0

According to my book: $$\ln(\lvert y \rvert)=\ln(\lvert 1+x \rvert)+ \ln(\lvert c \rvert)$$ $$\Rightarrow \ln(\lvert y \rvert)=\ln(\lvert c(1+x) \rvert)$$ $$\Rightarrow y=c(1+x)$$

I know that this is correct. However, I want to be able to understand the second step clearly. I want to make sure if this is why : $$\ln(\lvert y \rvert)=\ln(\lvert c(1+x) \rvert) \Rightarrow \lvert y \rvert=\lvert c(1+x) \rvert$$ $$\Rightarrow y=± c(1+x) \Rightarrow y=C(1+x)$$ where $C=±c$ is just a substitution for the constant. Am I correct? Is this an often-used method that is considered rigorous even if we do not rename the constant?

Edit: $c$ is not a fixed constant, but an arbitrary one. To give some context, the equation to be solved is obtained from integration.

1 Answers1

2

Yes, you are correct. Though I would not write $C=\pm c$. Note that $c$ is a fixed (nonzero) constant. You cannot define $C$ with two values.

In general, what you have is that $$ |y|=|kx| $$ if and only if $$ y=kx\quad \textrm{or}\quad y=-kx\tag{1} $$

(1) is sometimes written as $y=\pm kx$.


Notes 1.

If you get the identity when you solve an ODE and get two (sets of) solutions: $y=cx$ and $y=-cx$, you need to be careful if you want to combine them. In your case, the "arbitrary" constant $c$ is not zero. So you cannot conclude that $y=Cx$ where $C$ is "arbitrary". If you find further that $y=0$ is also a solution, then you can indeed claim that $y=Cx$, combining all the cases.


Notes 2. Consider for instance the ODE you mentioned in the comment: $$ \frac{dy}{dx}=\frac{y}{1+x}\tag{0} $$ To solve it, \begin{align} \frac{dy}{y}&=\frac{dx}{1+x}\\ \ln|y|&=\ln|1+x|+\ln|c|&(c\ne 0) \end{align}

(Remark. The reason why we can write $\ln|c|$ here is that the range of the function $x\mapsto \ln x$ ($x>0$) is the whole real line. So adding an "arbitrary" constant $c$ is the same as adding $\ln|c|$ for some arbitrary nonzero constant $c$.)

To go on, \begin{align} \ln|y|&=\ln|1+x|+\ln|c|&(c\ne 0)\\ \ln|y|&=\ln|c(1+x)|&(c\ne 0)\\ |y|&=|c(1+x)|&\textrm{(because $\ln x$ is injective)} \end{align}

Now since $c$ is an arbitrary nonzero constant, you have following set of solutions, $$ y=c(1+x),\quad c\ne 0.\tag{1} $$

Since $c$ is "arbitrary", you don't need to add $\pm$ in front of it.

Observe that $y\equiv 0$ is also a solution to (0), and it is not of the form in (1).

So combining (1) and the zero solutions together, you get that the general solution for (0) is $$ y=C(1+x) $$ where $C$ is an arbitrary constant. (Note that $C$ is not required to be nonzero anymore.)

  • I forgot to mention that c is an arbitrary constant, that is, an integration constant. So, it is actually not a fixed constant. I edited the post! – curiouss Jan 18 '21 at 16:27
  • 1
    @curiouss see my edits. –  Jan 18 '21 at 17:00
  • The ODE to be solved was $(1+x)dy-ydx=0$. The book recommended to use $\ln(\lvert c \rvert)$ rather than the usual $c$ for the integration constant and immediately got $y=c(1+x)$ as I have written. I can now notice that y=0 is a solution. So, what is the right approach to this problem? Did the book skip over some obvious steps? – curiouss Jan 18 '21 at 17:21
  • 1
    @curiouss: notes added. –  Jan 18 '21 at 19:12
  • That helps a lot, thanks! I still find it odd that the book just wrote: "From $\ln(\lvert y \rvert)=\ln(\lvert c(1+x) \rvert)$, we immediately get $y=c(1+x)$", without even mentioning that c≠0 or that it combined the two sets of solutions. – curiouss Jan 19 '21 at 09:43