Newton-Raphson Method
The Newton-Raphson method is an iterative way to find the roots of a function. The formula for this method is as follows: x[i+1] = x[i] - f(x[i])/f'(x[i]). Since the derivative is the denominator, an f' value of 0 can throw us off, as well as a discontinuous or non-existant derivative at a root. Our last major condition to take into account is if the function does not converge onto a root. In this case, you will want to cap on the number of iterations.
One common use of the Newton-Raphson method is to find the square roots of numbers. This is quite easy to do by setting up a quadratic: x^2 - n = 0, where n is the number we want to find the square root of. For example, let's say n = 2. So our derivative is 2x, and we set up our equation: x[i+1] = x[i] - (x^2-2)/2x. Now we just need a good initial estimate. Since we know the sqrt(2) is between 1-2, 1.75 is a good estimate. Try to guestimate rather than picking arbitrary numbers. Guestimation can improve accuracy and the number of iterations for convergence.
So for our iterations, we get:
x[1] = 1.75 - (1.0625)/3.5
x[2] = 1.45 - (1.45^2-2)/2.9
x[3] = 1.41 - (1.41^2 - 2)/2.82
After only a few iterations, we start to see x converge to the sqrt(2), which is 1.4142.
There are some functions though, for which the Newton-Raphson method is not effective. One such function is f(x) = x^2+3. If we try to solve this using the quadratic formula: x = +-sqrt(-4)/2, we get imaginary roots. So when we apply the Newton-Raphson method here, it will not converge.
A good initial guess here is x = 1. So applying that, we get:
x[1] = 1 - 4/2 = -1
x[2] = -1 + 4/2 = 1
As we see, the Newton-Raphson method will return a list of oscillating values.
Intermediate Value Theorem
The intermediate value theorem is pretty simple. If we think about driving on the highway, we can apply the intermediate value theorem. In the given context, it means we go from 60-65 mpr, then we have to go through 61-64 mph to get there. Generalizing it back to math, the formal definition is (given that the function is continuous) that if there is an interval [a,b], there is some number c between f(a) and f(B) inclusive such that f(x) = c.
The squeeze theorem is an extension of the intermediate value theorem. It states that given functions f(x), g(x), and h(x) such that f(x) <= g(x) <= h(x), if lim(x-->a)f(x) and lim(x-->a)h(x) converge on the same value at x = c, then lim(x-->a)g© converges on that same value. Essentially, g(x) is being squeezed between f(x) and h(x). The following image demonstrates graphically how this appears.
Mean Value Theorem
The mean value theorem states that at some point on the interval [a,b], there exists a point c where the instantaneous velocity is equivelant to the average velocity. Recall that the average velocity is simply the slope formula: deltaY/deltaX. So f'© = (y2 - y1)/(x2 - x1). If we know f'(x) or can derive it, solving for c becomes relatively straight-forward.
Extreme Value Theorem
As the name implies, the extreme value theorem deals with the extrema (maxima and minima) of a function on a closed interval [a,b]. It states that on such an interval, the function takes on a maximum value at some point, as well as a minimum value at another point. It is this theorem that tells us when examining an interval to check our endpoints as well as critical points (points where the derivative = 0) to find the local extrema.
Continuity
The one factor that makes Calculus possible is continuity. We can think of this graphically as us putting pen to paper and tracing over a graph without having to lift the pen up. That is, the function has no gaps or breaks in it. We can also evaluate continuity on an interval [a,b].
So why do we need continuity? We know that derivatives are slopes, and slopes require two points to calculate. So if we were limited to a discrete point, there would be know f(x+h) to evaluate. Note that continuity does not necessarily imply differentiability. As we saw with absolute value functions in my first tutorial, they are not differentiable at the vertex/cusp. This applies to all functions, not just absolute value functions. No function is differentiable at a cusp/vertex/sharp turn. As we will also see with integrals, which are used to find area under curves, there is no area under a discrete point. Without a continuous function, we cannot find the area.
So what exactly is continuity? Let's use game and graphics programming as an analogy. When animating an object programatically, we deal in how many frames/second an object moves. Each second, we might move an object 45 frames/second; and for each frame, we might move the object 3 pixels. So these are discrete movements that appear continuous. If an objects starts at (0,0) and moves to (3,3), it didn't actually go through (1,1), (1.5, 1.5), etc. What if we make our step-size 1 instead of 3? It is still discrete. We never go through (.5, .5) on the way to (2,2). So how is this relevant? Basically, if there are enough discrete points along a function, it can be called continuous. The more points we have, the more accurate our calculus is. In other words, a function becomes increasingly continuous as lim(numPoints-->infinity).
Conclusion
This tutorial provides a lot of the background and theorems to support the Calculus covered up to this point. In my next tutorial, I will introduce integration and the fundamental theorem of calculus.
MultiQuote
| 