A classical version of the inverse function theorem for 1D asserts the following.
Suppose that is an invertible function with inverse . If is differentiable at with and is continuous at , then is differentiable at and
Very often, some of the above hypotheses for is replaced by, say, is continuous “in a neighborhood” and one-to-one. But in the proof, while the one-to-one property immediately gives the existence of the inverse function , the “non-local” continuity yields the continuity of the inverse . Here we are interested in the “local” continuity/differentiablity.
In this note, we show by constructing counter-examples that in general the continuity of at cannot be dropped. In the sequel, we denote by the set set of positive integers, namely .
1. The case of continuity
We start with the simplest case, namely if we only assume that is continuous at . In this scenario, the following example due to R.E. Megginson is well-known. Consider the function defined by
It is not hard to see that is well-defined and invertiable. Moreover, is continous at . However, as , its inverse function fails to be continous at because
as . This concludes the construction.
2. The case of zero derivative
We now modify the above example to handle the case of differentiable functions at the given point. A careful calculation shows that in the above example, the funtion is not differentiable at because
Toward a new construction, let us denote
which is one-to-one. Having the function , we construct the desired function as follows.
It is not hard to see that is well-defined and invertiable. We now show that is differentiable at with . To see this, we just consider for . From the definition of and the monotonicity of we note that
for any . Hence
This shows that is differentiable at with . Arguing as in the case of continuity, we know that
and
as . This shows that is not continuous at .
3. The case of non-zero derivative
We are now arrive in the most delicated case. For simplicity, let us construct some bijective mapping . The following example is due to Mr. Nghia. The desired function is given as follows
It is now standard to verify that is well-defined and invertiable. Obviously, is not the whole because . However, there holds and . We now examine at 0 and its inverse at 0. We now show that is differentiable at with . Again, we just consider for . Clearly, it suffices to show that
for any sequence converging to 0. This is done once we can show that both
hold true, where and is any sequence having no term . This is because given arbitrary , there are two positive integers and such that
and
Simply taking we conclude that the above estimate holds for all regardless of either or . Going back to the limits while the case of is trivial by the definition of , the case is also simple because in this scenario, we easily get
for some and some . Hence
As , we must have . Hence from the above two-sided estimate we must have
Hence
as claimed. Thus, we have just shown that is differentiable at with . Finally, we show that the inverse function is not continuous at . Indeed, this follows from
as .
Our last observation concerns the case when is twice differentiable at . However, in this case is differentiable in a neighborhood of , which is enough to conclude that its inverse must be continuous at .
Leave a comment