6
$\begingroup$

Show that $$f(x) = 1 + x + \frac{x^2}{2!} + \frac{x^3}{3!} + \frac{x^4}{4!} > 0 ~~~ \forall_x \in \mathbb{R}$$

I can show that the first 3 terms are $> 0$ for all $x$:

$(x+1)^2 + 1 > 0$

But, I'm having trouble with the last two terms. I tried to show that the following was true:

$\frac{x^3}{3!} \leq \frac{x^4}{4!}$

$4x^3 \leq x^4$

$4 \leq x$

which is not true for all $x$.

I tried taking the derivative and all that I could ascertain was that the the function became more and more increasing as $x \rightarrow \infty$ and became more and more decreasing as $x \rightarrow -\infty$, but I couldn't seem to prove that there were no roots to go with this property.

$\endgroup$
3
  • 1
    $\begingroup$ Use the result on the first two terms you have and see that $f^{'''}(x)>0$. Further when $f^{'''}(x)>0$, then $f^{''}(x)$ is always increasing, so is $f^{''}(x)$ and last but not least $f(x)$. $\endgroup$
    – draks ...
    Jun 9, 2012 at 16:16
  • $\begingroup$ Or use the solutions of the Quartic polynomial and check whether it has real roots. $\endgroup$
    – draks ...
    Jun 9, 2012 at 16:29
  • $\begingroup$ The argument of Andrew Salmon works if we truncate the series for $e^x$ at any even exponent, call that $f_{2n}(x).$ The trick is that $f_{k}' = f_{k-1},$ so $f_k(x) = f_{k-1}(x) + \frac{x^k}{k!}.$ Back to $k$ even, we get $$ f_{2n}(x) = f_{2n}'(x) + \frac{x^{2n}}{(2n)!}. $$ At any local minimum, the derivative is $0.$ As the minimum does not occur at $x=0,$ the minimum is positive. Some of this involves induction on $n.$ $\endgroup$
    – Will Jagy
    Jun 9, 2012 at 18:51

4 Answers 4

12
$\begingroup$

Hint: $$f(x) = \frac{1}{4} + \frac{(x + 3/2)^2}{3} +\frac{x^2(x+2)^2}{24}$$

$\endgroup$
3
  • $\begingroup$ Oh! Very nice and short. I'd +1 twice if I could. $\endgroup$
    – Gigili
    Jun 9, 2012 at 16:56
  • $\begingroup$ Completing the square solves everything that ever existed! $\endgroup$
    – stariz77
    Jun 9, 2012 at 17:13
  • 1
    $\begingroup$ @stariz77, yes, it is true for single-variable polynomials but not for several variables: en.wikipedia.org/wiki/Hilbert's_seventeenth_problem $\endgroup$
    – lhf
    Jun 9, 2012 at 19:03
3
$\begingroup$

$f$ is a polynomial, and therefore, is differentiable at all points. Furthermore, as $x\to\infty$ or $x\to-\infty$, $f(x)\to+\infty$. Thus, if $f(x)\le0$ for some $x$, then $f(x)\le0$ for some relative minimum.

$$f'(x)=1+x+\frac{x^2}{2!}+\frac{x^3}{3!}$$

$f'(x)=0$ for all relative minima. However, if $f'(x)=0$, then $$f(x)=f'(x)+\frac{x^4}{4!}$$

Thus, $f(x)=\frac{x^4}{4!}>0$ for all relative minima $x\not=0$. $x=0$ is not a relative minimum, because $f'(0)\not=0$, so this equation holds for all relative minima of $f$. This contradicts our assumption, so $f(x)>0$ for all $x\in \mathbb R$.

