Graphing for Sanity Checks — Domains, Discontinuities, Sampling — GetCalcMaster
Use 2D/3D graphing to validate formulas: choose ranges, handle asymptotes, avoid misleading sampling, and cross-check numerically.
Graphing is the fastest “does this formula behave?” test
A calculator can tell you a number, but a graph can tell you whether the relationship makes sense. If you’ve ever trusted a result and later found an off-by-sign or wrong-domain error, graphing is the habit that prevents it.
Choose a domain before you trust a plot
Most graphing mistakes aren’t about the function — they’re about the viewing window.
- Start small: pick a domain that matches the scale of the problem (e.g., −10..10).
- Expand deliberately: widen the window only after you see the shape.
- Check units: if x is seconds, don’t graph a week-long domain unless you mean it.
Discontinuities and asymptotes are not “bugs”
Functions like 1/x, tan(x), and 1/(x-2) blow up at certain x values.
A good graphing tool shows gaps or breaks — it should not connect the curve through infinity.
2D examples to try
- Asymptote: plot
1/(x-2)and look for the break at x=2. - Oscillation: plot
sin(x)and confirm the period is2π(in radians). - Non-smooth: plot
abs(x)and confirm the cusp at 0.
Open /graph/2d and try those with a domain of −10..10.
Sampling can lie (aliasing)
Graphing tools don’t draw every point — they sample. If your function changes faster than the sampling rate, the curve can look smoother than it is, or it can miss narrow spikes entirely.
Signals example: if you plot a high-frequency sine wave on a wide domain, it may look like noise or a low-frequency wave. The fix is to zoom in or increase sampling resolution (if available).
Verify with a few numeric points
A graph should match numeric evaluation at specific x values. Pick 3–5 points:
- one near 0
- one near a suspected root
- one near an asymptote or boundary
Then evaluate those points in a calculator to confirm the plotted y-values are consistent.
Try it
3D graphs: surfaces add one more failure mode
In 3D, the “window” is an x/y domain, and the surface can have singularities (division by zero) or sharp ridges. Start with simple surfaces before jumping to complicated ones.
3D examples
- Bowl:
x^2 + y^2 - Saddle:
x^2 - y^2 - Gaussian:
exp(-(x^2+y^2))
Open /graph/3d and start with a domain of −3..3 to keep it stable.
A graphing checklist (fast)
- Did I pick a domain that matches the problem scale?
- Does the function have discontinuities I should expect?
- Do 3–5 numeric points match the plotted curve?
- Is my trig mode (degrees/radians) consistent with my source?
- If it looks weird, did I zoom in before concluding it’s wrong?
FAQ
Why does the graph “connect across” a vertical asymptote sometimes?
That’s a sampling artifact. If the tool samples too sparsely, it may draw a line between points on opposite sides of the asymptote. Zoom in or increase resolution if possible.
Is graphing enough to prove a result?
No. Graphing is a sanity check, not a proof. Use it to catch mistakes and build intuition, then verify numerically or analytically as appropriate.
What’s the best first plot for a new formula?
Start with a simple domain around 0 and plot a simplified version of the formula (or a special case) to confirm you understand its behavior.