$\int_0^{100}\frac{e^{-x}}{x+100}dx>0.005$? My attempt: $$\int_0^{100}\frac{e^{-x}}{x+100}dx>\int_0^{100}\frac{e^{-x}}{200}dx=\frac{1-e^{-100}}{200}$$ A little bit error. How to amend it?
-
Where is the error? $e^{-100} \approx 0$ so ${1-e^{-100}\over 200}\approx \frac 1{200}=0.05$ – For the love of maths May 31 '19 at 11:39
-
@MohammadZuhairKhan But when we use this inequality, we could not assure $\frac{1-e^{-100}}{200}>0.005$. – xldd May 31 '19 at 11:42
-
Ah, yes. It was a good attempt though. I will try to find another way. In reality, ${1-e^{-100} \over 200} < 0.005$ – For the love of maths May 31 '19 at 11:44
3 Answers
The idea: The integrand decreases rapidly on the given interval, so the idea is to estimate the integral from below by integrating over a shorter interval $[0, a]$, and then continue with your approach, but with a better bound for the denominator:
For $0 < a < 100$ we can estimate $$I = \int_0^{100}\frac{e^{-x}}{x+100}dx \ge \int_0^a\frac{e^{-x}}{x+100}dx \\ \ge \int_0^a\frac{e^{-x}}{a+100}dx = \frac{1-e^{-a}}{a+100} \, . $$ For $a=4$ this gives $$ I \ge \frac{1-e^{-4}}{104} \approx 0.00943927270299294 $$ which comes fairly close to the result $I \approx 0.009901942286733037$ (obtained by numeric integration with Maxima).
We can also avoid calculating $e^{-a}$ numerically and use $e^a \ge 1+a$ to further estimate $$ I \ge \frac{1-e^{-a}}{a+100} \ge \frac{a}{(a+1)(a+100)} \, . $$ For $a=10$ this gives $$ I \ge \frac{1}{121} \approx 0.008264462809917356 \,. $$
- 113,040
Use $e^u\ge 1+u$ to conclude $$ \frac1{1+u}\ge e^{-u}\implies \frac1{100+x}\ge\frac1{100}e^{-x/100}. $$ Then, again using $1-e^{-u}\ge1-\frac1{1+u}$, $$ \int_0^{100}\frac{e^{-x}}{100+x}dx \ge\frac1{100} \int_0^{100} e^{-\frac{101}{100}x} dx =\frac1{100} \frac{100}{101} \left( 1-e^{-101} \right) \ge\frac1{101}\left(1-\frac1{102}\right)=\frac1{102} $$ which is decidedly larger than $\frac1{200}$.
- 126,666
-
That is an excellent estimate, $1/102 \approx 0.0098$ comes very close to the actual value. – Martin R May 31 '19 at 13:22
Using the MatDeck function for numerical integration (Gaussian rule of integration in this case) you can get approximate value of 0.00990194
MatDeck numerical integration - Gaussian rule of integration
- 1