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.

I need assistance in plotting the graph of $\frac{1}{\ln x}$. wolframalpha gives this.

How to plot this function (both real and imaginary part) using calculus?

share|improve this question
add comment

3 Answers

Using this software you should get following graph for $\frac{1}{\ln x}$ ,when $x>0$

enter image description here

share|improve this answer
add comment

I don't know what alpha is doing for negative values of $x$. Usually, we take $\log x$ to be undefined for $x\le0$ - certainly, this is what is generally done at the level of introductory Calculus classes. For $x\gt0$, $\log x$ is real, so the question of plotting the imaginary part does not arise - there is no imaginary part. To plot the "real part", imagine the graph of $y=\log x$ - every student should have a picture of this graph in her head - and take the reciprocal. So it approaches 0 through the negative numbers as $x$ approaches zero from the right; it has $x=1$ as an asymptote; it goes to zero as $x\to\infty$; and so on.

share|improve this answer
add comment

The results from WolframAlpha do make sense (EDIT: though admittedly they only consider a limited case, see the edit below). To start at the beginning:

You know that, for some $a$, we define a number $b = ln(a)$ to be the number which solves this equation:

$$e^b = a$$

For positive $a$, it's not too difficult to see that this makes sense, when $a = e$, $b = 1$; when $a = 6$, $b \approx 1.792$. Cool

But what about $a = -1$? There's no real number $b$ such that $e^b = -1$. However, if we suppose $b$ is complex, and in particular

$$b = x + yi$$

Then we can solve this equation. By Euler's formula:

$$e^{(x+yi)} = e^x \times e^y = e^x (\cos y + i \sin y)$$

So our problem now is to find x and y such that:

$$a = e^x (\cos y + i \sin y)$$

When $a$ is positive, this is easy. We set $y = 0$, and the problem reduces to solving $a = e^x$, which we can do in the positive case.

When $a$ is negative, what we can do is set $y = \pi$. In this case, $\sin \pi = 0$, but $\cos \pi = -1$, so we get the equation:

$$a = -e^x$$

which is solvable as a is negative.

Combining what we know about x and y, we get a formula for $ln(a)$ when a is negative:

$$ln(a) = x + y i = ln(-a) + \pi i$$

Then, for $1/ln(a)$, you just need to re-arrange the expression above:

$$\frac{1}{ln(a)} = \frac{1}{ln(-a)+\pi i} = \frac{ln(-a) - \pi}{ln(-a)^2+\pi^2} = \frac{ln(-a)}{ln(-a)^2+\pi^2} - \frac{\pi}{ln(-a)^2+\pi^2} i$$

I stress again, this only holds when a is negative. Plotting the real and imaginary parts of this expression in WolframAlpha gives the same curves as it originally plotted whenever a is negative (c.f. here and here). For positive a, it's just a normal, real, $ln(a)$ graph.

EDIT: While Wolfram's answer does indeed 'make sense', it was correctly brought up in the comments that there are many other solutions to this problem. When we let $y = \pi$, we found that $e^{x+y i}$ was a negative real number. However, we could just have easily let $y = 3 \pi$ or in general $y = (2k+1)\pi$ for any integer $k$. These values will all lead to legitimate curves, and in general we will have: $$\frac{1}{ln(a)}= \frac{ln(-a)}{ln(-a)^2+(2k+1)^2\pi^2} - \frac{(2k+1)\pi}{ln(-a)^2+(2k+1)^2\pi^2} i$$

Which will give different real and imaginary curves for different integer values of k.

To get a range of these curves, you could use any CAS or graphing software capable of plotting multiple functions on the same axes.

share|improve this answer
You have given one formula for $\log a$ when $a\lt0$, but of course there are many, all differing by multiples of $2\pi i$, and no one of them any better than any other. You got the one that alpha uses, but that just means that alpha is sweeping the problem under the rug. –  Gerry Myerson Oct 31 '11 at 11:33
Thanks, you're right of course. Probably a reason I shouldn't be doing this late at night :) I hope the edit doesn't introduce any new silly errors :) –  tom Oct 31 '11 at 12:10
add comment

Your Answer


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.