Sensitivity (Meta) Examples: 45 Problems with Answers

Worked examples of Sensitivity (Meta) to try before you open each solution, then practice problems with answers.

Worked Examples

Example 1

easy
Compute f(x)=x3 at x=2 and x=2.1. Find the sensitivity: by what percentage does f change when x changes by 5%?
Show solution

Answer

5% change in x⇒≈15.8% change in f

First step

1.
f(2)=8, f(2.1)=9.261.

Full solution

  1. 2.
    Change in f: 9.261−8=1.261. Relative change in f: 1.2618≈15.8%.
  2. 3.
    Change in x: 0.12=5%.
  3. 4.
    Sensitivity: a 5% increase in x causes about a 15.8% increase in f. The function is sensitive — it amplifies errors by a factor of about 3.
Sensitivity measures how much the output changes relative to the input. For f(x)=x3, the exponent 3 amplifies percentage changes roughly threefold, which can be derived from the derivative: f′(x)/f(x)≈3Δx/x.

Example 2

medium
In the compound interest formula A=P(1+r)t, compute the sensitivity of A to a small change Δr in the interest rate, using the derivative dAdr.

Example 3

medium
A retirement formula A=P(1+r)t with P=10000, t=20 years. Compute A for r=0.06 vs r=0.07 to gauge sensitivity to the rate.

Example 4

hard
The quadratic formula computes x=−b+b2−4ac2a. For a=1,b=106,c=1, explain why a naive direct computation is highly sensitive to rounding, and give the stable form.

Example 5

challenge
Heron's formula for triangle area is sensitive when the triangle is thin (near degenerate). For sides a=1, b=1, c=1.9999, compute area with both the standard and a numerically stable form and discuss the gap.

Practice Problems

Try each one, then open the solution to compare your method.

Example 1

easy
For f(x)=x, compute f(100) and f(104). By what percentage does f change when x increases by 4%?

Example 2

medium
A model predicts exam score S=10log⁡10(h) where h is hours of study. If h changes from 10 to 11 hours, compute the change in S and discuss sensitivity.

Example 3

easy
A function is f(x) = 1000x. If x changes by 0.01, by how much does f change?

Example 4

easy
Function g(x) = 0.001x. If x changes by 100, by how much does g change? Is g sensitive to x?

Example 5

easy
Computing 1/x near x = 0.001, then near x = 1000. Where is 1/x more sensitive to a small change in x?

Example 6

easy
A measured side s of a square has 1% error. The area is s^2. Approximately what percent error appears in the area?

Example 7

easy
In y = 5x + 2, which input change causes a bigger output change: changing x by 1 or changing the constant 2 by 1?

Example 8

easy
A bank balance grows by interest rate r. A 0.1% change in r barely changes one month's interest but compounds over 30 years. This shows sensitivity can depend on what?

Example 9

easy
Two recipes: A's taste is ruined if salt is off by a pinch; B tolerates large salt swings. Which is more sensitive to the salt input?

Example 10

easy
A formula's answer changes from 5.00 to 5.01 when an input changes by 50%. Is the answer sensitive to that input?

Example 11

medium
For f(x) = x^2 at x = 3, estimate the output change when x increases by 0.1 using the derivative, and compare with the exact change.

Example 12

medium
A system solves A x = b. The condition number of A is 1000. If b has 0.1% relative error, what is the worst-case relative error in x?

Example 13

medium
Profit P = (price - 10) * (1000 - 20*price). Determine whether profit is more sensitive to price near price = 20 or near price = 40, using the slope.

Example 14

medium
A computation evaluates (1 + x) - 1 for x = 1e-16 in floating point and gets 0 instead of 1e-16. Which input region is this operation sensitive to, and what causes the failure?

Example 15

medium
Compound interest: A = P(1 + r)^t. With P = 1000, t = 30, compare final amounts for r = 0.05 and r = 0.06 to gauge sensitivity to r.

Example 16

medium
A formula y = a/(a - b). For a = 100, compare sensitivity of y to a small change in b when b = 1 versus b = 99.

Example 17

challenge
For f(x, y) = x*y at (x, y) = (4, 5), use partial derivatives to find which variable the output is more sensitive to, and the first-order output change if both increase by 0.1.

