Why is mathematics important in radio interferometry? Because an array of antennas does not take a conventional photograph. Each antenna pair measures a complex correlation—a visibility—associated with a projected baseline and wavelength. As Earth rotates, the array samples more spatial frequencies. Fourier mathematics, calibration and inverse-problem reasoning turn those samples into evidence about the sky.
This article is educational, not an observing or data-reduction manual. Professional imaging requires current observatory documentation, instrument-specific calibration, validated software and scientific review. A reconstructed image is a model constrained by measurements, not a direct snapshot.
Quick Reading Route
- Start with why combine antennas.
- Use baselines and resolution for geometry.
- Read complex visibility for the measured quantity.
- Work through a two-source example.
- Study aperture synthesis and imaging limits.
- Finish with calibration and FAQs.
Why Combine Antennas
The angular scale associated with an aperture is roughly λ/D. Building one dish as wide as the longest desired baseline can be impractical. An interferometer correlates signals from separated antennas, gaining information associated with their separation.
For two antennas separated by baseline vector B and a source direction s, geometric delay is approximately τ=B·s/c relative to a chosen reference. Correcting delay aligns wavefront samples before correlation. As the source direction changes with Earth rotation, projected delay and sampled spatial frequency change.
Did You Know? A longer baseline can miss large structures
Long baselines sample fine angular scales, but an array also needs short spacings for extended emission. The NRAO VLA status summary explains that an interferometric configuration acts as a spatial filter and can miss structures larger than scales sampled by its shortest baseline. Resolution and completeness are different goals.
Correlation, not simple addition
The correlator compares voltages with timing and phase information. Uncorrelated receiver noise tends not to produce the same baseline correlation as a coherent celestial signal, though calibration and systematics remain. The output is typically complex and frequency-resolved.
Baselines, Wavelengths and Angular Scales
A physical baseline B becomes dimensionless coordinates in wavelengths. With components projected relative to the source, u=B_u/λ, v=B_v/λ and possibly w=B_w/λ. The same metre baseline samples more wavelengths at higher frequency.
Suppose B=1,000 m and λ=0.21 m. B/λ≈4,762 wavelengths. A rough fringe spacing λ/B≈2.1×10⁻⁴ rad≈43 arcsec. At λ=0.03 m, the same baseline gives about 6.2 arcsec. These are order-of-magnitude scales; array weighting and geometry shape the synthesised beam.
Projection matters
Only the component perpendicular to the source direction contributes to the transverse spatial frequency. A long ground baseline can have a small projected length for some geometry. Observation time and source declination therefore affect uv coverage.
Primary and synthesised beams
Each dish’s diameter controls a primary field response, while the array baselines determine the synthesised response. Confusing these leads to wrong claims about field of view and resolution. Wide-field imaging also requires accounting for non-coplanar baselines and direction-dependent effects.
NRAO’s field-of-view guidance discusses time and bandwidth smearing and the assumptions behind synthesis imaging. Averaging choices affect which off-axis signals remain faithful.
Complex Visibility: Amplitude and Phase
For a small field under a simplified flat-sky approximation, visibility V(u,v) is related to sky brightness I(l,m) by a Fourier transform:
V(u,v)=∬I(l,m)e^{-2πi(ul+vm)} dl dm.
Visibility is complex: V=Ae^{iφ}=X+iY. Amplitude carries contrast on a sampled spatial scale; phase carries positional/asymmetry information relative to the phase centre and conventions.
A point source
A point of flux S at direction (l0,m0) gives V=S e^{-2πi(ul0+vm0)} in the simplified model. Its amplitude stays S while phase rotates with baseline. A point at phase centre has constant zero phase after ideal calibration.
Extended source
An extended Gaussian transforms into a Gaussian visibility that declines on long baselines. This does not mean the source fades physically; fine spatial-frequency response is smaller because the source is smooth.
Hermitian symmetry
For real sky brightness, V(−u,−v)=V*(u,v). This conjugate symmetry is a useful consistency property. Measurement noise and calibration affect observed values, while gridding often uses both mirrored coordinates appropriately.
