I am trying to prove the following equation: $$ \int_a^x\int_a^{t_1}\cdots\int_a^{t_{n-1}}\,dt_n\cdots\,dt_2\,dt_1=\frac{(x-a)^n}{n!}$$ I am not really sure where to begin. I would appreciate any help.
-
3Use induction on $n$. – Zardo Jun 12 '15 at 22:02
-
@zardo I'm guessing Op would appreciate more than a catchy one liner... ;) – Zach466920 Jun 12 '15 at 22:03
-
$$ \frac{(x-a)^{n+1}}{(n+1)!}=\frac{(x-a)^{n}(x-a)}{{n}!*(n+1)}$$ I kind of got stuck after that ... – Remuo Rissan Jun 12 '15 at 22:10
-
related: https://en.wikipedia.org/wiki/Cauchy_formula_for_repeated_integration – Vincenzo Tibullo Jun 12 '15 at 22:11
3 Answers
Let $I_n(x)$ be the integral above. It is straightforward to verify the equation for $n=1$.
Note that we have $I_{n+1}'(x) = I_n(x)$.
Suppose the equation is true for $1,...,n$.
Then integrating the equation above gives $I_{n+1}(x) = \int_a^x I_{n+1}'(t)dt = \int_a^x I_{n}(t)dt= \int_a^x {(t-a)^n \over n! }dt $, and evaluating the integral yields the desired result.
- 178,207
-
Thank you, I didn't see your answer until now, but it makes the most sense. – Remuo Rissan Jun 12 '15 at 22:52
@Zardo suggested mathematical induction on $n$. Let's try a crude form of that and see if you can take it from there. $$ \int_a^{t_{n-1}} \, dt_n = t_{n-1} - a. $$ \begin{align} \int_a^{t_{n-2}} \int_a^{t_{n-1}} \, dt_n\,dt_{n-1} & = \int_a^{t_{n-2}} (t_{n-1} - a)\,dt_{n-1} \\[8pt] & = \left.\vphantom{\frac11} \frac{(t_{n-1} - a)^2} 2 \right|_{t_{n-1}:=a}^{t_{n-1}:=t_{n-2}} = \frac{(t_{n-2}-a)^2} 2 \end{align}
\begin{align} \int_a^{t_{n-3}} \int_a^{t_{n-2}} \int_a^{t_{n-1}} \, dt_n\,dt_{n-1} \, dt_{n-2} & = \int_a^{t_{n-3}} \frac{(t_{n-2}-a)^2} 2\,dt_{n-1} \\[8pt] & = \left.\vphantom{\frac11} \frac{(t_{n-2} - a)^3} 6 \right|_{t_{n-2}:=a}^{t_{n-2}:=t_{n-3}} = \frac{(t_{n-3}-a)^3} 6 \end{align}
$\ldots$ and so on. That is how mathematical induction is done, but it's stated somewhat more abstractly.
Let's make it a little bit more abstract:
Induction hypothesis: $$ \int_a^{t_{n-(k-1)}} \int_a^{t_{n-(k-2)}} \cdots \int_1^{t_{n-1}} dt_n \, dt_{n-1} \cdots dt_{n-(k-2)} = \frac{(t_{n-(k-1)}-a)^{n-(k-1)}}{(n-(k-1))!}. $$
Induction step: \begin{align} & \int_a^{t_{n-k}} \int_a^{t_{n-(k-1)}} \int_a^{t_{n-(k-2)}} \cdots \int_a^{t_{n-1}} dt_n \, dt_{n-1} \cdots dt_{n-(k-2)} \, dt_{n-(k-1)} \\[10pt] = {} & \int_a^{t_{n-k}} \frac{(t_{n-(k-1)}-a)^{n-(k-1)}}{(n-(k-1))!} \, dt_{n-(k-1)} \\[10pt] = {} & \left. \frac{(t_{n-(k-1)} - a)^{n-k}}{(n-k)!} \right|_{t_{n-(k-1)}\,:=\,a}^{t_{n-(k-1)}\,:=\,t_{n-k}} \\[10pt] = {} & \frac{(t_{n-k} -a)^{n-k}}{(n-k)!}. \end{align} This is the same argument that was presented more concretely above.
-
Like my comment on Dr. MV's answer, I knew the pattern beforehand, but I am not really sure how to relate the pattern to induction. – Remuo Rissan Jun 12 '15 at 22:43
-
Patterns of just this kind are what induction is all about. Maybe I'll add more on that to this answer. – Michael Hardy Jun 13 '15 at 17:25
-
@RomeoDissan : OK, I've added to the answer. But if you want to understand mathematical induction, you should learn why the difference between the concrete version I posted originally and the abstract version I posted later is only that the abstract version says that the pattern will persist in all cases, and that should be clear already from an understanding of the concrete version. So if you don't understand that, then that is what you need to work on. ${}\qquad{}$ – Michael Hardy Jun 13 '15 at 18:25
HINT:
$$\begin{align} \int_a^{t_{n-1}}dt_n&=\frac{1}{1}(t_{n-1}-a)^1\\\\ \int_a^{t_{n-2}}(t_{n-1}-a)dt&=\frac{1}{2\cdot 1}(t_{n-2}-a)^2\\\\ \int_a^{t_{n-3}}\frac12 (t_{n-2}-a)^2dt&=\frac{1}{3\cdot 2\cdot 1}(t_{n-3}-a)^3 \end{align}$$
HINT 2:
We have established a benchmark that the result is true if $n=3$ ($x=t_{n-3}=t_0$). Now assume that the relationship is true for some number $k$. Show that it is true for $k+1$ to complete the proof by induction.
- 184,670
-
I already noticed the pattern before posting, but I'm trying to make the proof as rigorous as possible, and a pattern isn't that rigorous. Do you think you can give another hint? – Remuo Rissan Jun 12 '15 at 22:31
-
@romeodissan After one notices a pattern, one can then test to see if it is general. Here, the natural test is unductive. So, proceed with a quite simple proof by induction. – Mark Viola Jun 12 '15 at 23:05
-
Most proofs by induction require a little bit of algebraic manipulation, and as you can see in the comment under my question, I actually got stuck. – Remuo Rissan Jun 13 '15 at 00:52
-
@romeodissan integrate a $k+1$ time on the $k$th term. You will get the desired result. – Mark Viola Jun 13 '15 at 02:28