Evaluation of $\displaystyle \int^{2}_{1}\frac{1}{x}dx$ using limit as a sum

Try: Using The formula $$\int^{b}_{a}f(x)dx = \lim_{h\rightarrow 0}h\times \sum^{n-1}_{r=1}f(a+rh)$$

where $nh=b-a$

above $a=1,b=2$ and $\displaystyle f(x)=\frac{1}{x}$ and $nh=1$

$$\int^{2}_{1}\frac{1}{x}dx = \lim_{h\rightarrow 0} \sum^{n-1}_{r=1}f(a+rh)=\lim_{h\rightarrow 0}h\cdot \sum^{n-1}_{r=0}f(1+rh) $$

$$\lim_{h\rightarrow 0}h\cdot \sum^{n-1}_{r=0}\frac{1}{1+rh}=\lim_{h\rightarrow 0}\bigg[\frac{h}{1+h}+\frac{h}{1+2h}+\cdots \cdots +\frac{h}{1+(r-1)h}\bigg]$$ i did not know how i proceed, struck here

could some help me. thanks

  • 1
    Well the answer is $\log 2$ and you need to know something about $\log 2$ before you can actually deal with the problem. – Paramanand Singh Oct 12 at 6:48
  • 1
    If you accept that $\log 2=\lim_{n\to\infty}n(2^{1/n}-1)$ then you need to use a partition of $[1,2]$ via points $x_k=2^{k/n}$. If you accept that $\log 2=\sum_{i=1}^{\infty} (-1)^{i+1}(1/i)$ then your approach can be used to show that the integral is $\log 2$. – Paramanand Singh Oct 12 at 6:51
  • In particular see method 3 of this answer math.stackexchange.com/a/155248/72031 which is helpful for your approach. – Paramanand Singh Oct 12 at 7:03
  • I did not understand the line $x_{k}=2^{\frac{k}{n}}$ for $[1,2].$ thanks parmanand singh, Can you please explain me in detail. thanks – D Tiwari Oct 12 at 7:08
up vote 5 down vote accepted

Consider the points $x=c^k$ with $c^n=2$.

$$\int_1^2\frac{dx}{x}\approx\sum_{k=0}^{n+1} \frac{\Delta c^k}{c^k}=\sum_{k=1}^n \frac{ c^{k+1}-c^k}{c^k}=n(c-1)=n\left(\sqrt[n]2-1\right).$$

Then,

$$\lim_{n\to\infty}n\left(\sqrt[n]2-1\right)=\lim_{h\to0}\frac{2^h-1}h=\left.(2^h)'\right|_{h=0}=\log2.$$


Note that this is in fact a discrete version of an exponential change of variable, $x=e^t$, giving

$$\int_1^2\frac{dx}x=\int_{\log1}^{\log2}\frac{e^t\,dt}{e^t}=\int_0^{\log2}dt.$$

The latter integral can be trivially computed as a sum.

$$\int_1^2\frac{dx}{x}\approx\frac1n\sum_{k=n+1}^{2n}\left(\dfrac kn\right)^{-1}=\sum_{k={n+1}}^{2n}\frac1{k}=H_{2n}-H_{n}$$ where $H_n$ denotes an harmonic number.

Now,

$$\lim_{n\to\infty}(H_{2n}-H_n)=\lim_{n\to\infty}(\log2n+\gamma+o(1)-\log n-\gamma+o(1))=\log2.$$


It is questionable whether this approach makes sense, as deriving the asymptotic expression of the harmonic numbers is much harder than the initial problem.

  • 1
    +1 especially for the remark at the end. – Paramanand Singh Oct 12 at 8:28

Let's assume the following equation as given $$\log 2=\lim_{n\to\infty} n(2^{1/n}-1)\tag{1}$$ By definition of Riemann integral if $f:[a, b] \to\mathbb {R} $ is Riemann integrable on $[a, b] $ with integral $I$ then for every $\epsilon >0$ there is a $\delta>0$ such that for all partitions $P$ of $[a, b] $ with norm less than $\delta$ we have $|S(f, P) - I|<\epsilon $ and we write $\lim_{||P||\to 0}S(f,P)=I$.

To explain notation and terms we say that a set $P=\{x_0,x_1,\dots, x_n\} $ is a partition of interval $[a, b] $ if $$a=x_0<x_1<\dots<x_n=b$$ The norm of $P$ (denoted by $||P||$) is defined as $\max_{i=1}^{n}(x_i-x_{i-1})$. A Riemann sum for $f$ over a partition $P$ of $[a, b] $ (denoted by $S(f, P) $) is a sum of the form $$\sum_{i=1}^{n}f(t_i)(x_i-x_{i-1})$$ where points $t_i$ called tags are arbitrary points of the intervals $[x_{i-1},x_i]$ respectively.

Now it is well known that function $f$ defined by $f(x) =1/x$ is Riemann integrable on any interval $[a, b] $ if $0\notin[a,b]$ (because of continuity of $f$ in that interval) and in particular the Riemann integral $$I=\int_{1}^{2}\frac{dx}{x}$$ exists.

Let's choose partition $P$ of $[1,2]$ using points $x_k=2^{k/n}$. Then norm of $P$ is $\max_{k=1}^{n}(2^{k/n}-2^{(k-1)/n})$ and this tends to $0$ if $n\to \infty $. Also let's choose tags $t_k$ as $t_k=x_{k-1}=2^{(k-1)/n}$. Then the Riemann sum $$S(f, P) =\sum_{k=1}^{n}f(t_k)(x_k-x_{k-1})=\sum_{k=1}^{n}\frac{2^{k/n}-2^{(k-1)/n}}{2^{(k-1)/n}}=n(2^{1/n}-1)$$ and by our starting equation this tends to $\log 2$ as $n\to\infty$. Hence the value of the integral is $\log 2$.

Note: This is an expansion of my comment to the question and another answer based on same idea has been given by another user. I have tried here to give some more detail about Riemann integration and in particular highlighted the arbitrary nature of points of a partition. Typical introductory calculus texts almost exclusively use partition points in arithmetic progression and this answer in contrast uses points in geometric progression for creating a partition. The same technique can be used to evaluate the integral of $f(x) =x^p$ for any $p$.

  • 1
    Expanded version of mine, it seems ;-) – Yves Daoust Oct 12 at 8:16
  • 1
    @YvesDaoust: see updated note at the end. – Paramanand Singh Oct 12 at 8:20

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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