I read an example from "Principles of Mathematical Analysis" by Rudin under the section 'Continuity and Compactness'.

According to the example,

Let $X$ be the half-open interval $[0,2\pi)$ on the real line, and let $f$ be the mapping of $X$ onto the circle $Y$ consisting of all points whose distance from the origin is $1$, given by $f(t)=(\cos t, \sin t) \quad(0 \leq t < 2\pi)$

Now, we know that $f$ is continuous, since $f$ is one to one and onto, it has $f^{-1}$, but $f^{-1}$ is not continuous.


since, $f$ maps the points from $[0,\pi/2]$ to quarter of the circle in the I quadrant, $[\pi/2,\pi]$ to the II quadrant, $[\pi,3\pi/2]$ to the III quadrant, $[3\pi/2,2\pi)$ to the fourth quadrant, hence $f$ is continuous, similarly $f^{-1}$ maps the points in the quarter of the circle in the first quadrant to $[0,\pi/2]$, similarly the other $3$ quadrants..

it is given that $f^{-1}$ is not continuous at $(1,0)$, how is this possible?

According to the definition of continuity at $x_0$:

For every $\epsilon$ there exits a $\delta > 0$ such that $d(x_0,x)<\delta$ implies $d(f(x_0),f(x))<\epsilon$.

but a open ball at $(1,0)$ with $\delta$ radius, we can find the $\epsilon =2 \pi$ of radius, the above definition gets satisfied...

Obviously, when we move from I quadrant to the point $(1,0) f^{-1}$ converges to $0$ and when we move from from IV quadrant to the point $(1,0) f^{-1}$ converges to $2\pi$..hence left limit not equals right limit, hence it is discontinuous, but I want to know how it violates the definition, I cant find any, if any of my argument is wrong, please correct me.

Secondly, if $X$ is $[0,2\pi]$, then is $f^{-1}$ is continuous? if it is, how is that, how is the difference affects the continuity?


First of all, if $X=[0,2\pi]$, then $f$ is not bijective and therefore does not have an inverse.

Second, let's first remember how we can prove that something is not continuous. Since $f$ is continuous at a point $x_0$ if

For every $\epsilon>0$, there exists $\delta>0$ such that for every $x$, if $|x_0-x| < \delta$, then $|f(x_0)-f(x)| < \epsilon$.

Therefore, to prove that $f$ is not continuous, we need to find one point at which it is not continuous, and we can prove that it is not continuous if we show that

There exists some $\epsilon > 0$ such that for every $\delta >0$, there exists some $x$ such that $|x_0-x|<\delta$ and $|f(x)-f(x_0)|\geq \epsilon$

In our case, the inverse, $f^{-1}$ is not continuous at $(1,0)$, because there exists some $\epsilon$ (in particular, that value of $\epsilon$ can be $1$) such that for every $\delta$, you can find some point $(x,y)$ on $Y$ such that $f^{-1}((x,y))$ is more than $\epsilon$ away from $f^{-1}((1,0))$, which is $0$.

For example, if you take $\delta=0.1$, I can always choose $(x,y) = (\cos(-0.01), \sin(-0.01)$, and you will see that $f^{-1}((x,y))$ is very close to $2\pi$, so nowhere near $0$.


One way to do this is with sequences.

Suppose $f^{-1}$ is continuous at $(1,0).\ $ Then it must be the case that

$(x_n,y_n)\to (1,0)\Rightarrow f^{-1}((x_n,y_n))\to f^{-1}((1,0))=0 $ for any sequence $\left \{ (x_y,y_n) \right \}_n\subset S^1$.

But if we take $(x_n,y_n)=\left ( \cos \frac{(-1)^{n}}{n},\sin \frac{(-1)^{n}}{n} \right ),\ $ then $(x_n,y_n)\to (1,0)\ $ but $f^{-1}(x_n,y_n)$ does not converge, since the $f^{-1}(x_k,y_k)\to 0$ if $k=2n$ and $f^{-1}(x_k,y_k)\to 2\pi $ if $k=2n+1$.


The points $p_h=(\cos h,\sin h)$ and $q_h=(\cos h,-\sin h)$ can be made as close as desired by taking $h$ sufficiently small.

But $f^{-1}(p_h)=h$ and $f^{-1}(q_h)=2\pi-h$, so $f^{-1}(p_h)$ and$f^{-1}(q_h)$ cannot be made close by taking $h$ sufficiently small.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.