Worked Example: Two Sources and One Baseline
Consider two fictional equal point sources of strength S/2 at angular coordinates l=±α along one dimension. Their visibility is
V(u)=S/2 e^{-2πiuα}+S/2 e^{+2πiuα}=S cos(2πuα).
The visibility is real and oscillates between positive and negative values. Zeros occur when 2πuα=(π/2)+kπ, so uα=1/4+k/2.
Numerical scale
Let separation be 2α=20 arcsec, so α≈4.848×10⁻⁵ rad. The first zero occurs u≈1/(4α)≈5,157 wavelengths. At λ=0.21 m this corresponds to a projected baseline about 1,083 m.
One visibility zero does not by itself prove “two sources”; other brightness distributions can share samples. Broad uv coverage and model comparison constrain the interpretation.
Unequal sources
If strengths differ, the imaginary components no longer cancel generally, and phase contains asymmetry information. Shifting both sources together multiplies visibility by a phase ramp while preserving amplitude. This is the Fourier shift theorem in a physical setting.
Finite bandwidth
Across a channel, u=B/λ changes with frequency. Averaging wide bandwidth without modelling can wash out phase for off-centre sources. Narrow channels reduce smearing but increase data volume and per-channel noise.
Earth Rotation and Aperture Synthesis
As Earth rotates, a fixed physical baseline changes orientation relative to a celestial source. Its projected (u,v) coordinate traces a curve. Many antenna pairs and times build a sampling pattern that approximates a larger aperture.
NRAO’s Fundamentals of Radio Interferometry presents visibility and baseline ideas used in synthesis imaging. The mathematical attraction is clear: time converts a finite collection of dishes into many projected measurements.
Number of baselines
For N antennas, the number of distinct pairs is N(N−1)/2. Ten antennas yield 45 baselines; 27 yield 351. Adding antennas increases pair count quadratically, though sensitivity and layout depend on more than count.
Redundancy
Two pairs may sample similar uv coordinates. Redundancy can improve sensitivity or calibration, while diverse baselines improve coverage. Array design balances these aims.
Frequency synthesis
Different frequencies sample different radii because u and v scale as 1/λ. Multi-frequency synthesis can improve coverage, but sky brightness and instrument response may vary with frequency. A model must distinguish geometry from spectral behaviour.
The Dirty Image, Deconvolution and Missing Scales
Measured visibilities equal the full Fourier plane multiplied by a sampling/weighting function. Inverse transforming the sampled data produces a dirty image: the true sky convolved with the dirty beam under simplifying assumptions.
Sidelobes in the dirty beam can create patterns around real sources. They are not new celestial objects. Deconvolution algorithms model the sky and residuals to reduce sampling artefacts, but their outputs depend on assumptions and stopping rules.
Weighting
Natural weighting often favours sensitivity because densely sampled data receive substantial influence. Uniform weighting balances uv cells more evenly and can improve resolution or sidelobes at a sensitivity cost. Tapering downweights long baselines, producing a broader beam and greater emphasis on extended scales.
Missing short spacings
Without measurements near the uv origin, total or broad emission can be underestimated. Negative bowls or fragmented extended structures may appear. Combining interferometer data with single-dish or compact-array information can restore scales under a careful method.
Image cell size
Pixels should sample the synthesised beam adequately. Tiny pixels do not create real resolution and enlarge computation; large pixels undersample the beam. Field size and cell size also determine Fourier-grid extent and wrap-around risks.
Cleaning is not proof
An attractive restored image is one reconstruction. Inspect residuals, alternative weights, calibration tests and simulations. Scientific conclusions should survive reasonable processing choices.
Calibration, Noise and Uncertainty
Measured visibility on baseline ij can be modelled Vobs_ij=g_i g*_j Vtrue_ij + error, where each antenna has complex gain g. Calibration observes sources with known or modelled properties to estimate time-, frequency- and direction-dependent gains.
Amplitude and phase calibration
Amplitude errors rescale visibilities; phase errors blur or shift structure. Bandpass calibration addresses frequency response. Flux-density calibration establishes a scale. Polarisation and direction-dependent calibration add further layers.