$\endgroup$
3
  • $\begingroup$ +1 Nice, but I'd edit the answer, pointing out simply: since $$f(x)=f'(x)+\frac{x^4}{4!}\,\,\forall\,x\in\mathbb{R}$$, if we have a minimum at $\,x_0\,$ then $$f(x_0)=f'(x_0)+\frac{x_0^2}{4!}=\frac{x_0^2}{4!}>0$$as clearly $\,x_0\neq 0\,$ , which contradicts our initial assumption that $\,f\leq 0\,$ at some minimum... $\endgroup$
    – DonAntonio
    Jun 9, 2012 at 17:01
  • 1
    $\begingroup$ @DonAntonio We can polish it up even more, the contradiction is not required, the entire answer could be: $x=0$ is not a minimum, so if $x_0$ is a minimum then $f(x_0) = f'(x_0) + x^4_0/4! > 0.$ So $f> 0.$ $\endgroup$ Jun 9, 2012 at 17:45
  • 1
    $\begingroup$ This argument shows that, if we truncate the series for the exponential ending with any even exponent, the resulting polynomial is always positive. And, by induction, positive second derivative and a single global minimum. $\endgroup$
    – Will Jagy
    Jun 9, 2012 at 18:20
1
$\begingroup$

You have had some good ideas so far. You tried to see when this was true: $$\frac{x^3}{3!} \leq \frac{x^4}{4!}.$$

You rearranged this to $4x^3\leq x^4$ but you made an incorrect conclusion when you divided by $x^3$ (if $x<0$ then the inequality sign should flip). Instead, lets divide by $x^2$ to get $4x \leq x^2$ or $x(x-4)\geq 0.$ This is true when $x\leq 0$ or $x\geq 4$ so the desired inequality is true in that range.

For $0< x < 4$ we don't have $\frac{x^3}{3!} \leq \frac{x^4}{4!}$ but lets see if the other terms can save us. To do this, we need to see exactly how large $g(x) = x^3/3! - x^4/4!$ can be in $(0,4).$ We calculate that $g'(x) = -(x-3)x^2/6$ so $g$ increases when $0\leq x\leq 3$, the maximum occurs at $g(3)=9/8$, and then it decreases after that.

This is good, because the $1+x+x^2/2$ terms obviously give at least $1$ from $x=0$, and will give us more as $x$ gets bigger. So we solve $1+x+x^2/2=9/8$ and we take the positive solution which is $\frac{\sqrt{5}-2}{2} \approx 0.118.$ So the inequality is definitely true for $x\geq 0.12$ because $g$ is at most $9/8$ and $1+x+x^2/2$ accounts for that amount in that range.

Remember that $g$ was increasing between $x=0$ to $x=3$, so the largest $g$ can be in the remaining range is $g(0.12) = 873/315000 <1$, which is less than the amount $1+x+x^2/2$ gives us. So the inequality is also true for $0\leq x\leq 0.12$, so overall, for all $x.$

So all in all, the only trouble was for $x$ in $(0,4)$ and the contribution from the other terms was always enough to account for $x^3/3!$ when $x^4/4!$ wasn't enough.

$\endgroup$
0
$\begingroup$

Observe that

$$ e^x = f(x) + \frac{x^5}{5!} + \cdots$$

Then show that for $x<0$ $$ f(x) > \frac{x^5}{5!} + \cdots$$

But $e^x > 0, \forall x\in R$

So, $$f(x) > e^x - f(x) \Rightarrow 2f(x)>e^x>0$$

$\endgroup$
7
  • $\begingroup$ Well, showing $\,\displaystyle{f(x)>\frac{x^5}{5!}}\,$ may prove to be tricky... $\endgroup$
    – DonAntonio
    Jun 9, 2012 at 16:45
  • $\begingroup$ Well, not so much (unless I'm mistaken) but I don't want to give away the answer. $\endgroup$
    – Eelvex
    Jun 9, 2012 at 16:49
  • $\begingroup$ Well, perhaps I'm wrong...I'll expect your posting on this after some time has ellapsed. $\endgroup$
    – DonAntonio
    Jun 9, 2012 at 16:52
  • $\begingroup$ $f(x)\gt\frac{x^5}{5!}\dots$ is incorrect. Think of a sufficiently large $x$. This would imply that a polynomial function can be strictly greater than an exponential one. $\endgroup$
    – A.S
    Jun 9, 2012 at 16:52
  • $\begingroup$ @DonAntonio Sure :) $\endgroup$
    – Eelvex
    Jun 9, 2012 at 16:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.