A Fourier Transform pair refers to a function in the time or spatial domain and its corresponding representation in the frequency domain. The Fourier Transform, denoted by `F(s)` or `\mathcal{F}\{f(x)\}`, decomposes a function into its constituent frequencies. Pre-computed pairs for common functions serve as reference identities, allowing for the transformation of complex signals and the solution of differential equations without performing the integration manually.
The relationship is often written using a double-arrow notation to indicate the transform pair.
| Time/Spatial Domain Function, f(x) | Frequency Domain Function, F(s) |
|---|---|
| `\delta(x)` (Dirac Delta) | `1` |
| `1` (Constant) | `\delta(s)` |
| `e^{-a|x|}` (Exponential Decay) | `\frac{2a}{a^2 + 4\pi^2 s^2}` |
| `e^{-x^2/a^2}` (Gaussian) | `a\sqrt{\pi}e^{-\pi^2 a^2 s^2}` |
| `\cos(ax)` | `\frac{1}{2}[\delta(s - a/2\pi) + \delta(s + a/2\pi)]` |
| `\sin(ax)` | `\frac{1}{2i}[\delta(s - a/2\pi) - \delta(s + a/2\pi)]` |
| `\text{sinc}(x) = \frac{\sin(\pi x)}{\pi x}` | Rectangular Pulse (`\text{rect}(s)`) |
| Rectangular Pulse (`\text{rect}(x)`) | `\text{sinc}(s)` |
A common visualization shows a function in the time domain (x-axis) and its corresponding representation in the frequency domain (s-axis). For instance, a pure cosine wave in the time domain, which oscillates indefinitely, is represented in the frequency domain by two infinitely sharp peaks (Dirac delta functions). These peaks are located at the positive and negative frequencies that constitute the wave.
| Property | Time Domain | Frequency Domain |
|---|---|---|
| Linearity | `af(x) + bg(x)` | `aF(s) + bG(s)` |
| Time Scaling | `f(ax)` | `\frac{1}{|a|}F\left(\frac{s}{a}\right)` |
| Time Shifting | `f(x-a)` | `e^{-2\pi ias}F(s)` |
| Frequency Shifting (Modulation) | `e^{2\pi iax}f(x)` | `F(s-a)` |
| Differentiation | `\frac{d^n}{dx^n} f(x)` | `(2\pi is)^n F(s)` |
| Convolution | `(f * g)(x)` | `F(s)G(s)` |
| Multiplication | `f(x)g(x)` | `(F * G)(s)` |
Let's prove the time-shifting property, which states that a shift in the time domain corresponds to a phase shift in the frequency domain.
Step 1: Start with the definition of the Fourier Transform applied to the shifted function `f(x-a)`.
Step 2: Use a change of variables. Let `u = x - a`, which implies `x = u + a` and `du = dx`. The limits of integration remain unchanged.
Step 3: Separate the exponential term using the property `e^{A+B} = e^A e^B`.
Step 4: The term `e^{-2\pi i sa}` is a constant with respect to the integration variable `u`, so it can be factored out of the integral.
Step 5: Recognize that the remaining integral is the definition of the Fourier Transform of `f(u)`, which is `F(s)`. This completes the proof.
Fourier Transform pairs are fundamental tools in many scientific and engineering disciplines for analyzing signals and systems in the frequency domain.
Magnetic Resonance Imaging (MRI) machines collect data in the frequency domain, known as k-space. This raw data represents the Fourier transform of the actual patient anatomy. Powerful computers then perform a 2D or 3D inverse Fourier transform on this k-space data to reconstruct the detailed cross-sectional images used by doctors for diagnosis.
When you adjust the bass or treble on a stereo system, you are manipulating the frequency content of the audio signal. The equalizer is essentially a set of filters that boost or cut specific frequency bands. The entire process relies on the principle that any complex sound can be broken down into a sum of simple sine waves, which is the core idea of the Fourier Transform.
Engineers analyze vibrations in structures like bridges or aircraft wings to ensure their safety and stability. Sensors measure the structure's movement over time. By taking the Fourier Transform of this time-domain data, engineers can identify the structure's resonant frequencies—the specific frequencies at which it vibrates most intensely. This is crucial for avoiding catastrophic failure due to resonance.
The concept of the Fourier Transform adapts to different types of signals. The pairs listed here are for the standard continuous Fourier Transform. Other related transforms are used for periodic or discrete signals.
| Transform Type | Input Signal Characteristics | Output Spectrum Characteristics |
|---|---|---|
| Fourier Transform (FT) | Continuous and aperiodic | Continuous and aperiodic |
| Fourier Series (FS) | Continuous and periodic | Discrete and aperiodic |
| Discrete-Time Fourier Transform (DTFT) | Discrete and aperiodic | Continuous and periodic |
| Discrete Fourier Transform (DFT) | Discrete and periodic | Discrete and periodic |
The Discrete Fourier Transform (DFT) is the most important for practical applications, as it is the form that can be computed by a computer, typically using the highly efficient Fast Fourier Transform (FFT) algorithm.
Inconsistent Definitions: Different fields use different conventions for the Fourier Transform, primarily in the placement of the `2π` term (e.g., using angular frequency `ω = 2πs` vs. ordinary frequency `s`). Using a transform pair from a physics textbook with a formula from an engineering textbook can lead to scaling errors. Always check the definition being used.
Confusing Duality: The transform has a beautiful symmetry (duality), where the shape of a function in the time domain determines the shape in the frequency domain and vice-versa (e.g., a rect becomes a sinc, and a sinc becomes a rect). However, it's easy to forget the scaling factors and sign changes that may occur when applying the duality property.
Misinterpreting the Dirac Delta Function: The `δ` function in frequency domain pairs (like for `cos(ax)`) represents an infinitely narrow, infinitely tall spike at a specific frequency. It is not a standard function and should be treated as a spectral line representing a pure tone, not a value that can be simply plugged into other equations.