Sign in

username:

password:



Not a member?

Search Online Books



Search tips

Free Online Books

Ads

Chapters

Chapter Contents:

Search Mathematics of the DFT

  

Book Index | Global Index


Would you like to be notified by email when Julius Orion Smith III publishes a new entry into his blog?

  

Points of Infinite Flatness

Consider the inverted Gaussian pulse,E.1

$\displaystyle f(x) \isdef e^{-1/x^2}, % \frac{1}{x^2}},
$

i.e., $ f(1/x)$ is the well known Gaussian ``bell curve'' $ e^{-x^2}$. Clearly, derivatives of all orders exist for all $ x$. However, it is readily verified that all derivatives at $ x=0$ are zero. (It is easier to verify that all derivatives of the bell curve are zero at $ x=\infty$.) Therefore, every finite-order Maclaurin series expansion of $ f(x)$ is the zero function, and the Weierstrass approximation theorem cannot be fulfilled by this series.

As mentioned in §E.2, a measure of ``flatness'' is the number of leading zero terms in a function's Taylor expansion (not counting the first (constant) term). Thus, by this measure, the bell curve is ``infinitely flat'' at infinity, or, equivalently, $ f(x)$ is infinitely flat at $ x=0$.

Another property of $ f(x) \isdef e^{-\frac{1}{x^2}}$ is that it has an infinite number of ``zeros'' at $ x=0$. The fact that a function $ g(x)$ has an infinite number of zeros at $ x=x_0$ can be verified by showing

$\displaystyle \lim_{x\to x_0} \frac{1}{(x-x_0)^k} g(x) = 0
$

for all $ k=1,2,\dots\,$. For $ f(x)$, the existence of an infinite number of zeros at $ x=0$ is easily shown by looking at the zeros of $ f(1/x)$ at $ x=\infty$, i.e.,

$\displaystyle \lim_{x\to\infty} x^k f(1/x) = \lim_{x\to\infty} x^k e^{-x^2} = 0
$

for any integer $ k$. Thus, the faster-than-exponential decay of a Gaussian bell curve cannot be outpaced by the factor $ x^k$, for any finite $ k$. In other words, exponential growth or decay is faster than polynomial growth or decay. (As mentioned in §3.10, the Taylor series expansion of the exponential function $ e^x$ is $ 1 +
x + x^2/2 + x^3/3! + \dots$--an ``infinite-order'' polynomial.)

The reciprocal of a function containing an infinite-order zero at $ x=x_0$ has what is called an essential singularity at $ x=x_0$ [13, p. 157], also called a non-removable singularity. Thus, $ 1/f(x) = e^{\frac{1}{x^2}}$ has an essential singularity at $ x=0$, and $ e^{x^2}$ has one at $ x=\infty$.

An amazing result from the theory of complex variables [13, p. 270] is that near an essential singular point $ z_0\in{\bf C}$ (i.e., $ z_0$ may be a complex number), the inequality

$\displaystyle \left\vert f(z)-c\right\vert<\epsilon
$

is satisfied at some point $ z\neq z_0$ in every neighborhood of $ z_0$, however small! In other words, the function comes arbitrarily close to every possible value in any neighborhood about an essential singular point. This result, too, is due to Weierstrass [13].

In summary, a Taylor series expansion about the point $ x=x_0$ will always yield a constant approximation when the function being approximated is infinitely flat at $ x_0$. For this reason, polynomial approximations are often applied over a restricted range of $ x$, with constraints added to provide transitions from one interval to the next. This leads to the general subject of splines [78]. In particular, cubic spline approximations are composed of successive segments which are each third-order polynomials. In each segment, four degrees of freedom are available (the four polynomial coefficients). Two of these are usually devoted to matching the amplitude and slope of the polynomial to one side, while the other two are used to maximize some measure of fit across the segment. The points at which adjacent polynomial segments connect are called ``knots'', and finding optimal knot locations is usually a relatively expensive, iterative computation.


Order a Hardcopy of Mathematics of the DFT

Previous: Weierstrass Approximation Theorem
Next: Differentiability of Audio Signals

written by Julius Orion Smith III
Julius Smith's background is in electrical engineering (BS Rice 1975, PhD Stanford 1983). He is presently Professor of Music and Associate Professor (by courtesy) of Electrical Engineering at Stanford's Center for Computer Research in Music and Acoustics (CCRMA), teaching courses and pursuing research related to signal processing applied to music and audio systems. See http://ccrma.stanford.edu/~jos/ for details.


Comments


No comments yet for this page


Add a Comment
You need to login before you can post a comment (best way to prevent spam). ( Not a member? )