Example 18

challenge
A relative-sensitivity (elasticity) is defined as E = (x/f) * df/dx. Compute the elasticity of f(x) = x^n and interpret the result.

Example 19

challenge
A root-finding iteration for f(x) = 0 near a root r has error multiplied by |f'(r)|^(-1) per measurement error in f. If f(x) = (x - 2)^3, explain why the root x = 2 is highly sensitive to perturbations in f.

Example 20

medium
A thermostat formula output = k*(target - current) with k = 50. If the temperature reading is off by 0.2 degrees, how much does the output change, and is the system sensitive to reading errors?

Example 21

medium
For f(x) = ln(x), is the output more sensitive to a small change in x near x = 0.1 or near x = 10? Use f'(x) = 1/x.

Example 22

medium
A recipe scales servings by a factor s, and total cost is C(s) = 5s. A spreadsheet rounds s to the nearest 0.5. If true s = 2.2 rounds to 2.0, what is the resulting cost error, and is cost sensitive to this rounding?

Example 23

easy
For f(x)=20x, by how much does f change when x increases by 0.25?

Example 24

easy
For f(x)=x2, compute Δf when x goes from 4 to 4.1.

Example 25

easy
A measured side s of a cube has 1% error. Volume is s3. Approximately what percent error in volume?

Example 26

easy
For f(x)=1x, where is f more sensitive to small changes in x: near x=0.1 or near x=10?

Example 27

medium
For f(x)=x, estimate the change in f when x goes from 25 to 26 using the derivative.

Example 28

medium
A linear system Ax=b has condition number κ(A)=500. If b is known to 0.05% relative error, what is the worst-case relative error in x?

Example 29

medium
For f(x)=tan⁡x, compare sensitivity at x=0 vs x near π/2 (e.g., x=1.5).

Example 30

medium
Output y=3a2b. If a changes by 1% and b stays fixed, approximately what percent change in y?

Example 31

medium
A model uses T=10P. Compute the sensitivity dT/dP at P=100 and the approximate change in T if P shifts by 1.

Example 32

medium
Cancellation: computing x+1−x for large x loses precision. Rewrite to reduce sensitivity to rounding.

Example 33

medium
A function f(x,y)=x2+y at (3,4). Which input has greater first-order sensitivity?

Example 34

hard
For f(x)=1(x−2)2, identify the input region where the output is most sensitive to perturbations.

Example 35

hard
For f(a,b)=a⋅b at (a,b)=(10,2), compute the linearized change if both a and b each increase by 0.1.

Example 36

hard
The elasticity of f(x)=xn is defined as E=xf⋅f′(x). Compute E for f(x)=5x−2.

Example 37

hard
A bridge load model L=10000+50w−2w2 where w is wind speed. Find the wind speed where L is most sensitive (in magnitude) to w over 0≤w≤30.

Example 38

medium
In a chain f(g(x)) with f′=4 and g′=0.5 at the relevant points, by what factor does input error get amplified?

Example 39

challenge
For a 2x2 matrix A=(1111.01), the system Ax=b is ill-conditioned. Compute det⁡A and explain why solving Ax=b is sensitive to small changes in b.

Example 40

challenge
For f(x)=sin⁡(1000x), why is computing f at large x in floating point highly sensitive to representation of x?

What to focus on

Core idea: Sensitivity is the ratio of how much the output changes to how much the input changed.

Common stuck point: The procedure for sensitivity (meta) is the easy part; the trap is reporting the raw output change as sensitivity. Asking "Am I measuring how much the output moves PER unit change in the input?" first is what keeps a correct-looking calculation from being attached to the wrong concept.

Ask yourself: Am I measuring how much the output moves PER unit change in the input?

Concept recap

The degree to which a result or output changes in response to small changes in its inputs, parameters, or assumptions.

Sensitivity is the gearing between a dial and a needle: on a low-sensitivity dial a full turn barely moves the needle, while on a high one a hair's turn swings it across the scale. Unlike a fixed gear, the ratio can change with where the dial is set, as it does near a double root.

Read the full lesson on Sensitivity (Meta)

Background knowledge

These ideas may be useful before you work through the harder examples.

local vs global behavior