Is there a proof of the real form of Fourier's Theorem?

175 Views Asked by At

Complex form of Fourier's Theorem

Any periodic function can be decomposed into a linear combination of complex exponentials.

Proof

Consider a complex exponential with period $T_0$: $$e^{i\frac {2π}{T_0}t}$$ Hence, we have: $$\int^{T_0}_0 e^{i\frac {2π}{T_0}t} dt = 0 ……(1)$$ Let us claim that for a periodic function $x(t)$ we can write: $$x(t) ≈ \sum^{N}_{k=-N} C_ke^{i\frac {2πk}{T_0}t},$$ where $2N+1$ is the number of frequency components used. As $N→∞$, we have: $$x(t) = \sum^{∞}_{k=-∞} C_ke^{i\frac {2πk}{T_0}t} ……(2)$$ Consider: $$v_k(t):=e^{i\frac {2πk}{T_0}t}……(3)$$ Then we have: $$v_k(t+T_0)=v_k(t)$$ Hence, $v_k(t)$ is a periodic function. Now, $$\int^{T_0}_0 v_k(t)v^*_l(t) dt$$ $$=\int^{T_0}_0 e^{i\frac {2π(k-l)}{T_0}t} dt$$ $$=0 (k≠l)$$ $$T_0 (k=l)……(4)$$ Thus, $v_k(t)$ is orthogonal. Moreover, if we assume equation (2) to be valid , then we can multiply both its sides by $v_l^*(t)$ and integrate over $T_0$ we get: $$C_k=\frac {1}{T_0} \int^{T_0}_0 x(t)v_k^*(t) dt ……(5)$$ Putting this value in equation (2) we get back $x(t)$. Hence, any periodic function can be decomposed into a linear combination of complex exponentials.

Real form of Fourier's Theorem

An arbitrary periodic function $F(t)$ with period $T$ can be decomposed into a linear combination of the functions $f_n(t)$ and $g_n(t)$ where, $$f_n(t)=\sin \frac {2πnt}{T}$$ $$g_n(t)=\cos \frac {2πnt}{T}$$ Mathematically, $$F(t)=b_0 + b_1g_1(t) +b_2g_2(t) + …… + a_1f_1(t) +a_2f_2(t) + ……,$$ where $n$ is a non-negative integer and all of $a_i$,$b_i$ are real.

Problem

Is there a similar proof for the real form of Fourier's Theorem as for the complex form? I couldn't get one on the internet or in any book. Does this mean that it can be derived from the complex form? If yes then how? Any help would be appreciated.

1

There are 1 best solutions below

0
On BEST ANSWER

The functions $$ 1,\sin(2\pi x/T),\cos(2\pi x/T),\sin(4\pi x/T),\cos(4\pi x/T),\cdots $$ are mutually orthogonal. That is, if you choose any two different functions $f,g$ from the above, then $$ \int_{-T}^{T} f(x)g(x)dx = 0. $$ And, $$ \int_{-T}^{T}1^2dx=2T\\ \int_{-T}^{T}\sin^2(2\pi x/T)dx=T\\ \int_{-T}^{T}\cos^2(2\pi x/T)dx=T. $$ That's enough to carry about the same argument that you used before. It should be noted that your argument assumes convergence of the series, and does not prove it.