In the integration formula $\int dx/x = log x + c$, Is the log natural or log base 10? The formula appears in many problems and i just got a problem wrong for apparently using the wrong log. Could you please enlighten me about the right log to be used in integration.
-
3That log is the natural log. No other choice. – Parcly Taxel Nov 03 '21 at 15:03
-
2Some take that as the definition of $\ln x$ – J. W. Tanner Nov 03 '21 at 15:09
-
2Ten is a nice number, but seriously, it's not that special. – PM 2Ring Nov 03 '21 at 16:07
3 Answers
Consider the differential equation $y'=y$ with $y(0)=1$ on $\mathbb{R}$. It can be proven that this has a unique solution. We define the exponential function on $\mathbb{R}$ to be equal to this solution. We denote it as $\exp$, but many people abuse notation and denote it $e^x$, which in this context, is fine, as long as everyone knows what is being meant. Now, it can be proven that $\exp$ is function whose range is $\mathbb{R}^+$. In other words, $\exp[\mathbb{R}]=\mathbb{R}^+$. It can also be proven that $\exp$ is a function that has a compositional inverse. This compositional inverse has a name: the natural logarithm, and we denote it $\ln$, or $\log$. Of course, one must remember that $\ln[\mathbb{R}^+]=\mathbb{R}$. The relationship that this function has with other logarithmic functions is complicated, but in some sense, it is the logarithmic function, due to the fact that it can be defined independently of any notion of powers. Hence the name "natural" logarithm. In terms of other logarithmic functions, the natural logarithm can be interpreted as being base $e$, where $e=\lim_{n\to\infty}\left(1+\frac{1}{n}\right)^n$.
Now, given the equation $\exp'=\exp$, it can be proven via the chain rule that $\ln'(x)=\frac{1}{x}$ for every $x\in\mathbb{R}^+$. This is because $(\exp\circ\ln)'(x)=1$, yet by the chain rule, $(\exp\circ\ln)'(x)=(\exp'\circ\ln)(x)\ln'(x)=(\exp\circ\ln)(x)\ln'(x)=x\ln'(x)$, so $x\ln'(x)=1$, implying $\ln'(x)=\frac{1}{x}$. Because of this, $\ln(x)$ is an antiderivative of $\frac{1}{x}$ on $\mathbb{R}^+$.
However, this is not the complete story. Consider the function $\ln^-$ defined by $(\ln^-)(x)=\ln(-x)$ for every $\mathbb{R}^-$. It turns out that $(\ln^-)'(x)=\frac{1}{x}$ is true on $\mathbb{R}^-$ as well! So to talk about the antiderivatives of $\frac{1}{x}$ properly, we need to consider functions $f_{A,B}:\mathbb{R}\setminus\{0\}\rightarrow\mathbb{R}$ such that $$f_{A,B}(x)=\begin{cases} \ln(-x)+A & x\lt0 \\ \ln(x)+B & x\gt0 \end{cases}.$$ These functions satisfy the property that $f_{A,B}'(x)=\frac{1}{x}$ for every $x\in\mathbb{R}\setminus\{0\}$. And in fact, this is every function that satisfies the equation on that domain, there are no other functions. So you can say $$\int\,\frac{\mathrm{d}x}{x}=f_{A,B}(x).$$ If you want to limit yourself to $x\gt0$, though, then $$\int\,\frac{\mathrm{d}x}{x}=\ln(x)+B.$$ Hopefully, this covers everything.
