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

Using the standard notation $(x,y)$ for cartesian coordinates, and $(r, \theta)$ for polar coordinates, it is true that $$ x = r \cos \theta$$ and so we can infer that $$ \frac{\partial x}{\partial r} = \cos \theta.$$ This means that if we perturb $r$ to $r+\Delta r$, $x$ gets perturbed to approximately $x+ \cos(\theta) \Delta r$. This is equivalent to saying that if we perturb $x$ to $x+\Delta x$, then we perturb $r$ to approximately $r+ (1/\cos \theta) \Delta x$. Correspondingly, we expect that $$ \frac{\partial r}{\partial x} = \frac{1}{\cos \theta}.$$ In fact, $$ \frac{ \partial r}{\partial x} = \frac{\partial}{\partial x} \sqrt{x^2+y^2}= \frac{2x}{2 \sqrt{x^2+y^2}} = \frac{x}{r} = \cos \theta.$$ What gives?

share|cite|improve this question
If you are in the $(r,\theta)$ plane, a change in $r$ does not affect $\theta$; the change in $x$ is proportional to the change in $r$, with the constant of proportionality given by $\cos\theta$, which is constant because $\theta$ is not changing. But if you are in $(x,y)$-plane, a change in $x$ affects both $r$ and $\theta$, so you cannot consider $\cos\theta$ a fixed quantity; this means that the claim that perturbing $x$ by $\Delta x$ corresponds to perturbing $r$ by $\frac{1}{\cos\theta}\Delta x$ is unjustified and unclear: which $\theta$? $\theta$ changes! – Arturo Magidin Feb 7 '11 at 22:05
Good question. I had the same questions when I first learnt partial derivatives. My teachers did not emphasize what we were doing and used to simply write out the answer using the appropriate expressions. – user17762 Feb 7 '11 at 22:05
thanks, everyone – calc student Feb 7 '11 at 22:33
up vote 8 down vote accepted

When you take partial derivatives, what you are doing is to hold the other variable fixed. So for instance if $f$ is a function of $n$ variables, $f(x_1,x_2,\ldots,x_n)$, $\frac{\partial f}{\partial x_i}$ means you hold all the other $x_j$'s constant except $x_i$ and vary $x_i$ by $\delta x_i$ and find out what happens to $\delta f$.

We have

$x= r \cos(\theta)$,$y= r \sin(\theta)$, $r^2= x^2 + y^2$ and $\tan(\theta) = \frac{y}{x}$.

We are transforming from the $(x,y)$ space to $(r,\theta)$ space.

In the $(x,y)$ space, $x$ and $y$ are independent variables and In the $(r,\theta)$ space, $r$ and $\theta$ are independent variables.

$\frac{\partial x}{\partial r}$ means you are fixing $\theta$ and finding out how changing $r$ affects $x$.

So $\frac{\partial x}{\partial r} = \cos(\theta)$ since $\theta$ is fixed.

$\frac{\partial r}{\partial x}$ means you are fixing $y$ and finding out how changing $x$ affects $r$.

So $\frac{\partial r}{\partial x} = \frac{x}{\sqrt{x^2+y^2}}$ since $y$ is fixed.

When you do $r = \frac{x}{\cos(\theta)}$ and argue that $\frac{\partial r}{\partial x} = \frac{1}{\cos (\theta)}$ you are not holding $y$ constant.

If you were to hold $y$ constant, then $\theta$ would change as well.


I am adding this in the hope that this will make it a bit more clear. If you want to use $r = \frac{x}{\cos(\theta)}$ and still derive it, you need to do as follows:

$r = \frac{x}{\cos(\theta)}$, $\delta r = \frac{\delta x}{\cos(\theta)} + \frac{-x}{\cos^2(\theta)} (-\sin(\theta)) \delta \theta$.

$\tan(\theta) = \frac{y}{x} \Rightarrow x \tan(\theta) = y$

$\delta x \tan(\theta) + x \sec^2(\theta) \delta \theta = 0$ (Since $y$ is held constant)

$x \delta \theta = - \frac{\delta x \tan(\theta)}{\sec^2{\theta}} = - \sin(\theta) \cos(\theta) \delta x$.

Plugging the above in the previous expression, we get

$\delta r = \frac{\delta x}{\cos(\theta)} + \frac{\sin(\theta)}{\cos^2(\theta)} (-\sin(\theta) \cos(\theta) \delta x) = \frac{1-\sin^2(\theta)}{\cos(\theta)} \delta x = \cos(\theta) \delta x$ and hence we get

$\frac{\partial r}{\partial x} = \cos(\theta) = \frac{x}{\sqrt{x^2+y^2}}$

share|cite|improve this answer
Or, alternatively to "not holding $y$ constant", one is not applying the chain rule properly, since $\theta$ also changes with $x$. – Arturo Magidin Feb 7 '11 at 22:25
@Arturo: Right. I was adding that when you made this comment. – user17762 Feb 7 '11 at 22:27
I would only add a reference to the inverse function theorem:… – AppliedSide Feb 8 '11 at 0:08

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.