AI SL begins with numbers before it begins with algebra. Topic 1 asks you to round sensibly, work with standard form, find the bounds of a measurement, measure how wrong an estimate is, and then use arithmetic and geometric sequences as models of real situations. The two halves meet in one exam skill: making a prediction and saying how far to trust it.
Below are the ideas you need, four worked examples checked step by step, and the errors that cost marks.
Where this sits in the course
Approximation, standard form and sequences all belong to Topic 1, Number and algebra, which the current AI guide gives 16 suggested hours at SL. Both AI SL papers require a GDC. These skills appear as short Paper 1 questions and as the opening parts of longer Paper 2 questions. If you want regular practice with a tutor who teaches this course, see our AI SL maths tutoring.
The finance half of Topic 1 (compound interest, loans and the TVM solver) has its own deep dive in our sibling post on the AI SL TVM solver.
Part 1: approximation and error
Decimal places and significant figures. 0.004 718 to 2 s.f. is 0.0047; 47 180 to 2 s.f. is 47 000. Zeros at the front never count; zeros between digits always do. IB answers default to 3 significant figures unless the question says otherwise or the answer is exact.
Standard form. A number written as a × 10^k, where 1 ≤ a < 10 and k is an integer. For example, light travels about 3.00 × 10^8 m each second, so in a year (365.25 days) it travels
3.00 × 10^8 × 365.25 × 24 × 3600 ≈ 9.47 × 10^15 m.
Enter numbers in standard form with the calculator's ×10^x key rather than typing "× 10 ^", which can upset the order of operations.
Bounds. A measurement rounded to some accuracy could have been up to half a unit either side. "7.4 cm to 1 decimal place" means 7.35 ≤ L < 7.45.
Percentage error. The formula booklet gives it as ε = |(v_A − v_E) ÷ v_E| × 100, given as a percentage, where v_A is the approximate value and v_E the exact value. Always divide by the exact value.
Estimation. Rounding to 1 significant figure before calculating gives a quick check: 48.7 × 21.3 ≈ 50 × 20 = 1000, so an answer of 1037.31 is plausible and 10 373.1 is not.
Worked example 1: bounds and percentage error
A rectangular tile is measured as 7.4 cm by 3.2 cm, each to 1 decimal place.
(a) Write down the bounds of each length.
7.35 ≤ length < 7.45 and 3.15 ≤ width < 3.25.
(b) Find the bounds for the area.
Lower bound: 7.35 × 3.15 = 23.1525 cm².
Upper bound: 7.45 × 3.25 = 24.2125 cm².
The calculated area 7.4 × 3.2 = 23.68 cm² sits between them, as it must.
(c) A student uses π ≈ 3 to find the area of a circle. Find the percentage error in the value of π.
ε = |3 − π| ÷ π × 100 = 4.51%.
With π ≈ 22/7, the error drops to 0.0402%.
Part 2: sequences as models
| Arithmetic | Geometric | |
|---|---|---|
| Pattern | Add d each time | Multiply by r each time |
| nth term | uₙ = u₁ + (n − 1)d | uₙ = u₁ × r^(n − 1) |
| Sum of first n terms | Sₙ = (n/2)(2u₁ + (n − 1)d) or (n/2)(u₁ + uₙ) | Sₙ = u₁(rⁿ − 1)/(r − 1), r ≠ 1 |
| Real-life shape | Steady growth: seats per row, weekly savings | Percentage growth or decay: users, populations, value |
You may also see sigma notation: Σ from k = 1 to 20 of (4k + 1) means 5 + 9 + 13 + … + 81, which is an arithmetic series with sum 20/2 × (5 + 81) = 860. Infinite geometric series are part of AA rather than AI SL, so AI SL questions stick to finite sums.
For the basics of both formulas and their link to finance, our sequences and series overview is a good refresher.
Worked example 2: an arithmetic model and the "find n" step
A theatre's first row has 18 seats, and each row has 3 more seats than the row in front. There are 25 rows.
(a) Seats in the last row. u₂₅ = 18 + 24 × 3 = 90.
(b) Total seats. S₂₅ = (25/2)(18 + 90) = 1350.
(c) Which is the first row with more than 60 seats?
18 + 3(n − 1) > 60 gives n − 1 > 14, so n > 15. Row 15 has exactly 60 seats, which is not more than 60. The answer is row 16 (63 seats).
