Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

$\displaystyle \qquad y'[x] = \frac{y}{x-y}=\frac{ \frac{y}{x} }{ 1-\frac{y}{x} }=\frac{u}{1-u}$


$\displaystyle \qquad u = \frac{y}{x} \rightarrow y=ux,\qquad y'[x] = u$

$\displaystyle \qquad u = \frac{u}{1-u} \rightarrow u^{2}=0$

But wolframalpha states that it a first order non-linear ODE here and its solution: $$y(x) = - \frac{x} {W(-e^{-C_{1} x})}$$ where $W(z)$ is a product log function(?!).

Could someone explain which way is the right and which is really the solution. My friend just said that he solved it by integrating, separating $x$'s and $y$'s to different sides but I cannot get it that way.

[update] thanks to joriki: $$\frac{d u}{d x} = \frac{u}{1-u}-u=\frac{u^{2}}{1-u}$$ and after integrating $$\frac{-1}{u}-\ln(u)-x+C=0$$ but not yet the solution, thinking...but how to get to the solution suggested by WA now? I cannot see a way to solve it now just for x or just for y. Ideas?

share|cite|improve this question
The step from $y=ux$ to $y'=u$ is wrong; it should be $y'=u+u'x$ instead, since $u$ may depend on $x$. – joriki May 25 '11 at 19:10
note that, in the update, you used $y'=u+u'$ instead of $y'=u+u'x$. – Shai Covo May 25 '11 at 20:56
If you are interested in telling if it's linear, then use the linear operator format L[y] to describe the equation, and check whether L[$y_1+y_2$]= L[$y_1]+L[y_2$], and L[cy]=cL[y] – gary Jun 9 '11 at 14:10
up vote 6 down vote accepted

As Shai pointed out in the comments, the factor $x$ is missing in your update. It should be


and thus

$$ \begin{eqnarray} u'\frac{1-u}{u^2} &=& \frac{1}{x}\;, \\ -\frac{1}{u}-\log u &=& \log x + c\;, \end{eqnarray} $$

which leads to Shai's result if you plug in $u=y/x$ and cancel $\log x$. To get to the WolframAlpha result, you can switch to $v=-1/u=-x/y$ to get

$$ \begin{eqnarray} v+\log(-v) &=& \log x + c\;, \\ v\mathrm e^v &=& -\mathrm e^c x\;, \\ v &=& W(-\mathrm e^c x)\;, \\ -\frac{x}{y} &=& W(-\mathrm e^c x)\;, \\ y &=& -\frac{x}{W(-\mathrm e^c x)}\;. \end{eqnarray} $$

Note that there's a typo in your rendition of WolframAlpha's result; the $x$ shouldn't be in the exponent.

share|cite|improve this answer

You may prefer the following approach.

First rewrite the equation as $ - y'y = y - y'x$, leading to $$ - \frac{{y'}}{y} = \frac{{y - y'x}}{{y^2 }}, $$ and in turn $$ - \frac{d}{{dx}}\log y = \frac{d}{{dx}}\frac{x}{y}. $$ Integrating both sides then gives $$ - \log y = \frac{x}{y} + C. $$

share|cite|improve this answer
...but it is not anything that Wolframalpha gives but the form is similar to mine. What is the solution of this function then? You don't have $y$ as a function of just $x$ but you have $y$ there as parameter -- so what is the conclusion? How does WA solves the solution? – hhh May 25 '11 at 22:52
As stated, this is only an alternative approach, which you may prefer. To continue, you may consider the ProductLog function; see – Shai Covo May 25 '11 at 23:09

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.