Everywhere Continuous Nowhere Differentiable Functions

Here I discuss the use of everywhere continuous nowhere differentiable functions, as well as the proof of an example of such a function. First, I will explain why the existence of such functions is not intuitive, thus providing significance to the construction and explanation of these functions. Then, I will provide a specific detailed example along with the proof for why it meets the requirements not only of being continuous over the whole real line but also of not having a derivative at any real number. I will next analyze the proof and compare the function to another example of a continuous everywhere nowhere differentiable function in order to pull out how these functions sidestep intuition. 1. Continuity and Differentiability: The Failure of Intuition In mathematics, not all concepts which seem intuitive at first are indeed correct. Fortunately, often having enough experience working in a field will expose a person to the field’s intricacies, and thus change the person’s intuition or at least make it weary of certain ideas. On the other hand, sometimes the counterexamples to an idea can be rare or contrived, preventing the intuition from naturally adapting and leaving the person vulnerable to making a fatal assumption. Hence, with such concepts, going out of one’s way to remember a counter-example can keep one on one’s toes, and understanding it will give one a more nuanced understanding of the ideas involved. The relationship between the continuity and differentiability of a function contains an example of a deceptively intuitive link, though a number of intuitive ideas about the relationship apply. In one direction, the straight forward idea holds: if a function is differentiable at a point, then it must be continuous there as well. It is also not counter-intuitive why the opposite implication, that a derivatives exist wherever a function is continuous, does not hold. Often one of the first things a calculus student learns is that if a function exhibits a ‘sharp turn’ that the derivative does not exist, and the absolute value function provides an easily conceptualized example of this concept. At 0 the left handed derivative of the absolute value function equals -1, as the function in question is a simple linear function with a slope of -1, and similarly the right handed derivative equals 1. Because derivatives are based on limits, if the right handed derivative and the left handed derivative are not equal then the derivative as a whole cannot exist. Hence, though the absolute value function is continuous at 0, it is not differentiable there. Even without the formalism, one can see that at ’sharp turns’ the tangent line wants to have two distinct slopes: the slope as conceived from 1 the left and the slope as conceived from the right. The logical choice is to not choose between them and to declare the derivative non-existent, as is done formally. The deception occurs in that for continuous functions failure of differentiation seems rare. Given a continuous function, one typically assumes that the derivative exists at most points, though the derivative could exist nowhere. A couple of factors fuel the incorrect intuition. For one, a number of the functions that math students first learn to work with behave nicely with respect to the derivative over their domain. These include polynomial, rational, trigonometric, exponential, and logarithmic functions, which in fact are differentiable everywhere on their domain. Of the power functions, which have the form f(x) = x for some a ∈ R, only have points for which they are not differentiable for 0 < a < 1. This is because by the power rule of differentiation we will have f ′(x) = axa−1, if a 6= 0. If a ≥ 1, the derivative continues to exist everywhere. If a < 0, it does not exist at 0, but that was not a part of the functions domain. For 0 < a < 1, then, f exists at everywhere and is differentiable everywhere but at 0. In the case that a = 0, we merely have f(x) = 1, which has derivative 0 everywhere. Even when these commonly used functions have continuous points without a derivative, they don’t typically have many. For example, the absolute value function and the power functions described above only lack a derivative at a continuous point at 0. One can think of methods such as forcing periodicity to increase the number of continuous non-differentiable points given that one such point exists, but even that will not force the number of non-differentiable points into the uncountable range if it were not there already. Hence these functions give the idea that non-differentiability is actually relatively rare for continuous functions. Moreover, both the ideas of a function’s continuity and differentiability relate, at least intuitively to ideas of smoothness. Before learning the formal definition of continuity, often students hear the idea that if one can draw a function without picking up one’s pencil then it is continuous. As stated, this intuitive test manifests a logically correct idea. If a function is not continuous at a point c then no mater how small of an interval one creates around c, there will always be points at least away for some > 0. Hence, when drawing the function one would need to move away from f(c), for all intents and purposes, hence requiring the pencil to be picked up. Unfortunately, until students reach more formal classes such as Real Analysis, students typically learn how to find discontinuities in a few types of functions and that elsewhere these functions are continuous. These functions are the one’s noted above: polynomial, rational, trigonometric, exponential, and logarithmic functions. Hence, students do not confront early on functions that are nearly impossible to draw, let alone draw without picking up one’s pencil in terms of continuity. Yet, these functions can still have points at which they are continuous. the Thomae function, which provides an example of a function continuous on the irrationals, where it has non-zero values at all the rational numbers and is valued 0 at the irrationals 1. This would be quite impractical to draw because both the rationals and the irrationals are