4

A particle moves in a straight line starting from rest and finishing at rest, and covers unit distance in unit time. Prove that at some point its acceleration has a magnitude of at least 4 units. It is assumed that $v(t)=\frac{dx}{dt}$ and $a(t)=\frac{dv}{dt}$ are continuous functions of time.

The proof given is the following:
If we plot $v$ against $t$, the area under the curve must be the same (1 square unit) as that of an isosceles triangle having the same base and an altitude of 2. graph

The slopes of the sides of the triangle are $\pm 4$. Part of the $v-t$ curve must fall outside the triangle or coincide with its sides. Thus the slope, $a$, of the curve is $\geq 4$ at some point.

I was looking for a pure analytic proof which doesn't rely on the graphical representation and intuition.

user5402
  • 3,092
  • 19
  • 35

2 Answers2

3

Here's a more general result:

Let $f:[a, b] \to \mathbb{R}$ is differentiable, not constant and $f(a) =f(b) =0.$ Show that there exists $c \in (a, b)$ such that $|f'(c)| > \dfrac{4}{(b-a)^2}\displaystyle \int_{a}^{b}f(x) dx$

Proof: The result is immediate if $f'$ is not bounded on $[a, b].$ If not then $|f'|([a, b])$ is non-empty and bounded above and hence has supremum $M >0.$ Then for all $x \in [a, b], $ $|f'(x)| \leq M.$ If $x \in [a, \frac{a+b}{2}]$ then by the mean value theorem applied to $f$ on $[a, x]$ there exists $c_x \in (a, x)$ such that $\dfrac{f(x) -f(a)}{x-a} =f'(c_x)$ and since $f(a) =0$ and $f'$ is bounded we get $|f(x)|\leq M|x-a| = M(x-a)$

Similarly for $x \in [\frac{a+b}{2}, b]$ we obtain $|f(x)|\leq M|x-b| = M(b-x)$ Note that both the above relations cannot be equalities simultaneously or else the right hand derivative of $f$ at $\frac{a+b}{2}$ would be $-M$ and the left hand derivative would be $M,$ a contradiction. The rest is straightforward:

$\begin{align} \displaystyle \int_{a}^{b}f(x) dx &= \int_{a}^{\frac{a+b}{2}}f(x) dx+ \int_{\frac{a+b}{2}}^{b}f(x)dx \\&< \int_{a}^{\frac{a+b}{2}}(x-a)dx + \int_{\frac{a+b}{2}}^{b}(b-x)dx\\&= M\dfrac{(b-a)^2}{4} \end{align}$

$\Rightarrow \dfrac{4}{(b-a)^2} \displaystyle\int_{a}^{b}f(x)dx <M$

Since $M = \sup\{|f'(x)|: x\in [a,b]\}, $ corresponding to $\varepsilon = M - \dfrac{4}{(b-a)^2} \displaystyle\int_{a}^{b}f(x)dx >0$ there exists $c \in [a , b]$ such that $|f'(c)| > M-\varepsilon = \dfrac{4}{(b-a)^2}\displaystyle\int_{a}^{b}f(x)dx, $ thus completing the proof.

For the problem in question, let $f(t)$ denote the velocity function as a function of time so that $f'(t)$ denotes acceleration as a function of time. Given: $f(0)=f(1)= 0$ and $\displaystyle \int_{0}^{1}f(x)dx =1.$ Then from the result above there exists $c \in [0, 1]$ such that $|f'(c)|>4.$

Note that for equality we would require the velocity function to not be differentiable at $0.5$, a case shown in the diagram posted by the OP which is a physical impossibility.

Aryaman Jal
  • 1,595
  • Great answer! What is the source of the theorem? I mean where did you find such a result? – vidyarthi Dec 06 '17 at 18:40
  • Thanks. If I remember right, it was problem 121 in the continuous functions chapter in Polya and Szego's book called Problems and Theorems in Analysis – Aryaman Jal Dec 06 '17 at 19:08
  • ok, that book has lots of interesting elementary and classic problems. By the way, are you a research scholar or so in Mathematics in TIFR or IMSc? – vidyarthi Dec 06 '17 at 19:11
3

Let $v(t)$, $a(t)$ be the velocity and acceleration at time $t\in [0,1]$. We express distance as the integral of velocity and integrate by parts, giving $$ \begin{eqnarray*} 1&=&\int_0^1 v(t)\,dt\\ &=&-\int_0^1 a(t)\left(t-\frac{1}{2}\right)\,dt. \end{eqnarray*} $$ Suppose $|a(t)|<4$ for all $t$. Then $$ \begin{eqnarray*} -\int_0^1 a(t)\left(t-\frac{1}{2}\right)\,dt&\leq&\int_0^1 |a(t)|\left|\left(t-\frac{1}{2}\right)\right|\,dt\\ &<&\int_0^{\frac{1}{2}}4\left(\frac{1}{2}-t\right)\,dt+\int_{\frac{1}{2}}^{1}4\left(t-\frac{1}{2}\right)\\ &=&\frac{1}{2}+\frac{1}{2}=1. \end{eqnarray*} $$

Julian Rosen
  • 16,600
  • Sorry would you mind spelling out the integration by parts in a bit more detail? I assume it's $\int_0^1 v(t)\times 1 ,\rm{dt}$, but I end up with only $-\int_0^1 a(t)t,\rm{dt}$. – TooTone Jul 21 '13 at 21:19
  • I chose $t-\frac{1}{2}$ as my anti-derivative of $dt$ (instead of just $t$) – Julian Rosen Jul 21 '13 at 21:26
  • cool! I hadn't seen that done before but it works of course! – TooTone Jul 21 '13 at 21:32