ODE: Even and Odd Function Extensions

Expanding a function defined only on a limited interval, like $[0,L]$, into a full Fourier series requires defining the function's behavior on the symmetric interval $[-L,L]$. The concepts of even and odd function extensions provide a powerful, systematic way to do this, leading directly to Fourier cosine series and Fourier sine series. These half-range expansions are not just mathematical curiosities; they are essential tools in engineering for solving partial differential equations that model physical phenomena like heat distribution, wave propagation, and signal processing, where boundary conditions dictate the natural form of the solution.

Core Concept 1: Symmetry and Fourier Series

A function $f(x)$ is even if $f(-x)=f(x)$ for all $x$ in its domain, meaning its graph is symmetric about the $y$-axis. A function is odd if $f(-x)=-f(x)$, meaning its graph is symmetric about the origin. The standard Fourier series representation on $[-L,L]$ contains both cosine and sine terms. However, the integrals for the coefficients simplify dramatically when the function possesses symmetry.

For an even function on $[-L,L]$, the product $f(x)\sin (n\pi x/L)$ is odd, so its integral over the symmetric interval is zero: $b_n=0$. The Fourier series reduces to a Fourier cosine series: $$f(x)\sim \frac{a_0}{2}+\sum _{n=1}^{\infty }{a}_n\cos \left(\frac{n\pi x}{L}\right)$$ with coefficients given by: $$a_n=\frac{2}{L}{\int }_0^{L}f(x)\cos \left(\frac{n\pi x}{L}\right)dx,\text{for }n=0,1,2,...$$

Conversely, for an odd function, the product $f(x)\cos (n\pi x/L)$ is odd, forcing $a_n=0$. The result is a Fourier sine series: $$f(x)\sim \sum _{n=1}^{\infty }{b}_n\sin \left(\frac{n\pi x}{L}\right)$$ with coefficients: $$b_n=\frac{2}{L}{\int }_0^{L}f(x)\sin \left(\frac{n\pi x}{L}\right)dx,\text{for }n=1,2,3,...$$

Notice the critical factor: the coefficient integrals for both series are computed only over the interval $[0,L]$, with a multiplier of $2/L$ instead of $1/L$. This is the first glimpse of the computational efficiency symmetry provides.

Core Concept 2: Half-Range Expansions on [0, L]

In applied problems, you are often given a function $f(x)$ defined only on the interval $[0,L]$. To represent it using a Fourier series, you must extend it to the full interval $[-L,L]$. You have two canonical choices, leading to half-range expansions.

The even extension of $f(x)$ is defined as: $$f_{\text{even}}(x)=\{\begin{array}{ll}f(x)& \text{for }0\le x\le L,\\ f(-x)& \text{for }-L\le x<0.\end{array}$$ This creates an even function on $[-L,L]$. The Fourier series of $f_{\text{even}}(x)$ is the Fourier cosine series of the original $f(x)$ on $[0,L]$.

The odd extension of $f(x)$ is defined as: $$f_{\text{odd}}(x)=\{\begin{array}{ll}f(x)& \text{for }0\le x\le L,\\ -f(-x)& \text{for }-L\le x<0.\end{array}$$ This creates an odd function on $[-L,L]$. Its Fourier series is the Fourier sine series of the original $f(x)$ on $[0,L]$.

For example, consider $f(x)=x$ defined on $[0,1]$. Its even extension would create a "V" shape (absolute value function), yielding a cosine series. Its odd extension simply continues the line $y=x$ into negative territory, yielding a sine series. The two series will converge to $f(x)=x$ on the open interval $(0,1)$, but their behavior at the endpoints $x=0$ and $x=1$ will differ significantly, a point crucial for applications.

Core Concept 3: Choosing the Extension Based on Boundary Conditions

The choice between an even or odd extension is not arbitrary in boundary value problems; it is dictated by the boundary conditions (BCs) of the associated partial differential equation. This is where the engineering application becomes clear.

Consider the one-dimensional heat equation on a rod of length $L$. The temperature $u(x,t)$ satisfies $u_t=\alpha u_{xx}$.

