Heat Kernel Explained with Worked Examples

The heat kernel solves for ; it is the temperature that results from a unit of heat released at at time 0. The solution for any initial profile is .

Write 2x or 2*x to multiply, ^ for powers, and ; between simultaneous equations. You can also type factor, expand, derivative of, integrate … from a to b, limit … as x->0 or 15% of 80.

Try an example

Before you start

This page is easier if you are comfortable with:

  • the heat equation
  • the Gaussian function
  • square roots and exponentials

What Heat kernel (fundamental solution) means

It is a normal density with variance (the normal density), with total area 1 for every (the Gaussian integral). For , , the height at is , and at it is . For , , : , the value of the standard normal density at 1.

The peak falls like and the width grows like : diffusion spreads a distance proportional to the square root of time. Check that it solves the equation: with , the combination vanishes (see the heat equation). The kernel arises as the limit of the series solution as the rod becomes infinite.

Brownian motion has transition density exactly with (Brownian motion), the link between the heat equation, random walks and the Fokker–Planck equation. The solution formula is a convolution (the convolution theorem) and the Black–Scholes equation reduces to it (the Black–Scholes PDE).

What is it used for?

The heat kernel is the temperature profile produced by a single burst of heat; it is a spreading Gaussian.

How to use heat kernel (fundamental solution)

  1. Identify , and .
  2. Compute .
  3. Compute .
  4. Divide.

Graph

Heat kernel at t = 0.25 and t = 1 (k = 1)peak 0.5642x = 1: 0.2076-4-3-2-10123400.10.20.30.40.50.6Heat kernel at t = 0.25 and t = 1 (k = 1)xΦt = 0.25t = 1Graph with x from -4 to 4 and y from 0 to 0.65. Shows the curve t = 0.25; the curve t = 1; point peak 0.5642 at (0, 0.5642); point x = 1: 0.2076 at (1, 0.2076).

The kernel gets lower and wider as time goes on, keeping area 1.

Worked examples

These examples are worked by hand, step by step, so you can compare your own working.

Example 1: centre value

Find for .

Worked example
Answer
0.56420.5642
  1. Formula

Example 2: off centre

Find for .

Worked example
Answer
0.20760.2076
  1. Formula

Example 3: k = 0.5

Find for .

Worked example
Answer
0.24200.2420
  1. Formula

Special cases

A few situations come up again and again. Knowing them saves time and helps you spot answers that cannot be right.

k = 1, x = 0, t = 1

Φ = 1/√(4π)

0.2821.

x = 1, t = 1

Φ = 0.2821 e^(−1/4)

0.2197.

Smaller t = 0.25

x = 0: 1/√π

0.5642 (sharper peak).

Area

∫Φ dx = 1

Heat is conserved.

Word problems

Φ(x, t) = e^(−x²/(4kt))/√(4πkt) solves u_t = ku_xx. The solution with initial data f is u = ∫Φ(x − y, t)f(y)dy.

Peak

Find Φ(0, 1) for k = 1.

  1. 1/√(4π)
  2. Check: ✓0.2821

Off centre

Find Φ(1, 1).

  1. 0.2821 × e^(−0.25)
  2. Check: ✓0.2197

Earlier time Higher

Find Φ(0, 0.25).

  1. 1/√π
  2. Check: ✓0.5642

Conservation Higher

What is ∫Φ dx for any t?

  1. The total heat
  2. Check: 1 ✓1

Common mistakes

Using 2kt

x²/(2kt) ✗

x²/(4kt) ✓

Heat equation has 4kt.

Forgetting normalisation

dropping 1/√(4πkt) ✗

keep it ✓

Area must be 1.

t = 0

evaluate directly ✗

limit is δ(x) ✓

Singular at t = 0.

Spreading

narrowing with time ✗

widening with time ✓

Variance 2kt grows.

Exam tips

  1. Peak 1/√(4πkt).
  2. Width grows like √t.
  3. Area stays 1.
  4. Convolve with f.
  5. t > 0 only.

Practice questions

Try each one on paper, then open the answer. Use the solver above to see the full working.

1. Φ(0, 1) for k = 1 to 4 d.p. Foundation

0.2821

2. Φ(0, 0.25) to 4 d.p. Foundation

0.5642

3. Φ(2, 1) to 4 d.p. Foundation

0.1038

4. Φ(0, 4) to 4 d.p. Higher

0.141

5. Φ(1, 0.25) to 4 d.p. Higher

0.2076

6. Variance 2kt for k = 1, t = 3 Higher

6

Practice generator

Unlimited fresh questions with instant marking. Choose a level, work it out on paper, then type your answer.

Solve: …

Key terms

Heat kernel
Fundamental solution.
Gaussian
Bell curve.
Convolution
Integral of Φ(x − y)f(y).
Diffusivity (k)
Spreading rate.
Variance 2kt
Spread of the kernel.
Delta function
Limit as t → 0.

Frequently asked questions

What is the heat kernel?

Φ = e^(−x²/(4kt))/√(4πkt).

What does it represent?

Temperature from a point source of unit heat.

How do I use it?

Convolve it with the initial temperature.

What is its total area?

1.

Which page covers the heat equation?

heat equation.

Which page covers characteristics?

method of characteristics.

More in Differential equations and transforms and the formula sheets.