EART60071 Chap.2 Forward Euler, Convergence and Numerical Error
Forward Euler, Convergence and Numerical Error
Forward Euler, Convergence and Numerical Error covers the first practical's central lesson: running a model involves far more than placing its equation on a computer. Starting from radioactive decay, the chapter replaces the derivative by a finite difference and obtains a stepping rule that multiplies the current value by one minus the decay constant times the step.
It separates the plotting interval, which only samples the exact curve, from the numerical time-step, which changes the answer. It explains convergence, why an oversized step can produce a physically impossible negative count, and why the same method makes a harmonic oscillator's amplitude grow, so that you can diagnose step-size problems before running a model.
The practical asks you to explain each of its runs, so the chapter keeps the arithmetic short and the interpretation central: what the factor predicts, what the plot shows and why the behaviour belongs to the method rather than to the decay or oscillation being modelled.
What this chapter covers
- 01
The decay equation and its exact solution
- 02
Deriving the Forward Euler update
- 03
Plotting interval versus numerical time-step
- 04
Convergence and its computational cost
- 05
Negative values from an oversized step
- 06
Growing amplitude in the oscillator
Worked example · free
Three Forward Euler steps against the exact decay curve
- 1The one-step factor is 1 − λΔt = 1 − 100/500 = 0.8, so each step removes 20% of the atoms.
- 2Steps: 8.0 → 6.4 → 5.12 → 4.096 × 1018 atoms at 300 s.
- 2Exact: 8.0 × 1018 × e−0.6 = 4.39 × 1018. The error is (4.096 − 4.390)/4.390 = −6.7%.
Key terms
- Time-step
- The finite interval by which a numerical method advances its solution in time.
- Analytical solution
- An exact formula for the solution of a differential equation, against which numerical results can be checked.
- Harmonic oscillator
- A system whose acceleration is proportional to its displacement and opposite in sign, giving constant-amplitude oscillation.
Forward Euler, Convergence and Numerical Error FAQ
Why does a smaller time-step give a more accurate answer?
Forward Euler assumes the gradient at the start of each interval holds across the whole interval. The shorter the interval, the less the true gradient changes within it, so each step's error shrinks and the numerical solution approaches the exact one. That approach is convergence, and its price is more steps and therefore more computation.
Why did the practical's largest step give a negative number of atoms?
Each step multiplies the count by one minus the decay constant times the step. With a 10,000 s step and a decay constant of 1/2000 per second that factor is minus four, so a single step flips the sign. The physics has not changed; the step is far too long for the decay timescale.
Why does the oscillator's amplitude grow with Forward Euler?
The exact oscillator keeps a constant amplitude, but each Forward Euler step multiplies it by the square root of one plus omega squared times the step squared. With a half-second step the growth is visible within a few cycles; cutting the step to 0.05 s makes it far smaller, which is again convergence at work.
How is Forward Euler written in the practical script?
The decay script applies the update that comes straight from replacing the derivative with a forward difference: the new value equals the old value minus the time-step times the decay constant times the old value. Searching the file for the labelled Forward Euler step finds that line, and the practical asks you to match it to the equation and compute the first step by hand.
Exam move
Practise this chapter with a pencil before a computer. Write the decay equation, derive the Forward Euler update in two lines, and take three steps by hand for a few combinations of decay constant and step, always writing the factor first. Compare each with the exponential and express the error as a percentage, then halve the step and confirm the error falls.
Repeat with the oscillator, tracking the quantity y squared plus v over omega squared to see the amplitude grow. Next, explain in one sentence each the difference between the plotting interval and the time-step, and why an analytical solution has no step at all.
Finally, look back at your practical plots for the large-step runs and write the physical impossibility each shows and the numerical cause, because the assessment can ask you to interpret exactly these experiments.