Chapter 4 - Periodic Functions & Fourier Series Flashcards
Fourier Series
Overview
- considering trigonometric sine and cosine functions as a basis for the linear space of periodic functions on the line
- any reasonable (to be defined) periodic function can be written as an expansion in sine and cosine functions known as Fourier series
Periodic Function
Definition
-the function f(x) is said to be periodic, with (minimum) period 2L, if f(x+2L)=f(x) for all x in the domain of f
Linear Superposition
-clearly if f and g are both periodic functions with same period 2L, then any linear combination of them, c1f(x)+c2g(x) is also periodic with the same period
Fourier Series
Theorem
-given a function such that, f(x+2L)=f(x) we can find an, bn such that:
fN(x) = ao/2 + Σ[ancos(nπx/L) + bnsin(nπx/L)]
-where sum is from 1 to N and 1≤N
-and the Fourier coefficients are given by:
an = 1/L * ∫ f(x)cos(nπx/L) dx
bn = 1/L * ∫ f(x)sin(nπx/L) dx
-both integrals taken between -L and +L
Continuity
Definition
-a function f(x) is continuous at xo, if
lim f(x) = lim f(x)
-where the first limit is taken as x tends to xo from below and the second limit is taken as x tends to xo from above
-both limits must exist and be equal
Jump Discontinuity
Definition
-f(x) has a jump discontinuity at x=xo if both limits (x->xo-)lim f(x) and (x->xo+)lim f(x) exist but are not equal
Piecewise Continuity
Definition
-f(x) is piecewise continuous on interval [a,b] if it only has a finite number of jumps in the interval [a,b]
Piecewise Smooth
Definition
- f(x) is piecewise smooth on interval [a,b] if both f(x) and f’(x) are piecewise continuous on [a,b]
- this means that wherever the function jumps, the gradient also changes
Integral Over a Period
-when f(x) is piecewise continuous and has period T:
(0,T)∫ f(x) dx = (a,a+T) ∫ f(x) dx
-i.e. integral of f(x) between 0 and T is equal to the integral of f(x) between a and a+T
Orthogonality Relations
sine-sine
∫sin(mπx/L)sin(nπx/L)dx = {0, m≠n or L, m=n
Orthogonality Relations
cosine-cosine
∫cos(mπx/L)cos(nπx/L)dx = {0, m≠n or L, m=n≠0 or 2L, m=n=0
Orthogonality Relations
sine-cosine
∫sin(mπx/L)cos(nπx/L)dx = {0, for all m,n
Double Angle Identites
sinAsinB = 1/2[cos(A-B)-cos(A+B)] cosAcosB = 1/2[cos(A-B)+cos(A+B)] -in particular sin²A = 1/2[1-cos(2A)] cos²A = 1/2[1+cos(2A)]
Fourier Series
Inner Product
⟨f,g⟩ = ∫f(x) g(x) dx
-where the integral is taken between -L and +L
**check last paragraph from L16
Fourier Series
Differentiation Theorem
- given f(x), a continuous periodic function with piecewise continuous derivative f’(x)
- the Fourier series for f’(x) can be obtained by term by term differentiation of the Fourier Series of f(x)
Fourier Series
Convergence Theorem
-let f(x) be a piecewise smooth periodic function
-then, for all x, we have the pointwise convergence:
1/2[f(x+)+f(x-)] = ao/2 + Σ[ancos(nπx/L)+bnsin(nπx/L)]
-where an and bn are given by the Fourier coefficient formulae
-therefore, at a jump discontinuity, the Fourier series converges to the average value whilst at a point of continuity it just converges to the common value
Gibbs Phenomenon
- each finite expansion is an analytic function, so cannot truly represent a discontinuous function
- whilst the graph passes through the average at the precise point of discontinuity, the graph overshoots on either side
- this is known as the Gibbs phenomenon
Odd Function
Definition
-an odd function satisfies:
f(-x) = -f(x)
Even Function
Definition
-an even function satisfies:
f(-x) = f(x)
Odd or Even and Linear Combinations
-the property of being either even or odd is preserved under linear combinations
af(-x)+bg(-x) = ±(af(-x)+bg(-x))
Odd and Even Functions Under Multiplication
f(-x)g(-x) = {f(x)g(x) if f and g are both even or both odd OR -f(x)g(x) if one is odd and one is even}
Odd and Even Functions
Fourier Series
f(-x)=f(x) => bn=0 for all n => an = 1/L ∫ f(x)*cos(nπx/L) dx -integral btwn -1 and 1 => an = 2/L ∫ f(x)*cos(nπx/L) dx -integral btwn 0 and 1
f(-x)=-f(x) => an=0 for all n => bn = 1/L ∫ f(x)*sin(nπx/L) dx -integral btwn -1 and 1 => bn = 2/L ∫ f(x)*sin(nπx/L) dx -integral btwn 0 and 1
f(-x)=f(x) => f(x)cos(nπx/L) is even
f(-x)=-f(x) => f(x)sin(nπx/L) is odd
Even and Odd Parts of a Function
-given any function f(x), the even and odd parts are defined: fe(x) = 1/2[f(x)+f(-x)] fo(x) = 1/2[f(x)-f(-x)] -and we have: f(x) = fe(x) + fo(x)
What is a half-range Fourier series?
-in applications to separation of variables for partial differential equations we are often given a function f(x) on only the half-range 0
Even Extension
f(x) = ao/2 + Σancos((nπx/L)
-where the sum is between 1 and infinity and:
an = 2/L ∫ f(x)cos(nπx/L) dx
Odd Extension
f(x) = Σbnsin((nπx/L)
-where the sum is between 1 and infinity and:
bn = 2/L ∫ f(x)sin(nπx/L) dx