Maths Formulae Transforms Convolution

Convolution Theorem – Frequency and Time Domain Relationship

Learn how the convolution theorem links multiplication in the frequency domain with convolution in the time domain.
🔑

Definition of Convolution

The Laplace Convolution Theorem establishes that convolution in the time domain corresponds to simple multiplication in the s-domain. This fundamental relationship is the mathematical foundation for analyzing linear time-invariant (LTI) systems, where the system output is the convolution of the input with the system's impulse response. It transforms complex integral operations into straightforward algebraic multiplication, making system analysis and design dramatically simpler and more intuitive.

SymbolDescription
\[ f(t), g(t) \]Time-domain functions being convolved
\[ f(t) * g(t) \]The convolution of f(t) and g(t)
\[ F(s), G(s) \]The Laplace transforms of f(t) and g(t) respectively
\[ F(s) \cdot G(s) \]Simple multiplication of the functions in the s-domain
\[ H(s) \]The transfer function of a system, where H(s) = L{h(t)}, the Laplace transform of the impulse response
\[ \tau \]A dummy variable of integration for the convolution integral
\[ * \]The convolution operator
🔢

Key Formulas

\[ f(t) * g(t) = \int_{0}^{t} f(\tau) g(t-\tau) d\tau \]
Convolution Integral
\[ \mathcal{L}\{f(t) * g(t)\} = F(s) \cdot G(s) \]
Laplace Convolution Theorem
\[ \mathcal{L}^{-1}\{F(s) G(s)\} = f(t) * g(t) \]
Inverse Convolution Theorem
\[ \text{If } y(t) = x(t) * h(t), \text{ then } Y(s) = X(s) H(s) \]
System Response Formula
📊

System Diagram

Laplace Convolution Theorem f(t) ∗ g(t) Time domain F(s) · G(s) s-domain (multiply) Key Advantage Convolution in time → Multiplication in the s-domain ℒ{f∗g} = F(s)·G(s), (f∗g)(t) = ∫₀ᵗ f(τ)g(t−τ)dτ
Laplace Convolution: time-domain convolution f∗g becomes simple multiplication F(s)·G(s) in the s-domain

A block diagram for a Linear Time-Invariant (LTI) system illustrates the convolution principle. In the time domain, an input signal x(t) enters a block representing the system, characterized by its impulse response h(t). The resulting output is y(t) = x(t) * h(t). In the frequency (s) domain, the same system is represented by its transfer function H(s). The input X(s) is simply multiplied by H(s) to produce the output Y(s), demonstrating how convolution becomes simple multiplication.

⚖️

Properties of Convolution

Convolution is a mathematical operation with several important algebraic properties that mirror those of multiplication.

\[ f * g = g * f \]
Commutative Property
\[ f * (g * h) = (f * g) * h \]
Associative Property
\[ f * (g + h) = (f * g) + (f * h) \]
Distributive Property
\[ a(f * g) = (af) * g = f * (ag) \]
Scalar Multiplication Associativity
📝

Proof of the Convolution Theorem

The proof of the Laplace Convolution Theorem, \( \mathcal{L}\{f(t) * g(t)\} = F(s)G(s) \), begins with the definition of the Laplace transform applied to the convolution integral.

\[ \mathcal{L}\{f(t) * g(t)\} = \int_0^\infty e^{-st} \left[ \int_0^t f(\tau)g(t-\tau)d\tau \right] dt \]
Step 1: Apply the Laplace Transform definition

Next, we change the order of integration. The region of integration is over \( 0 \le \tau \le t \) and \( 0 \le t < \infty \). This is equivalent to integrating over \( \tau \le t < \infty \) and \( 0 \le \tau < \infty \).

\[ = \int_0^\infty f(\tau) \left[ \int_\tau^\infty e^{-st} g(t-\tau)dt \right] d\tau \]
Step 2: Change the order of integration

Perform a change of variable in the inner integral. Let \( u = t - \tau \), which means \( t = u + \tau \) and \( du = dt \). When \( t = \tau \), \( u = 0 \). When \( t \to \infty \), \( u \to \infty \).

\[ = \int_0^\infty f(\tau) \left[ \int_0^\infty e^{-s(u+\tau)} g(u)du \right] d\tau \]
Step 3: Substitute u = t - τ

Separate the exponential term and rearrange the integrals.

\[ = \int_0^\infty f(\tau) e^{-s\tau} \left[ \int_0^\infty e^{-su} g(u)du \right] d\tau \]
Step 4: Separate the exponential

The inner integral is, by definition, the Laplace transform of \( g(t) \), which is \( G(s) \). Since \( G(s) \) is not a function of \( \tau \), it can be moved outside the outer integral.

\[ = \left[ \int_0^\infty e^{-su} g(u)du \right] \int_0^\infty f(\tau) e^{-s\tau} d\tau = G(s) \cdot F(s) \]
Step 5: Recognize the definitions of F(s) and G(s)

This completes the proof, showing that the Laplace transform of a convolution of two functions is the product of their individual Laplace transforms.

