10

I have a polynomial $P(x) = -x^3+12x+3$, and I am asked to find the turning points of it, and hence state how many zeroes it has. Since this chapter is separate from calculus, we are expected to solve it without differentiation.

However, I'm not sure how I could solve this. I'm having trouble factorising it as well since the zeroes seem to be irrational.

TerryA
  • 211
  • Use Descartes rule of signs to find the number of positive and negative roots... Just testing a few values should give you enough to conclude on number and rough location of roots. A sketch helps too. – Macavity Apr 20 '16 at 04:13
  • For the answers below, I'm just worried about triple roots, which you might want avoid... – Simply Beautiful Art Apr 21 '16 at 00:18

4 Answers4

21

You want to know for which $c$ it is the case that $P(x)+c$ has a double root. We could mess around with the discriminant of the cubic, but that's probably too much work. Instead, suppose $P(x) + c = -(x-a)^2(x-b)$, so that $$ -x^3 + 12 x + 3 + c = - x^3 + (2a+b)x^2 -(a^2 + 2ab)x +a^2 b $$ From this, we read off $2a+b = 0$, $a^2 + 2ab = -12$, and $3+c = a^2 b$. From the first two, solutions $(a,b)$ are $(-2,4)$ and $(2,-4)$. We don't even need to solve for $c$ because the double root (the turning point) occurs at $x=a$, so the turning points are $(-2,P(-2)) = (-2, -13)$ and $(2,P(2)) = (2,19)$.

Barry
  • 2,268
Eric Towers
  • 67,037
  • Beat me by 30 s or so I think. +1 – Macavity Apr 20 '16 at 05:03
  • 3
    @Macavity : "Great minds ..." and all that... :-) – Eric Towers Apr 20 '16 at 05:06
  • Interesting. This seems to be the next topic of the chapter so I haven't actually learnt it yet, but thank you for your answer it looks very clean :). Just one question though, why do we want to have a double root? – TerryA Apr 20 '16 at 05:08
  • 2
    @TerryA : Draw a "random" cubic with two turning points and add a horizontal line through one of the turning points. When we translate, the $x$-axis lands on the horizontal line and the turning point is a double root. – Eric Towers Apr 20 '16 at 05:15
  • @TerryA If we have a multiplicity of $2$ or more of one root, then at that root, the graph turns, hence, a turning point. The value $c$ is meant to make your turning point a root so that we may observe this property. (If the multiplicity is even, it is a turning point, if it is odd, there is no turning, only an inflection point I believe.) – Simply Beautiful Art Apr 21 '16 at 00:15
  • @EricTowers Do we have to worry about $b=a$? From my thoughts, this should make that point not an actually turning point... – Simply Beautiful Art Apr 21 '16 at 00:16
  • Generally, we should check for that. Such a triple root is not a turning point. Didn't happen in this problem. Also, these are easy to detect. The coefficients are binomial coefficients, from expanding $(x-a)^3$, except for the constant, which may be shifted. – Eric Towers Apr 21 '16 at 12:38
10

For a cubic, any turning point is a double root of a suitable translation on the Y-axis. i.e. we seek double roots of $P(x)+k = -x^3+12x+3+k$ for some value of $k$. So we must have for some $a, b$, $$-x^3+12x+3+k = (x-a)(x-b)^2$$

Equating coefficients, we get $$a+2b=0,\quad 2ab+b^2=-12, \quad ab^2 = 3+k$$ The first two gives you $b=\pm2$, which are the turning points!


Here is another way with AM-GM. Note that from symmetry it is enough to find the positive turning point of $f(x) = -x^3+12x$. As $f(0) = 0, f(1) = 11, f(2\sqrt3) = 3$, we must have a local maximum in $[1, 2\sqrt3]$.

So we maximize $2f^2 = (2x^2)(12-x^2)(12-x^2)$ which is a product of three positive terms with a constant sum, which means the maximum is when the terms are all equal, viz. $2x^2=12-x^2 \implies x^2=4$.

Macavity
  • 46,381
2

. Do this: let $Q(x)=P(x)-3=-x^3+12x=-x(x+\sqrt{12})(x-\sqrt{12})$. This has $2$ turning points, because there are three roots, hence there are two turning points between the two pairs of consecutive roots. $P$ is just a translation of $Q$, hence has $2$ turning points as well.

To find the turning points, I thought I would make this little manipulation on the negative of the function: $x^3-12x-3 = (x^3+3x^2+3x+1) -(3x^2+15x+4)=(x+1)^3 - (3x^2+15x+4)$. Now $(x+1)^3$ changes sign only at $-1$, so it is down to us to see the behaviour of the function $3x^2+15x+4$. This factorizes as $3(x+\frac{5}{2})^2 - 14.75$. Finally, $x^3-12x-3 = (x+1)^3 - 3(x+\frac{5}{2})^2 + 14.75$.

This inspection tells us that the turning points on both sides are between $1$ and $2.5$ by plugging these values in. I'm not sure more can be done analytically other than "imitate calculus".

0

If you know how to differentiate, just find the roots of $P'(x) = -3x^2+12 $, which are $x = \pm 2$.

If you don't, $P(x) = -x^3+12x+3 $, so, imitating the invention of calculus,

$\begin{array}\\ P(x+h) &= -x^3+12x+3\\ &= -(x+h)^3+12(x+h)+3\\ &= -x^3-3x^2h-3xh^2-h^3+12x+12h+3\\ &= (-x^3+12x+3)-h(3x^2+3hx+h^2-12)\\ &= P(x)-h(3x^2+3hx+h^2-12)\\ \end{array} $

so, as expected $\dfrac{P(x+h)-P(x)}{h} =-(3x^2+3hx+h^2-12) $.

Letting $h \to 0$, we find that the turning point of $P$ are when $3x^2 = 12$ or $x = \pm 2$.

There is probably a clever way to do this, and this ain't it.

marty cohen
  • 107,799