Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

How do I get $$ \int_a^b \frac{1}{x}dx = \ln\left(\frac{b}{a}\right)$$ as a limit of sum. The constant width partition of the interval $(a,b)$ doesn't seem to work.

share|improve this question
    
What makes you think the fixed-width Riemann sum doesn't work in the limit? –  anon Jun 22 '12 at 14:14
2  
Typically the logarithm is defined $\log t:= \int_1^t\frac{1}{x}dx$, so your question follows nearly by definition. –  nullUser Jun 22 '12 at 14:15
    
@Anon I tried it but couldn't get to the answer. –  Prakash Gautam Jun 22 '12 at 14:25
    
@PrakashGautam Write up what you tried below your question. –  rschwieb Jun 22 '12 at 14:29

4 Answers 4

Take $x_i = a \left( \frac{b}{a} \right)^{i/n}$ then:

$$\lim\limits_{n \to \infty} \sum_{i = 0}^{n-1} \frac{x_{i+1} - x_i}{x_i} = \lim\limits_{n \to \infty} \frac{ \left(\frac{b}{a}\right)^{\frac{1}{n}} - 1 }{\frac{1}{n}} = \ln \frac{b}{a}$$

share|improve this answer
    
(+1), but I guess it is good to append that $$\log x = \lim\limits_{k \to 0} \frac{x^k-1}{k}$$ –  Pedro Tamaroff Jun 24 '12 at 22:43

$$\int _a^b {f(t) dt}=\lim_{n\to\infty} \frac{(b-a)}{n}\sum \limits_{k=1}^n f(\frac{k(b-a)}{n}+a)$$

$$\int _a^b \frac{1}{t} dt=\lim_{n\to\infty} \frac{(b-a)}{n}\sum \limits_{k=1}^n \frac{n}{k(b-a)+na}=\lim_{n\to\infty} \frac{(b-a)}{1}\sum \limits_{k=1}^n \frac{1}{k(b-a)+na}$$

$$\int _a^b \frac{1}{t} dt=\lim_{n\to\infty} \frac{(b-a)}{n}\sum \limits_{k=1}^n \frac{n}{k(b-a)+na}=\lim_{n\to\infty} \frac{(b-a)}{(b-a)}\sum \limits_{k=1}^n \frac{1}{k+\frac{na}{b-a}}$$

$$\int _a^b \frac{1}{t} dt=\lim_{n\to\infty} \sum \limits_{k=1}^n \frac{1}{k+\frac{n}{(b/a)-1}}$$

$b/a=x$

$$f(x-1)=\int _a^b \frac{1}{t} dt=\lim_{n\to\infty} \sum \limits_{k=1}^n \frac{1}{k+\frac{n}{x-1}}$$

$$f(x)=\lim_{n\to\infty} \sum \limits_{k=1}^n \frac{1}{k+\frac{n}{x}}$$

$$f(x)=\lim_{n\to\infty}\frac{x}{n} \sum \limits_{k=1}^n \frac{1}{1+\frac{kx}{n}}=\lim_{n\to\infty} \frac{x}{n} \sum \limits_{k=1}^n (1-\frac{kx}{n}+\frac{k^2x^2}{n^2}-\frac{k^3x^3}{n^3}+....)=\lim_{n\to\infty} \frac{x}{n} \sum \limits_{k=1}^n 1 -\lim_{n\to\infty} \frac{x^2}{n^2} \sum \limits_{k=1}^n k+\lim_{n\to\infty} \frac{x^3}{n^3} \sum \limits_{k=1}^n k^2-\lim_{n\to\infty} \frac{x^4}{n^4} \sum \limits_{k=1}^n k^3+.....$$

$$f(x)=\lim_{n\to\infty} \frac{x}{n} \sum \limits_{k=1}^n 1 -\lim_{n\to\infty} \frac{x^2}{n^2} \sum \limits_{k=1}^n k+\lim_{n\to\infty} \frac{x^3}{n^3} \sum \limits_{k=1}^n k^2-\lim_{n\to\infty} \frac{x^4}{n^4} \sum \limits_{k=1}^n k^3+.....=\lim_{n\to\infty} \frac{x}{n} n -\lim_{n\to\infty} \frac{x^2}{n^2} (\frac{n^2}{2}+\frac{n}{2})+\lim_{n\to\infty} \frac{x^3}{n^3} (\frac{n^3}{3}+\frac{n^2}{2}+\frac{n}{6})-\lim_{n\to\infty} \frac{x^4}{n^4}(\frac{n^4}{4}+\frac{n^3}{2}+\frac{n^2}{4})+....$$


$$\sum \limits_{k=1}^{n} k^m=\frac{n^{m+1}}{m+1}+a_mn^m+....+a_1n=\frac{n^{m+1}}{m+1}+\sum \limits_{j=1}^m a_jn^j$$ where $a_j$ are constants.where aj are constants. More information about summation http://en.wikipedia.org/wiki/Summation