- 3,774
- 12
- 14
-
Unless one really wants to solve some ODEs with a certain "maximum" discussion, one rarely considers the natural domain as this detailed answer above does; instead, some open connected subset of the natural domain is often used. This is particularly so in "exercises" of Calculus. For instance, one writes $$\int\frac{2x-3}{(x-1)(x-2)};dx=\int \frac{1}{x-1}+\frac{1}{x-2};dx=\log|x-1|+\log|x-2|+C$$ instead of specifying three different constants for antiderivatives in the natural domain $\mathbb{R}\setminus{1,2}$. – Nov 08 '21 at 13:30
-
@LouisPan I am interested in providing an accurate, enlightening answer that will prevent further confusion and doubts down the line, not a slightly inaccurate context-dependent oversimplification that is bound to eventually create confusion and require an answer of this caliber anyway. Besides, your objection is not accurate either: in a calculus classroom, the domain of a function is almost never specified, and they certainly never used terminology such as "open subset of the natural domain" in such classrooms. – Angel Nov 08 '21 at 13:34
-
I do not have any "objection" here and I am not criticizing your answer. I simply state a different point of view both in this comment and my answer. You keep saying my statements "inaccurate" without being able to point out any fallacy makes me really confused. – Nov 08 '21 at 13:37
-
-
@LouisPan in a calculus classroom, the domain of a function is almost never specified, and they certainly never used terminology such as "open subset of the natural domain" in such classrooms. Did you miss this part? Should I highlight it with bold font? I did not think this was necessary, but I will edit my reply if necessary. – Angel Nov 08 '21 at 13:39
-
"open subset of the natural domain" I did not say that. You missed the key term "connected". Besides, "in a calculus classroom..." is not a mathematical statement but simply a personal taste of teaching, isn't it. // Still, if you insist on using the natural domain, what is your answer to the integral mentioned above? Same answer with interpretation of $C$ as three different constants $C_1,C_2,C_3$ in each component? Or writing three cases as you did in your answer? – Nov 08 '21 at 13:50
-
in a calculus classroom, the domain of a function is almost never specified, and they certainly never used terminology such as "open subset of the natural domain" in such classrooms. This addresses nothing, and it does not change the fact that you said this. Me omitting a word from the paraphrasing is inconsequential to the point I am making against your quote. – Angel Nov 08 '21 at 14:03
-
Besides, "in a calculus classroom..." is not a mathematical statement but simply a personal taste of teaching, isn't it This is disingenuous on your part. You are the one who brought up exercises in calculus, not me. I am merely responding in kind: what my reply does is show that your claim about calculus classrooms is false, and it contradicts the very stance you are attempting to spouse: that connected open subsets of the natural domain are what these exercises consider - they do not, as a matter of fact. You will be hard pressed to find textbooks using that terminology nowadays. – Angel Nov 08 '21 at 14:05
-
I don't see any harm in simplification of things regarding the domain issue, do you? – Nov 08 '21 at 14:05
-
Still, if you insist on using the natural domain, what is your answer to the integral mentioned above? Same answer with interpretation of C as three different constants C1,C2,C3 in each component? Or writing three cases as you did in your answer? Not only is this addressed by my answer by implication, but this also fails to ask anything that helps expand on the topic. If you want to have a long-term discussion, then you should move this to chat. But frankly, I am also not really interested right now. – Angel Nov 08 '21 at 14:08
-
"Not only is this addressed by my answer by implication," Okay, I take that as you prefer considering the "maximum" domain, and writing out all the cases in different connected components; I'm okay with that. Obviously, we have very different points of view in the "calculus classroom" regarding this "domain" issue. I am okay with the disagreement. I will stop here. – Nov 08 '21 at 14:17
In the formula $$ \int \frac{1}{x}\; dx = \log(x)+C $$ the logarithm is with base $e$. Most calculus books write $$ \int \frac{1}{x}\; dx = \log|x|+C\tag{1} $$ with the absolute value on the right-hand side. The formula (1) is understood as $\log|x|$ is an antiderivative of $\frac1x$ on any open interval contained in the set $\mathbb{R}\setminus\{0\}$.
[If one wants to consider any domain for the function $\frac1x$ that is not connected, for instance, its natural domain $(-\infty,0)\cup(0,\infty)$, then one needs more than one "artibrary constants" and different cases for each connected component of the domain on the right-hand side. In other words, the general solutions to the differential equation $$ \frac{dy}{dx}=\frac{1}{x} $$ on an open subset that is not connected has more than one "arbitrary constant" C's.]
In many other contexts, the base may not be important due to the fact that they only cause a factor of a constant. But when doing integrals, the natural log is the one that is an antiderivative of $\frac1x$.
-
This is not quite right, since even if $\log(|x|)$ is an antiderivative of the function everywhere in $\mathbb{R}\setminus{0}$, the constants of antidifferentiation need not be the same on both sides of $0$, even though your answer assumes they must be. – Angel Nov 08 '21 at 12:44
-
@Angel: "everywhere" is not what is said here. It is written in the answer that "The formula (1) is understood as $\log|x|$ is an antiderivative of $\frac{1}{x}$ on any open interval contained in the set $\mathbb{R}\setminus{0}$". Different intervals may or may not give you a different constant "C". – Nov 08 '21 at 12:49
-
I never said that you said that. I was not quoting you. You seem like you are antagonizing, rather than accepting the still-valid criticism to your comment. Besides, what I said is correct: it is indeed an antiderivative everywhere in $\mathbb{R}\setminus{0}$, regardless of whether you said so or not. We are not considering $h(x)=\frac{1}{x}$ to be a function restricted to an open interval, unless otherwise specified. – Angel Nov 08 '21 at 12:54
-
@Angel: Fair enough. I only consider functions defined in an open connected subset of $\mathbb{R}$, not a natural domain. Thus the formula interpreted here does not apply to questions such as what is the set of antiderivatives for $\frac{1}{x}$ in the set $\mathbb{R}\setminus{0}$. – Nov 08 '21 at 13:00
The formula is $$\int\,\frac{\mathrm{d}x}{x}=\ln(|x|)+C$$
For the definition of the natural logarithm as an integral we have
$$ Ln (x) = \int _1^x \frac {1}{t} dt , x>0 $$
- 68,728