We're about to begin Lecture 57, Bonus material. Well, that seems like a bad way to end the course, talking about what we can't do. Let's change things up a little bit. Go back to the very beginning of this class. I began with the question, what is e to the x? Well by now, you all know exactly what e to the x is and means, on so many different levels. But let's reconsider this question, one more time from the point of view in Discrete Calculus. And what I'd like to emphasize is a particular point of unity, between Discrete and Continuous Calculus. This is Newton's formula, something that I've been saving for a special occasion. Let's begin. Recall that the forward difference operator Delta can be expressed as the Shift e minus the identity I. If we rearrange terms a little bit, then we can express the left Shift E as Delta plus I. Now, let's say that we want to shift over not one slot, but, n places. Then, we can express this operator as a power, E to the n, which thus can be expressed as quantity Delta plus I to the n. Now, what happens when we multiply that out? Well, we can use the binomial theorem to express E to the n. As the sum, k goes from zero to n of the coefficient. And choose k times delta to k times I to the n minus k. These binomial coefficients you recall, comes from say Pascal's Triangle or the definition in terms of Factorials of n and k and n minus k, and something like that. Anyhow, let's look a little bit more carefully. Recall that the identity operator, I, is the, do nothing operator. So we can effectively ignore it and just look at powers of the forward difference Delta.
Now, remember, one of the
useful formulate from Discrete Calculus, is the notion of a falling power. N to the falling k is n times n minus 1, times n minus 2, all the way to n minus k plus 1. But this is related to the binomial coefficient as follows, n to the falling k can be expressed as n choose k times k factorial. Now that means that we could rewrite E to the n as the sum, k goes from 0 to n of n to the falling k, divided by k factorial, times delta to the k. And at this point, if you've been paying attention in this class, you see something that looks slightly familiar. If we take a sequence, a, and we consider the nth term, a sub n, this is really the shift E to the N applied to a, evaluated at 0 at the initial slot. That means that we can express a sub , as the sum k goes from 0 to n on one of the k factorial, times the kth forward difference Delta to the k of a. Evaluated at 0, times n to the falling k. This is Newton's formula in slightly a different language, than you'll find it in most books, but in the language of Discrete Calculus. A language that makes you see immediately the relationship to Taylor Series. Indeed Newton's formula is the descriptive calculus version of our classical formula for the Taylor Series of F at X. Let's look at the similarities, in both cases, we have a one over k factorial and we have in the discreet version the kth forward difference of a evaluated at zero. In the smooth version we have the kth derivative of f, evaluated at zero. These look a little bit closer. If we replace that kth derivative of f by the kth power of the differentiation operator, capital D applied to f, and then evaluate it at zero. Whereas in the smooth case, we have x to the k, in the discrete case, we have the discrete version of this monomial, that is, n to the falling k. This beautiful and fearful symmetry
between the notion of a Taylor Series for
Discrete and Continuous Calculus is one of the gems in this course. If you don't mind, I would like to make a few observations, here at the very end. Recall the shift operator E, applied to a sequence a, and evaluated at the nth term. Is really just a sub n plus one, and in operator language, it can be expressed as I plus Delta. There's a corresponding shift operator in the continuous world as well. E applied to a function f and evaluated at x is simply f at x plus 1. Now, way back when, we argued using Taylor Series, that the shift operator, in the smooth world, is really the exponentiation of the differentiation operator. That is, E is I plus D plus 1 over 2 factorial times D squared, plus 1 over 3 factorial D cubed, etcetera. And seeing the relationship between the smooth and the discrete versions of a shift lead us to a few interesting observations. First of all, from the observation that the shift operator is the exponentiation of the differentiation operator. We can express the discrete derivative, the forward difference delta as e to the D minus I. This gives us an explicit connection between the notions of differentiation in the discrete and the continuous world. Lastly, we could say that differentiation D is really the natural log of the shift operator. Or, if you like, the natural log of the identity plus the forward discrete difference delta. Those are all beautiful observations, and all are connected to this simple initial question which with we began, what is e to the x? By now, you have seen so many answers to that question from so many different perspectives. But you haven't seen it all Indeed, there is much more to Mathematics than just Calculus. Even the type of calculus that you have seen in this course. I hope that this course has been for you more than a bag of tricks, or formulae written into boxes. I hope that you have seen the beauty and the wonder that is implicit
both in the world around us, and the
mathematics that we use to model it. I hope that this course has been for you More than just a few video lectures in which to up your skill set. Pay attention to the things around you, go deeper. Maybe, take a few more math classes to see more of what's out there. It has been my privilege and pleasure to be your professor this term.
(Seminars in Mathematics 11) N. M. Ivochkina, A. P. Oskolkov (Auth.), O. A. Ladyzhenskaya (Eds.) - Boundary Value Problems of Mathematical Physics and Related Aspects of Function Theory-Springer