✍️

Worked Example

Find the convolution of \( f(t) = e^{-t} \) and \( g(t) = e^{-2t} \) using the convolution theorem.
  1. First, find the Laplace transforms of f(t) and g(t). \( F(s) = \mathcal{L}\{e^{-t}\} = \frac{1}{s+1} \) and \( G(s) = \mathcal{L}\{e^{-2t}\} = \frac{1}{s+2} \).
  2. According to the convolution theorem, \( \mathcal{L}\{f(t) * g(t)\} = F(s)G(s) \). Multiply the two transforms: \( F(s)G(s) = \frac{1}{s+1} \cdot \frac{1}{s+2} = \frac{1}{(s+1)(s+2)} \).
  3. To find the time-domain result, we need to take the inverse Laplace transform of the product. Use partial fraction expansion: \( \frac{1}{(s+1)(s+2)} = \frac{A}{s+1} + \frac{B}{s+2} \).
  4. Solving for A and B gives A = 1 and B = -1. So, \( F(s)G(s) = \frac{1}{s+1} - \frac{1}{s+2} \).
  5. Take the inverse Laplace transform of each term: \( \mathcal{L}^{-1}\left\{ \frac{1}{s+1} - \frac{1}{s+2} \right\} = \mathcal{L}^{-1}\left\{ \frac{1}{s+1} \right\} - \mathcal{L}^{-1}\left\{ \frac{1}{s+2} \right\} \).
\[ f(t) * g(t) = e^{-t} - e^{-2t} \]
🚀

Applications of Convolution

🎛️ Control System Analysis

The convolution theorem is fundamental to control theory. It allows engineers to determine a system's output for any given input by simply knowing the system's impulse response. By transforming to the s-domain, the complex convolution operation becomes a simple multiplication of the input's transform and the system's transfer function, H(s).

⚡ Circuit Analysis and Design

In electrical engineering, the theorem is used to analyze the response of RLC circuits to various inputs like step or impulse voltages. The circuit's behavior is characterized by a transfer function, and the output voltage or current is found by multiplying this function with the input's Laplace transform.

📡 Signal Processing Systems

Convolution is the core of digital and analog filtering. A filter's effect on a signal is described by convolving the input signal with the filter's impulse response. In the frequency domain, this corresponds to multiplying the signal's spectrum by the filter's frequency response, which simplifies the design and analysis of filters for tasks like noise reduction or equalization.

🏗️ Mechanical System Dynamics

In mechanics and structural engineering, the theorem is used to analyze vibrations and dynamic responses. The response of a structure (like a bridge or building) to a time-varying force (like wind or an earthquake) can be calculated by convolving the force with the structure's impulse response, simplifying complex dynamic analysis.

🌍

Real-World Examples

An RC circuit with R = 1 MΩ and C = 1 μF has an impulse response of \( h(t) = e^{-t} \). If a unit step voltage, \( x(t) = u(t) \), is applied at t=0, what is the output voltage \( y(t) \)?
  1. Find the Laplace transforms of the input and impulse response. For the input step function, \( X(s) = \mathcal{L}\{u(t)\} = \frac{1}{s} \). For the impulse response, \( H(s) = \mathcal{L}\{e^{-t}\} = \frac{1}{s+1} \).
  2. In the s-domain, the output Y(s) is the product of the input and the transfer function: \( Y(s) = X(s)H(s) = \frac{1}{s} \cdot \frac{1}{s+1} \).
  3. Use partial fraction expansion to simplify Y(s): \( \frac{1}{s(s+1)} = \frac{1}{s} - \frac{1}{s+1} \).
  4. Find the inverse Laplace transform to get the output voltage in the time domain: \( y(t) = \mathcal{L}^{-1}\left\{ \frac{1}{s} - \frac{1}{s+1} \right\} = u(t) - e^{-t}u(t) \).
The output voltage across the capacitor is \( y(t) = (1 - e^{-t})u(t) \) Volts.
An audio low-pass filter has a transfer function \( H(s) = \frac{100}{s+100} \). An input signal \( x(t) = 5e^{-10t} \) is passed through it. Find the output signal \( y(t) \).
  1. Find the Laplace transform of the input signal: \( X(s) = \mathcal{L}\{5e^{-10t}\} = \frac{5}{s+10} \).
  2. Multiply the input transform by the transfer function to find the output transform: \( Y(s) = X(s)H(s) = \frac{5}{s+10} \cdot \frac{100}{s+100} = \frac{500}{(s+10)(s+100)} \).
  3. Apply partial fraction expansion: \( \frac{500}{(s+10)(s+100)} = \frac{A}{s+10} + \frac{B}{s+100} \). Solving gives \( A = 50/9 \) and \( B = -50/9 \).
  4. The output transform is \( Y(s) = \frac{50/9}{s+10} - \frac{50/9}{s+100} \).
  5. Take the inverse Laplace transform to find the output signal in the time domain.
The output signal is \( y(t) = \frac{50}{9}(e^{-10t} - e^{-100t}) \).
🏙️

Real-World Scenarios

speed response to step
Motor Speed Control
A motor's speed response to a voltage input is the convolution of input with the motor's impulse response. In the Laplace domain: Ω(s) = G(s)·V(s) — simple multiplication makes controller design tractable.
τ reverb = convolution
Audio Reverb & Echo
Echo effects in music are Laplace convolution: the output is the dry signal convolved with a decaying impulse train. In s-domain, Y(s) = X(s)·H(s) — multiplication instead of convolution integration.
T(x,t) = T₀ * Gaussian heat diffusion = convolution
Heat Conduction
Heat spreading through a material is a convolution: temperature distribution is the initial profile convolved with a Gaussian kernel that widens over time. Laplace transforms make the heat diffusion PDE algebraically solvable.