Closure quantities
The sum of phases around a triangle cancels ideal antenna-based phase terms: arg(V12V23V31). Closure phase therefore contains source information robust to those simple antenna errors, though baseline-based errors remain. Closure amplitude uses four antennas to cancel gain magnitudes under assumptions.
Thermal noise
Noise in real and imaginary visibility components is often approximated as Gaussian under suitable averaging. Image noise depends on system temperature, bandwidth, integration time, antennas, flagging and weights. Doubling integration time improves ideal thermal noise by about √2, not two.
Low-SNR amplitude bias
Magnitude √(X²+Y²) is non-negative, so noise produces positive amplitude even for zero true signal. Analyse complex components or use suitable bias-aware statistics rather than interpreting every positive amplitude as detection.
Flagging
Radio-frequency interference, hardware faults or bad weather can invalidate data. Flagging changes uv coverage and beam structure. Criteria should be reproducible; deleting samples because they complicate an image invites bias.
Time and Bandwidth Smearing
During finite integration, an off-axis visibility phase rotates. Averaging vector samples with changing phase reduces amplitude. Longer baselines, greater offset and longer integration generally worsen time smearing.
Across finite channel width, phase also varies with frequency. Averaging causes bandwidth smearing radially away from the phase centre. NRAO guidance recommends choosing channel width and integration time based on observation goals and field.
Data-volume trade-off
Finer time and frequency resolution protects field fidelity but expands data and processing. The optimisation is scientific: preserve the scales and field needed for the question without assuming every observation needs maximum resolution.
Wide-field geometry
When non-coplanar effects matter, a simple 2D Fourier transform is insufficient. Techniques such as w-projection or faceting address the extra phase term. Model dimensionality grows with the field and accuracy goal.
Misconceptions Worth Correcting
“The array photographs the sky directly”
It samples correlations in spatial-frequency space. Imaging is an inverse reconstruction.
“The longest baseline determines everything”
It influences fine resolution, while short baselines, layout, sensitivity, weighting and calibration determine other essential properties.
“Empty uv cells mean dark sky”
They mean unmeasured spatial frequencies, not zero brightness information.
“Deconvolution reveals the unique true image”
Incomplete sampling makes the inverse problem non-unique. Algorithms impose constraints and should be stress-tested.
“More observing time always fills every gap”
Earth rotation traces geometry-dependent tracks. Some spacings remain unsampled without different antennas, configurations or frequencies.
Learning Pathways for Students
Radio interferometry unites geometry, vectors, complex numbers, trigonometry, Fourier transforms, statistics and optimisation. Students can simulate point sources and sparse Fourier sampling with synthetic arrays.
Four-week plan
- Week 1: wavelength, delay, phase and projected baselines;
- Week 2: complex visibility and simple source transforms;
- Week 3: uv coverage, dirty beams, weighting and missing scales;
- Week 4: calibration, noise, smearing and a reproducible image report.
The careers span astronomy, antennas, communications, imaging and scientific computing. Mathematics expands access to these questions but does not guarantee a particular pathway.
Frequently Asked Questions
What is a baseline?
It is the vector separation between an antenna pair, projected relative to the observed source and often expressed in wavelengths.
What is a visibility?
It is a complex correlation measurement associated with a baseline, time and frequency under the interferometer convention.
Why are there N(N−1)/2 pairs?
Each unordered pair of distinct antennas forms one baseline; combinatorics counts the pairs.
Why use complex numbers?
They store correlation amplitude and phase compactly, and Fourier relationships become natural.
What is aperture synthesis?
It combines measurements from many projected baselines, often over Earth rotation and frequency, to sample an effective aperture.
What is the dirty beam?
It is the response associated with weighted uv sampling; the dirty image is convolved with this response under the model.
Why can extended emission disappear?
An array without sufficiently short baselines does not measure the spatial frequencies corresponding to large angular scales.
What does weighting change?
It changes the trade among sensitivity, resolution and sidelobe structure; it does not create missing measurements.
Why observe a calibrator?
