Identify what is changing in Linearization
Read the independent variable, dependent quantity, interval, and initial or boundary information before choosing an operation. Rebuild missing background in the calculus subject guide, then use calculus practice to retrieve the relationship without being told its name.
Conditions required before the rule can be used
Far from the centre the tangent drifts away from the curve, and it drifts fastest where the graph bends hardest. Estimating the square root of 101 from a centre of 100 is fine; using that same tangent line at x = 4 predicts 5.2 when the true root is 2.
Build the derivative, integral, or model in stages
- Name the unknown, and the unit the answer has to come out in.
- Match the symbols to your values. $a$ is the center and $L$ is the tangent-line approximation.
- Check the conditions before substituting. $f$ must be differentiable near $a$; accuracy usually decreases farther from $a$.
- Substitute, keep exact values to the last line, then test the sign, size, and unit against a rough estimate - The check that catches most calculus slips.
Linearization through one changing case
Track what each line says about rate, accumulation, or evolution:
Check change against behavior and units
After using Linearization, repeat the calculation with units attached and test the result against a boundary case or a quick numerical estimate.
Where a change calculation commonly breaks
What goes wrong: Writing the second term as the slope times x instead of the slope times x − a.
Why it fails: That measures the walk from 0 rather than from the centre a, so the estimate is off by the slope multiplied by a, which is usually a large error.
Repair: Always subtract the centre. The second term is the slope at a times the distance you moved away from a.
Questions about change in Linearization
How do I choose the centre a?
Take the nearest input whose exact value you already know, such as a perfect square or a familiar angle. The closer the centre, the smaller the error.
Does the estimate come out too high or too low?
It sits above the curve wherever the graph is concave down and below it wherever the graph is concave up, because the tangent line lies on the outside of the bend.
How far off is the estimate 2.025 for the square root of 4.1?
The true value is about 2.0248457, so the tangent line is high by roughly 0.00015, which is about one part in thirteen thousand.
Is this the same thing as a differential?
Almost. The differential reports only the change predicted by the slope, while linearization adds the starting height back on to give an actual estimated value.
Change one input, predict whether the output should rise, fall, or stay fixed, and only then recalculate.