After solving limits. We get:

$$f(x)=\frac{x}{1} -\frac{x^2}{2} + \frac{x^3}{3} - \frac{x^4}{4}+ ....=\sum \limits_{k=1}^{\infty} (-1)^{k+1} \frac{x^k}{k}=\ln(x+1)$$

$$f(x-1)=\ln(x)$$

$b/a=x$

$$\int _a^b \frac{1}{t} dt=f(x-1)=\ln(x)=\ln(b/a)$$

share|improve this answer

The constant width partition works fine. No need for fancy partitions. To wit:

Let $x_k = a+\frac{k}{N}(b-a)$, with $k = 0,...,N$. We have, of course, $x_0=a$, $x_N = b$.

Using the Taylor series expansion of $x \mapsto \ln x$, and the mean value theorem, we obtain the bounds, for $x \geq y$:

$$ \ln y + \frac{x-y}{y} - \frac{(x-y)^2}{2 y^2} \leq \ln x \leq \ln y + \frac{x-y}{y}.$$ Letting $y=x_k$, $x=x_{k+1}$, we get: $$\ln x_k + \frac{1}{N x_k} - \frac{1}{2 N^2 x_k^2} \leq \ln x_{k+1} \leq \ln x_k + \frac{1}{N x_k}.$$

Rearranging gives: $$- \frac{1}{2 N^2 x_k^2} \leq \ln x_{k+1} - \ln x_k - \frac{1}{N x_k} \leq 0.$$ Now sum the inequality over $k = 0,...,N-1$, to get: $$- \frac{1}{2 N^2} \sum_{k=0}^{N-1}\frac{1}{x_k^2} \leq \ln \frac{b}{a} - \frac{1}{N}\sum_{k=0}^{N-1} \frac{1}{x_k} \leq 0.$$

Since $x_k\geq a$, we obtain: $$- \frac{1}{2 N a^2} \leq \ln \frac{b}{a} - \frac{1}{N}\sum_{k=0}^{N-1} \frac{1}{x_k} \leq 0.$$

Now let $N \to \infty$, to obtain the desired result, $$ \lim_{N \to \infty} \frac{1}{N}\sum_{k=0}^{N-1} \frac{1}{x_k} = \int_a^b \frac{1}{t} dt = \ln \frac{b}{a}.$$

share|improve this answer
    
Well, fine is obviously not the same as "nicely". –  Pedro Tamaroff Jun 24 '12 at 22:42
    
No symbols where none intended. –  copper.hat Jun 24 '12 at 23:34
    
I don't follow. –  Pedro Tamaroff Jun 24 '12 at 23:36
    
Unlike mathematics, it loses meaning when explained; it is the last line of Beckett's idiosyncratic novel Watt. Seemed an appropriate response to your enigmatic comment... –  copper.hat Jun 24 '12 at 23:47
    
What I mean is that if one chooses an appropriate "fancy" partition the solution is much manageable (i.e. qoqozs solution or Amihai's), which I guess you're refering to. –  Pedro Tamaroff Jun 25 '12 at 0:09

Lets do it for $\int\limits_1^b\frac{1}{x}dx$:

Consider the following partition of our interval $\,[1,b]$:

$$P_n=\{1, \sqrt[n]{b},\sqrt[n]{b^2},\dots,\sqrt[n]{b^n}=b\}$$

Now, we need to calculate the upper and lower Darboux sums in respect to the $P_n$'s. Notice that $\frac{1}{x}$ is monotonic decreasing on the interval so the $\sup$ and the $\inf$ are obtained on the edges:

$$\mathcal{U}\left(\frac{1}{x},P_n\right)=\sum_{k=1}^n b^{-\frac{k}{n}}(b^{\frac{k}{n}}-b^{\frac{k-1}{n}})=\sum_{k=1}^n 1-b^{-\frac{1}{n}}=n(\sqrt[n]{b}-1)$$

define $g(x)=b^x$:

$$n(\sqrt[n]{b}-1)=\frac{g(\frac{1}{n})-g(0)}{\frac{1}{n}-0}\to_{n\to\infty}g'(0)=b^x\ln b|_{x=0}=\ln b$$

I'll leave it to you to figure out the lower sums as it's practically the same procedure.

share|improve this answer
    
Well, how do you define $\ln b$? This seems circular. THe derivative of $b^x$ is defined using that of $e^x$, which might or might not be defined in terms of the $\log x$. –  Pedro Tamaroff Jun 22 '12 at 16:05
    
Why can't I define the logarithm as the inverse of the $\exp$? –  Amihai Zivan Jun 23 '12 at 17:36
    
I said "might or might not" in the sense you can choose that method, but there are others available. –  Pedro Tamaroff Jun 24 '12 at 0:06

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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