STRATA
Skip to study
XLVII

XLVII / Experiments

Workbench

Change a condition, inspect what happens, and keep the result. Six experiments connect an equation with a visible outcome, a numerical record, and questions you can pursue.

Sampling and aliasing

What motion could have produced these samples?

A sequence of measurements contains values at particular instants, not the entire motion between them. Draw a sinusoid, sample it for one second, and compare it with a lower-band sinusoid that agrees at every sampled instant. Change the sampling rate to see when the two curves separate.

The curves agree at the sample times.

Original signalFolded candidate
Normalized amplitude-1.2-0.600.61.200.250.50.751Time (seconds)
The cream curve is the 6 Hz signal. The orange candidate has signed frequency 6 Hz. Dots mark 33 samples, including both endpoints.
Nyquist frequency
16.0 Hz

Half the sampling rate.

Folded frequency
6.00 Hz

The lower-band candidate at these sample times.

Samples per cycle
5.33

32 samples per second / 6 Hz.

Sample disagreement
0.0e+0

Largest absolute difference; roundoff only.

The table shows the first eight samples and the endpoint. The CSV includes every sampled instant.
Time (s)Original signalFolded candidate
0.00000.000000.00000
0.03130.923880.92388
0.06250.707110.70711
0.0938-0.38268-0.38268
0.1250-1.00000-1.00000
0.1563-0.38268-0.38268
0.18750.707110.70711
0.21880.923880.92388
1.0000-0.00000-0.00000

Keep a reproducible record.

The JSON includes parameters, model assumptions, sources, and full results. The CSV contains all sample times, original values, and folded-candidate values.

Method

The calculation follows these rules.

For a sampling rate fₛ, adding an integer multiple of fₛ to a sinusoid’s frequency leaves its samples unchanged. Fold the signed frequency into [−fₛ/2, fₛ/2) and evaluate both curves at identical sample times.

x(t) = sin(2πft + φ); f_alias = ((f + fₛ/2) mod fₛ) − fₛ/2.

Read the result within its assumptions.

  • The input is one noiseless sinusoid. No antialiasing filter, interpolation algorithm, timing jitter, or measurement noise is modeled.
  • The signed candidate preserves phase and sample values. Its displayed magnitude is an apparent frequency, not proof that it was the original source.
  • Unique recovery requires a suitable bandlimit and sampling above twice the highest included frequency; the exact boundary can lose phase information.

Take the experiment further.

01

Trace the ambiguity.

Load the folded example and compare the two curves only at the dots. Then inspect what each curve does between those dots.

02

Move the boundary.

Keep the signal at 18 Hz and increase the sampling rate past 36 samples per second. Identify the setting where the candidate stops folding.

03

Test a silent sample sequence.

Use the Nyquist-boundary preset. Compare phases of 0°, 45°, and 90° while keeping the signal frequency unchanged.

Read the sources behind the model.

  1. MIT 6.003, Sampling

    Dennis Freeman explains sampling, aliasing, reconstruction, and antialiasing filters in the Signals and Systems course.

Watch one model run.

12 seconds · Original model visualization · Silent

Sampling and aliasing

See how different continuous signals can pass through the same samples.

Read the explanation and method

The cream curve is an 18 hertz sinusoid. Orange dots show measurements taken at the displayed sampling rate.

The orange curve is a signed lower-band candidate that passes through the same sample values. Agreement at the dots cannot establish what happened between them.

The sampling rate rises above 36 samples per second. With the stated bandlimit, sampling strictly above twice the signal frequency removes this ambiguity.

Method. For a sampling rate fₛ, adding an integer multiple of fₛ to a sinusoid’s frequency leaves its samples unchanged. Fold the signed frequency into [−fₛ/2, fₛ/2) and evaluate both curves at identical sample times.

Apply the model to a research question.

These dossiers explain how this experiment relates to a larger question, including the observation or comparison that the model leaves unresolved.

Try another experiment.