, the resulting quadratic is $x^2+5$. Finding that and noting it has no other rational (or real) solutions is by far the fastest way of solving the problem. – almagest May 05 '16 at 12:14
  • If you have time, you can apply Sturm's algorithm to check there are only $2$ real roots… – Bernard May 05 '16 at 12:16

7 Answers7

4

You can use Vieta's relations: the sum of the roots (in $\mathbf C$) is $\frac 52$. Now the sum of the found roots is already $\frac 52$. So if there were other rational roots, they would be opposite. Now the equation can be written as $$2x^4+7x^2+15=5x(x^2+5),$$ and having opposite roots would imply they're roots of $x^2+5$, which has no real root.

KonKan
  • 7,344
Bernard
  • 175,478
  • Clever solution! – Ege Erdil May 05 '16 at 12:28
  • @Konkan: I didn't go through factorisation, if you read well: I only separate the part with negative coefficients from the part with positive coefficients. – Bernard May 05 '16 at 13:11
  • @Bernard, ok it is a clever trick indeed, but i think it is essentially the same argument as factorization or long division itself only in a different wrapping. And the OP asked explicitly for something different. – KonKan May 05 '16 at 15:49
  • @Konkan: I'm sorry I don't see how separating positive coefficients from negative ones has something to do with factorisation of the polynomial. The argument is that on the r.h.s., after $x$ has been factored out, there can be no root. And probably, it works because the initial polynomial has degree $4$. – Bernard May 05 '16 at 16:10
3

You can use the Rational Roots Test:

Lemma Let $P(X)=a_nX^n+...+a_1x+a_0$ be a polynomial with integer coefficient and $m,n$ integers. If $x =\frac{m}{k}$ is a root of $P(X)$ then $$m|a_0 \,;\, k|a_n$$

The proof is a pretty simple divisibility problem.

For your exercise, the Lemma tells you that $m \in \{ \pm 1, \pm 3, \pm 5 \pm 15 \}$ and $k \in \{1, 2\}$.

Therefore, the only potential rational roots are $$\{ \pm 1, \pm 3, \pm 5 ,\pm 15, \pm \frac12 , \pm \frac32, \pm \frac52, \pm \frac{15}2 \}$$

The only thing you need is now to test all of them. ANd remember, you don't need to necessarily calculate them, you only need to see if you get 0 or not [For example $P(-\frac{15}{2})$ is clearly much larger than 0].

N. S.
  • 132,525
2

$$p(x)=2x^4-5x^3+7x^2-25x-15$$ $$p'(x)=8x^3-15x^2+14x-25$$ $$p''(x)=24x^2-30x+14$$ The discriminant of $p''(x)$ is, $$D=30^2-4\cdot14\cdot24=900-1344\lt0$$ Thus, $p''(x)\gt0$ for all $x\in\mathbb R$. Thus, $p'(x)$ is a strictly increasing function.

Thus, the $p(x)$ is concave up everywhere and can cut the $x$-axis at most twice. Hence, $p(x)$ has at most two real, and hence at most two rational roots.

GoodDeeds
  • 11,185
  • 3
  • 22
  • 42
1

Alternate method: Exhaust search of potential rational roots

Imagine the polynomial can be written as four linear factors with integer coefficients:

$$p(x)=(a_1x-b_1)(a_2x-b_2)(a_3x-r_3)(a_4x-b_4)$$

If you expand this out you can see that $a_1a_2a_3a_4=2$ and $r_1r_2r_3r_4=-15$.

So the possible values for the $a_i$'s are $\pm1$ and $\pm2$. Also the possible values for the $b_i$'s is $\pm1$, $\pm3$, $\pm5$, $\pm15$.

The roots of the above factorized polynomial will be $x=\frac{b_i}{a_i}$

So the only possible rational roots are: $\pm1$,$\pm3$,$\pm5$,$\pm15$,$\pm\frac{1}{2}$,$\pm\frac{3}{2}$,$\pm\frac{5}{2}$,$\pm\frac{15}{2}$

You can then exhaustively test these 16 potential roots to find that the two you found are the only ones.

$p(1)=-36$

$p(-1)=24$

$p(3)=0$

$p(-3)=420$

$p(5)=660$

$p(-5)=2160$

$p(15)=85560$

$p(-15)=120060$

$p(\frac{1}{2})=-\frac{105}{4}$

$p(-\frac{1}{2})=0$

$p(\frac{3}{2})=-\frac{87}{2}$

$p(-\frac{3}{2})=\frac{261}{4}$

$p(\frac{5}{2})=-\frac{135}{4}$

$p(-\frac{5}{2})=\frac{495}{2}$

$p(\frac{15}{2})=4410$

$p(-\frac{15}{2})=\frac{36015}{4}$

Ian Miller
  • 11,844
0

Alternate method: Polynomial Short Division

For $p(x)=2x^4−5x^3+7x^2−25x−15$ you have found two factors: $(x-3)$ and $(2x+1)$.

Like you would do for long division short division needs us to multiple the two factors together:

$$(x-3)(2x+1)=2x-5x-3$$

Next start writing $p(x)$ as a multiple of this new quadratic expression. Compare the first terms: $2x^4$ and $2x^2$ the factor there is $x^2$. So:

$$2x^4−5x^3+7x^2−25x−15=x^2(2x^2-5x-3)+10x^2−25x−15$$

Adjust the bit at the end as needed for equality. Then repeat this step on the next largest term: $10x^2$ compared to $2x^2$ gives a factor of 5. So:

$$2x^4−5x^3+7x^2−25x−15=x^2(2x^2-5x-3)+5(2x^2−5x−3)+0$$

As you've gotten to $+0$ at the end you can stop. Do one last factorize of the quadratic expression and:

$$2x^4−5x^3+7x^2−25x−15=(x^2+5)(2x^2-5x-3)$$

You can then look at the new bracket and consider if it has any rational factors. It doesn't factorize so you know there are only two.

Ian Miller
  • 11,844
0

Common sense and the rational roots theorem. You have $3, -1/2$ as roots so you know

$p(x)=2x^4-5x^3+7x^2-25x-15 = (x - 3)(x + 1/2)(2x^2 + bx + 10)=(x-3)(2x + 1)(x^2 + b'x + 5)$.

So by rational roots the remaining rational roots, if any are $\pm(1,5)$. Very short list to check.

Although with very little extra work you can figure what $b'$ is and use the quadratic equation.

... or at least that be is not $\pm 6$ so $(x^2 + b'x + 5)$ is not factorable and has no rational roots.

====

The thing people tend to forget about the ration roots test is that although with the rational roots test you get a long list of potential solutions: $\pm 1, \pm 3, \pm 5 \pm 15, \pm 1/2, \pm 3/2, \pm 5/2, \pm 15/2$, the product of the rational roots must multiply to end coefficient divided by the leading coefficint (-15/2) or a factor if there are not all rational roots.

So as you have 3 and -1/2. There are either 2 rational roots and they must multiply to -15/2(3x-1/2)=5, or there are none. So that's either 1 and 5 or -1 and -5 or no more rational roots.

fleablood
  • 124,253
-2

Develop $(x-3)(x+\frac{1}{2})(ax^2+bx+c)$. Find $a,b,c$ by identification. Then, you can calculate the other roots.

Vincent
  • 2,318