Its known or modelled behaviour helps estimate instrumental gain, phase, bandpass and flux scale.
What is closure phase?
It is a triangle combination that cancels ideal antenna-based phase errors, preserving source-dependent information.
Does a small image pixel improve resolution?
No. It samples the reconstructed beam more finely but cannot exceed information in the baselines.
What is the transferable lesson?
When data sample a transform incompletely, every image carries the measurement pattern and the reconstruction assumptions.
Useful Next Reading
Read Why Mathematics? | LiDAR, Time of Flight and 3D Point Clouds for a different route from waves to geometry. Why Mathematics? | Ultrasound Imaging, Echo Timing and Beamforming compares coherent arrays in another medium. Why Mathematics? | Comparing Percentages Fairly supports signal-to-noise comparisons.
Radio interferometry shows mathematics at its most imaginative: separate antennas, Earth’s rotation and complex correlations become a synthetic aperture. The result is powerful precisely because the missing information and assumptions can be made visible.
A Practical Investigation Studio
These activities use synthetic or openly released teaching data. They are designed to make every assumption visible. For each investigation, begin with a labelled diagram or data dictionary, keep unrounded intermediate values, and finish with a short limitation statement. The goal is not to imitate a professional instrument with a spreadsheet; it is to understand which mathematical operation gives the instrument its meaning.
Investigation 1: Baseline in wavelengths
Convert physical baselines to u and v coordinates at two frequencies. Show that the same metre baseline samples different spatial scales when wavelength changes. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 2: Fringe spacing
Estimate lambda/B in radians and arcseconds for several baselines. Label it as an order-of-magnitude angular scale, not a full imaging guarantee. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 3: Complex visibility
Represent amplitude and phase as a complex number and rotate its phase. Plot real and imaginary components and keep the sign convention explicit. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 4: Two-point sky
Compute a simple sum of complex exponentials for two fictional sources. Connect changing baseline to oscillation in visibility rather than a literal sky image. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 5: Earth rotation
Sketch how a projected baseline traces points in the uv plane over time. Explain why source declination and array geometry change the track. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 6: Dirty beam
Fourier transform a sparse sampling mask. Relate sidelobes to missing uv samples and avoid calling them celestial sources. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 7: Weighting trade-off
Compare natural, uniform and tapered weights on fictional samples. Report the trade between sensitivity, resolution and sidelobe behaviour. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 8: Missing short spacings
Remove measurements near the uv origin from a simulated extended source. Show how large-scale flux or structure can be missed. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 9: Time averaging
Average a rotating complex visibility over longer intervals. Quantify amplitude loss away from the phase centre and connect it to smearing. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 10: Phase calibration
Add one antenna-based phase error and form baselines to other antennas. Use closure reasoning to distinguish antenna terms from source structure. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 11: Noise and uncertainty
Add complex Gaussian noise and repeat an amplitude estimate. Show why low-SNR amplitude can be biased positive and report intervals. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.
Investigation 12: Reproducible image
Save frequency, array, flagging, calibration, weights, cell size and deconvolution settings. Ask a peer to reproduce the dirty image before comparing restored products. Include one graph or table, one dimensional or limiting-case check, and one paragraph separating the calculated result from the real-world conclusion. Change one input across a sensible range to test sensitivity. If the result changes sharply, report the instability instead of hiding it with extra decimal places. Preserve the original data and record every correction so a classmate can audit the route from measurement to claim. Before calculating, predict the direction of change and explain the physical reason. After calculating, compare the prediction with the result and investigate any disagreement. Repeat the analysis with one deliberately flawed input—wrong unit, reversed sign, missing sample or premature rounding—and describe how the error appears. This makes the activity an error-detection exercise rather than a search for one polished answer. Finish by naming which quantities were observed, which were assumed, which were fitted and which were derived. State the domain over which the model is credible and one condition that would require a richer model. A strong submission includes the raw synthetic data, formulas or code, labelled axes, units, intermediate checks, final result, uncertainty note and a concise explanation suitable for a younger student. Do not claim professional validation from a classroom simulation.