• If the ends of the rod are insulated (Neumann condition: $u_x(0,t)=0$ and $u_x(L,t)=0$), the natural solution method involves a cosine series. This corresponds to an even extension of the initial temperature distribution $f(x)$. The derivative of an even function at $x=0$ is zero, satisfying the insulation condition.
• If the ends of the rod are held at zero temperature (Dirichlet condition: $u(0,t)=0$ and $u(L,t)=0$), the natural solution uses a sine series. This corresponds to an odd extension of the initial condition $f(x)$. An odd function is zero at $x=0$, and if we also perform an odd extension about $x=L$, it will force the function to zero at $x=L$ as well.

Choosing the extension that matches the boundary conditions ensures the series satisfies the BCs term-by-term, simplifying the solution process immensely.

Core Concept 4: Coefficient Computation and Symmetry Tricks

The formulas for $a_n$ and $b_n$ for half-range expansions are already simplified, integrating from $0$ to $L$. However, you can leverage the symmetry of the integrand within $[0,L]$ to simplify calculations further, especially for piecewise-defined functions.

The key is to recognize that $\cos (n\pi x/L)$ is even about $x=L/2$ for even $n$ and odd about $x=L/2$ for odd $n$, with similar (but phase-shifted) behavior for sine. A more general and powerful approach is to use the integral properties directly: $${\int }_0^{2L}g(x)dx={\int }_0^{L}g(x)dx+{\int }_0^{L}g(2L-x)dx$$

If your function $f(x)$ on $[0,L]$ has a specific symmetry, you can combine it with the symmetry of the sine or cosine kernel. For instance, if you are computing a cosine series for a function that is symmetric about $L/2$ (i.e., $f(L-x)=f(x)$), then the product $f(x)\cos (n\pi x/L)$ has predictable properties that can halve the integration interval. Always sketch the extended function and the kernel to identify these symmetries before integrating; it can turn a long, tedious integral into two short, simple ones.

Common Pitfalls

1. Misapplying the Coefficient Formula: The most frequent error is using the full-range coefficient formula $(1/L){\int }_{-L}^L...$ when computing a half-range expansion. Always remember: for half-range sine or cosine series, the formula is $(2/L){\int }_0^L...$. The factor of $2$ compensates for integrating over only half the period.

1. Ignoring Boundary Condition Compatibility: Choosing an odd extension for a problem with insulated (zero-derivative) boundaries will lead to a solution that does not satisfy the physical constraints, and vice-versa. Always let the boundary conditions guide your choice of extension. The rule of thumb: Dirichlet (fixed value) conditions suggest a sine series (odd extension), while Neumann (zero-derivative) conditions suggest a cosine series (even extension).

1. Confusing the Function with its Extension: The cosine and sine series both converge to the original function $f(x)$ on the open interval $(0,L)$. However, at the endpoints $x=0$ and $x=L$, they converge to the average of the limits of the extended periodic function. For an odd extension, this often forces the series to zero at $x=0$ and $x=L$. For an even extension, it preserves the function's value at $x=0$ but may cause a jump at $x=L$ if $f(L)\ne f(-L)$. Be clear about what your series represents: the periodic extension of $f_{\text{even}}(x)$ or $f_{\text{odd}}(x)$.

1. Overcomplicating the Integral: Before diving into integration by parts, check for symmetry in the product $f(x)\sin (...)$ or $f(x)\cos (...)$ on $[0,L]$. If the function is defined piecewise, splitting the integral at the point of symmetry (often $L/2$) can simplify the algebra significantly.

Summary

• Even and odd extensions allow you to represent a function defined on $[0,L]$ as either a pure Fourier cosine series (even extension) or a pure Fourier sine series (odd extension), collectively known as half-range expansions.
• The coefficient computations simplify to integrals over $[0,L]$ with a factor of $2/L$, leveraging the symmetry of the extended function.
• The choice between a cosine or sine series is critically determined by the boundary conditions of the physical problem being solved (e.g., Dirichlet conditions favor sine series, Neumann conditions favor cosine series).
• You can use symmetry properties within the interval $[0,L]$ to further simplify the computation of the coefficients $a_n$ and $b_n$.
• Always interpret the resulting series as representing the periodic extension of your chosen (even or odd) expansion, not just the original function on $[0,L]$, paying close attention to convergence behavior at the endpoints.