Lemonmath
Tiếng Việt

Approximate numbers and errors

Lesson 5.1 · 15 min read · Updated

A ruler says your textbook is 24.4 cm wide. A calculator says 2=1.414213562…\sqrt{2} = 1.414213562\ldots and the display just stops. In real life almost every number you meet, from a length to a population, is approximate. The real question is not "is it exact?" but "how far off can it be?". This lesson gives you the tools to answer that.

Approximate numbers and absolute error

Sometimes the exact value is impossible to write down (π\pi, 2\sqrt{2}) and sometimes it is impossible to know (the exact mass of a bag of rice). Then we use a nearby number instead.

Let aa be the exact value and aˉ\bar{a} an approximation of it.

The smaller dd is, the better the approximation. For π≈3.14\pi \approx 3.14 we can take d=0.002d = 0.002, so π\pi lies between 3.1383.138 and 3.1423.142.

3.143.143.143.143.143.143.143.143.14-1-0.8-0.6-0.4-0.20.20.40.60.81xyO3.1383.142āπ
The number π\pi lies inside the interval 3.14±0.0023.14 \pm 0.002

Relative error

Is an error of 1 cm serious? For the length of a pencil, yes. For the length of a football pitch, not at all. We compare the error with the size of the number.

Rounding and significant figures

To round a number to a chosen place, look at the digit just to the right of that place.

  • If that digit is less than 5, keep the digits on the left and replace the others by zeros (or drop them after the decimal point).
  • If that digit is 5 or more, add 1 to the last kept digit and then do the same.

For example, 7 4597\,459 rounded to hundreds is 7 5007\,500; 3.141593.14159 rounded to 2 decimal places is 3.143.14.

When you round to a place, the error is at most half a unit of that place. So 3.143.14 (rounded to hundredths) has d=0.005d = 0.005, and 7 5007\,500 (rounded to hundreds) has d=50d = 50.

The significant figures of a number are its digits starting from the first non-zero digit on the left. So 0.03050.0305 has 3 significant figures (3,0,53, 0, 5), and 2.402.40 has 3 (2,4,02, 4, 0). To round to kk significant figures, keep the first kk of them and round as above. For example, 0.00718490.0071849 to 3 significant figures is 0.007180.00718, and 12 34512\,345 to 3 significant figures is 12 30012\,300.

Reliable digits

Suppose a=aˉ±da = \bar{a} \pm d. A digit of aˉ\bar{a} is reliable (a "correct digit") if dd is at most half a unit of the place where that digit stands. All digits to the left of a reliable digit are reliable too. When we report a result, we keep only reliable digits and round off the rest.

Key terms

Summary

  • An approximation aˉ\bar{a} of the exact value aa has absolute error Δ=∣a−aˉ∣\Delta = |a - \bar{a}|. An error bound dd satisfies Δ≤d\Delta \leq d, and we write a=aˉ±da = \bar{a} \pm d.
  • The relative error is δ=Δ/∣aˉ∣\delta = \Delta/|\bar{a}|, bounded by d/∣aˉ∣d/|\bar{a}|. Use it to compare the accuracy of different measurements.
  • Rounding to a place gives an error of at most half a unit of that place.
  • A digit is reliable if dd is at most half a unit of its place. Report only reliable digits.

Practice

Show solutions
  1. To tens: 7 459→7 4607\,459 \to 7\,460 (the units digit 9≥59 \geq 5). To hundreds: 7 459→7 5007\,459 \to 7\,500 (the tens digit 5≥55 \geq 5). Then 3.14159→3.143.14159 \to 3.14 (the next digit is 11) and 3.14159→3.14163.14159 \to 3.1416 (the next digit is 99).
  2. 0.03050.0305 has 3 significant figures, 2.402.40 has 3, and 40.00740.007 has 5. For 0.00718490.0071849 keep 77 and 11; the next digit is 8≥58 \geq 5, so round up to 0.00720.0072.
  3. Δ=∣1.7320508−1.73∣=0.0020508≈0.0021\Delta = |1.7320508 - 1.73| = 0.0020508 \approx 0.0021 (rounded up). So d=0.0021d = 0.0021 and δ≤0.00211.73≈0.0012=0.12%\delta \leq \dfrac{0.0021}{1.73} \approx 0.0012 = 0.12\%.
  4. Tailor: 0.058.5≈0.0059=0.59%\dfrac{0.05}{8.5} \approx 0.0059 = 0.59\%. Builder: 0.5120≈0.0042=0.42%\dfrac{0.5}{120} \approx 0.0042 = 0.42\%. The builder's measurement is more accurate because its relative error is smaller.
  5. ∣227−π∣≈0.00126\left|\dfrac{22}{7} - \pi\right| \approx 0.00126 and ∣3.1416−π∣≈0.0000073|3.1416 - \pi| \approx 0.0000073. The second error is much smaller, so 3.14163.1416 is the better approximation.
  6. Half a unit of the tens place is 5≥45 \geq 4, so the tens digit 33 is reliable (and so are 11 and 55). Half a unit of the units place is 0.5<40.5 < 4, so the digit 66 is not reliable. Report 1 5401\,540 (rounded to tens).
  7. Half a unit of the tenths place is 0.05≥0.030.05 \geq 0.03, so the digit 66 is reliable, as are 44 and 55. Half a unit of the hundredths place is 0.005<0.030.005 < 0.03, so the last digit 00 is not reliable. Report 45.645.6.
  8. Perimeter Pˉ=2(25.4+12.6)=76.0 cm\bar{P} = 2(25.4 + 12.6) = 76.0\text{ cm}. Each side may be off by 0.10.1, so d=2(0.1+0.1)=0.4 cmd = 2(0.1 + 0.1) = 0.4\text{ cm} and P=76.0±0.4P = 76.0 \pm 0.4. Relative error: δ≤0.476.0≈0.0053=0.53%\delta \leq \dfrac{0.4}{76.0} \approx 0.0053 = 0.53\%.
  9. No. The last digit 00 is in the thousandths place. The error d=0.001d = 0.001 is larger than half a unit of the thousandths place (0.00050.0005), so the last 00 is not reliable. It is smaller than half a unit of the hundredths place (0.0050.005), so the digits 2,5,02, 5, 0 are reliable. Report 2.502.50.
  10. The smallest area is 25.3×12.5=316.2525.3 \times 12.5 = 316.25 and the largest is 25.5×12.7=323.8525.5 \times 12.7 = 323.85. The estimate is 25.4×12.6=320.04 cm225.4 \times 12.6 = 320.04\text{ cm}^2. The distances are ∣323.85−320.04∣=3.81|323.85 - 320.04| = 3.81 and ∣316.25−320.04∣=3.79|316.25 - 320.04| = 3.79, so d=3.81d = 3.81. Half a unit of the tens place is 5≥3.815 \geq 3.81, so the digits 33 and 22 are reliable, but half a unit of the units place is 0.5<3.810.5 < 3.81, so 00 is not. Report the area as about 320 cm2320\text{ cm}^2.