Fourier Series

(redirected from Fourier mode)
Also found in: Dictionary, Thesaurus.

Fourier series

[‚fu̇r·ē‚ā ‚sir·ēz]
(mathematics)
The Fourier series of a function ƒ(x) is Also known as Fourier expansion.
McGraw-Hill Dictionary of Scientific & Technical Terms, 6E, Copyright © 2003 by The McGraw-Hill Companies, Inc.
The following article is from The Great Soviet Encyclopedia (1979). It might be outdated or ideologically biased.

Fourier Series

 

a trigonometric series that provides a decomposition of a periodic function into its harmonic components. If a function f(x) is periodic with period 2T, then its Fourier series has the form

and a0, an, and bn (n > 1) are its Fourier coefficients. Depending on the nature of the integrals in the formulas for the Fourier coefficients, we speak of Fourier-Riemann series, Fourier-Lebesgue series, and so on. Periodic functions with period 2π are usually considered; the general case reduces to this case by a transformation of the independent variable.

Fourier series are the simplest class of expansions with respect to an orthogonal system of functions, namely, the trigonometric system 1, cos x, sin x, cos 2x, sin 2x,. . ., cos nx, sin nx. This system of functions possesses the important properties of closure and completeness. The Fourier sums

minimize the integral

where tn(x) is an arbitrary trigonometric polynomial of order ≤ n and f(x) is square integrable. Here,

so that the square integrable function f(x) can be approximated in the means by its Fourier sums with an arbitrary degree of closeness (seeAPPROXIMATION AND INTERPOLATION OF FUNCTIONS).

B. Reimann and H. Lebesgue showed that if f(x) is integrable then its Fourier coefficients an and bn tend to zero as x → ∞. Reimann also established that if f(x) has an improper Riemann integral then its Fourier coefficients need not tend to zero. If f(x) is square integrable, then the series

converges and we have the Parseval equality:

One variant of this equality was discovered by the French mathematician M. Parseval in 1799; the general formula (with the integral denoting the Lebesgue integral) was proved by Lebesgue. Conversely, given a sequence of real numbers an, bn for which the series

converges, we can find a square integrable (in the sense of Lebesgue) function with the Fourier coefficients an and bn; this result is due to the Hungarian mathematician F. Riesz and the German mathematician E. Fischer. For Riemann integrals the Riesz-Fischer theorem is false.

There are many convergence criteria for Fourier series, that is, sufficient conditions that guarantee the convergence of the series. The following condition, for example, is due to P. Dirichlet: if f(x) has a finite number of minima and maxima on a period interval, then its Fourier series converges everywhere. A more general condition is that of C. Jordan: if f(x) is of bounded variation (seeVARIATION OF A FUNCTION), then its Fourier series converges everywhere, and the convergence is uniform on every interval contained in the interval of continuity of f(x). The Italian mathematician U. Dini proved the following criterion in 1880: if f(x) is continuous and its modulus of continuity ω(δ, f) satisfies the condition

then its Fourier series converges uniformly.

The comprehensive investigation of convergence conditions of Fourier series is very difficult, and final results have not yet been obtained. Riemann showed that the convergence or divergence of a Fourier series at a point x0 depends on the behavior of the corresponding function f(x) in an arbitrarily small neighborhood of x0 (the localization principle for Fourier series). If different limits f(x0 – 0) and f(x0 + 0) exist and the Fourier series converges at x0, then it converges to {f(x0 – 0) + f(x0 + 0)}/2. In particular, if the Fourier series of a continuous periodic function f(x) converges everywhere, then its sum is f(x).

The German mathematician P. du Bois-Reymond showed in 1875 that there exist continuous functions whose Fourier series diverge at infinitely many points. Furthermore, A. N. Kolmogo-rov proved in 1926 that there exist Lebesgue integrable functions whose Fourier series diverge everywhere. L. Carleson, however, established in 1966 that the Fourier series of a square integrable function converges almost everywhere. As R. Hunt showed in 1968, this result holds for functions in any space Lp(–π, π) with p < 1. These “convergence defects” gave rise to various summability methods of Fourier series. Investigation of the behavior of Fourier sums is replaced by the study of their means, which in many cases exhibit far more regular behavior. For example, L. Fejér proved in 1904 the following: if f(x) is a continuous periodic function, then for n → ∞ its Fejér sum

converges uniformly to f(x).

REFERENCES

Tolstov, G. P. Riady Fur’e, 2nd ed. Moscow, 1960.
Bari, N. K. Trigonometricheskie riady. Moscow, 1961.
Zygmund, A. Trigonometricheskie riady, vols. 1–2. Moscow, 1965. (Translated from English.)
The Great Soviet Encyclopedia, 3rd Edition (1970-1979). © 2010 The Gale Group, Inc. All rights reserved.
References in periodicals archive ?
In the spectral domain, the Green's function is directly and efficiently available per Fourier mode but now the field-material interactions need careful attention, in view of the Fourier factorization rules that have become so widespread in RCWA and the differential method.
The computations were performed using 15 Fourier modes in x and in y and eight sample points in the z-direction per layer of holes or pillars, i.e., a total of 48 sample points in the z-direction.
Figure 2 shows the diffraction efficiency for the 0-th order reflection versus the truncation order, i.e., the upper index M of the Fourier modes in one periodic direction.
Owing to the structure of the separate matrices in the above equations, an efficient O([N.sub.z]M log M) matrix-vector product is obtained, where [N.sub.z] is the total number of unknowns, i.e., samples, in the aperiodic direction (z) and M is the total number of unknowns, i.e., Fourier modes, in the periodic direction (x).
When more Fourier modes are considered, the convergence rate appears to decrease and it seems that the relative error is bounded by a minimum value, which is smaller when no slicing is applied.
We have ascertained that field behavior in the distance of the slot (strip) edges is determined mainly by sums of Fourier components with low spatial frequencies, whereas the field asymptotics directly on the edges is caused by higher-order Fourier modes. An exponent of such power-type asymptotics has been computed with the help of the ratio of higher-order Fourier amplitudes.