Image Processing

In digital photography and computer vision, convolution is used to apply filters to images. A 'kernel' (a small matrix of numbers, analogous to an impulse response) is convolved with the image's pixel data to achieve effects like blurring (e.g., a Gaussian kernel), sharpening, or edge detection. The entire operation modifies each pixel based on the values of its neighbors.

Audio Engineering

Reverb effects in music production are created using convolution. An impulse response is recorded in a real acoustic space (like a concert hall or a cave). To apply that reverb to a dry audio recording (like a vocal track), the vocal signal is convolved with the hall's impulse response, making it sound as if the singer was performing in that hall.

Economics and Finance

Moving averages, a common tool for smoothing out volatile data like stock prices, are a form of convolution. The raw price data is convolved with a simple function (e.g., a rectangular window) to calculate the average over a specific time period. This helps analysts identify underlying trends more easily.

📂

Types and Classifications

Continuous-Time vs. Discrete-Time Convolution

The convolution integral \( \int f(\tau)g(t-\tau)d\tau \) applies to continuous signals, such as analog voltages or physical motion. In digital systems (computers, smartphones), signals are discrete samples. For these, discrete convolution is used, which replaces the integral with a summation: \( y[n] = \sum_{k=-\infty}^{\infty} x[k]h[n-k] \). The convolution theorem has a parallel in the discrete domain, where the Discrete Fourier Transform (DFT) of a convolution is the product of the individual DFTs.

⚠️

Common Mistakes

⚠️ Confusing Convolution with Multiplication in the Time Domain: A very common error is to think that \( \mathcal{L}\{f(t)g(t)\} \) is \( F(s)G(s) \). This is incorrect. The product of transforms \( F(s)G(s) \) corresponds to the convolution \( f(t) * g(t) \) in the time domain, not simple multiplication.
⚠️ Incorrectly Setting Up the Integral: When forced to compute the convolution integral directly, students often make mistakes with the limits of integration or the \( g(t-\tau) \) term. The convolution theorem is a powerful tool to avoid these complex and error-prone integral calculations entirely.
💡 Memory Trick: "CONVOLUTION = CONVert Operations Leaving Ultra-Tough Integration Operations Null." Think of the theorem as a way to make difficult integrals disappear by converting them to simple algebra in the s-domain.
🚀

Study Strategy

1 🔍 Grasp Core Concepts
  • Focus on the 'Definition of Convolution' section to understand the integral as a 'blending' or 'weighted averaging' operation.
  • Use the 'System Diagram' to visualize how an input signal `x(t)` interacts with a system's impulse response `h(t)` to produce an output `y(t)`.
  • Review the 'Properties of Convolution' (Commutative, Associative, Distributive) and understand how they simplify complex problems.
  • Differentiate between the continuous-time integral and the discrete-time summation formulas, noting their structural similarity.
2 🧠 Commit Formulas to Memory
  • Write out the primary convolution integral `y(t) = ∫x(τ)h(t-τ)dτ` multiple times until it becomes second nature.
  • Memorize the Convolution Theorem: `convolution in the time domain is multiplication in the frequency domain` (`y(t) = x(t) * h(t) ↔ Y(F) = X(F)H(F)`).
  • Create flashcards for the key properties, linking the name (e.g., Commutative) to its formula (`x*h = h*x`).
  • Practice reciting the formula for convolution with a Dirac delta function, `x(t) * δ(t-T) = x(t-T)`, as it's a common and important case.
3 ✍️ Practice with Worked Examples
  • Replicate the 'Worked Example' on your own, then compare your solution to the provided one, analyzing any discrepancies.
  • Use the graphical 'flip and slide' method to solve a convolution problem involving simple rectangular or triangular pulses.
  • Solve a problem both in the time domain (using the integral) and the frequency domain (using the Convolution Theorem) to verify your answer.
  • Work through problems listed under 'Common Mistakes' to actively avoid these pitfalls in your own calculations.
4 🌍 Apply to Real-World Problems
  • Select one topic from 'Applications of Convolution' (e.g., image processing) and explain how the input, system, and output relate.
  • Analyze a 'Real-World Scenario', like audio reverberation, by describing the input signal (dry sound) and the impulse response (room acoustics).
  • Consider how a different impulse response (e.g., a sharpening kernel vs. a blurring kernel in an image) changes the output.
  • Attempt to formulate a simple, real-world problem yourself, such as calculating the daily average temperature from hourly readings, and identify the convolution-like process.
By systematically understanding the concept, memorizing the tools, practicing the mechanics, and applying the logic, you can master convolution.

Frequently Asked Questions

×

×