(d) A smaller hall must seat at least 1000 people with the same pattern. How many rows are needed?
Sₙ = (n/2)(36 + 3(n − 1)) ≥ 1000. Solving on the GDC, the equation Sₙ = 1000 gives n ≈ 20.9.
Check whole numbers: S₂₀ = 930 and S₂₁ = 1008. So 21 rows.
Worked example 3: a geometric model and logarithms on the GDC
An app has 5000 downloads in its first month, and monthly downloads grow by 8% each month.
(a) Downloads in month 12. u₁₂ = 5000 × 1.08^11 = 11 658 (to the nearest whole number).
(b) In which month do monthly downloads first exceed 20 000?
5000 × 1.08^(n − 1) > 20 000, so 1.08^(n − 1) > 4.
The GDC (or logarithms) gives n − 1 > 18.01, so n − 1 = 19 and n = 20.
Check: month 19 gives 19 980 (just under), month 20 gives 21 579.
This is the classic trap. Students see 18.01, round to 18, and answer month 19.
(c) Total downloads in the first 12 months. S₁₂ = 5000(1.08^12 − 1)/(1.08 − 1) = 94 886 (about 94 900).
(d) In which month does the running total first pass 100 000? S₁₂ ≈ 94 886 and S₁₃ ≈ 107 476, so month 13.
Worked example 4: data that is nearly arithmetic
A new gym records its membership at the end of each of its first five months:
| Month n | 1 | 2 | 3 | 4 | 5 |
|---|---|---|---|---|---|
| Members | 412 | 437 | 465 | 488 | 514 |
(a) Explain why an arithmetic model is reasonable.
The differences are 25, 28, 23 and 26: not constant, but close to each other with no clear trend.
(b) Build the model. Use the average difference: d = (514 − 412) ÷ 4 = 25.5.
Model: uₙ = 412 + 25.5(n − 1).
(c) Predict the membership at month 10. u₁₀ = 412 + 25.5 × 9 = 641.5, so about 642 members.
(d) The actual number at month 10 was 610. Find the percentage error of the prediction.
ε = |641.5 − 610| ÷ 610 × 100 = 5.16%.
(e) Comment. Month 10 is well outside the data, so this was extrapolation. The model overestimated, which suggests growth was slowing. A model that fits five months closely can still drift once conditions change. Comments like this earn the "interpret" marks; our guide to choosing and judging AI SL models goes further into choosing and judging models.
Mistakes our tutors see
- Dividing by the approximate value in percentage error. The exact value goes underneath.
- Upper bound written with ≤. For rounded measurements the upper bound is not included: 7.35 ≤ L < 7.45.
- Rounding n the wrong way. Any "first time it exceeds" question needs a whole-number check either side.
- Using n instead of n − 1 in the geometric nth term. Check with u₁.
- Mixing up uₙ and Sₙ. "In month 12" is a term; "in the first 12 months" is a sum.
- Rounding too early. Store 1.08^11 and similar values in memory; round once at the end.
Exam technique
- Name the model and give its formula before using it: "arithmetic, u₁ = 18, d = 3".
- When solving for n with the GDC, write the equation you solved, then the whole-number check.
- State accuracy: "to 3 s.f." or "to the nearest whole number".
- When a question says "comment", talk about the context, the data range and whether the pattern is likely to continue.
Our post on exact answers and approximations in IB maths explains why examiners care so much about stated accuracy.
Self-check
- A length is 2.6 m to 1 decimal place. Write down its bounds. Answer: 2.55 ≤ x < 2.65.
- An exact value is 48 and an estimate is 50. Find the percentage error. Answer: 4.17%.
- An arithmetic sequence has u₁ = 7 and d = 4. Find u₃₀ and S₃₀. Answer: 123 and 1950.
- A geometric sequence has u₁ = 200 and r = 0.9. Find S₁₀. Answer: 1302.6 (1300 to 3 s.f.).
- Write 0.000 4071 in standard form, to 2 significant figures. Answer: 4.1 × 10^(−4).
Where this leads next
Percentage error returns whenever you estimate: for example, judging a trapezoidal-rule estimate in our AI SL calculus guide. Geometric growth leads to exponential models, and at HL to growth that levels off; see our AI HL logistic models post. The full list of AI SL topics shows how Topic 1 connects to the rest.
How Maths Bodhi helps
Topic 1 is usually taught in the first weeks of the DP, so gaps here show early. In the free demo our tutor checks rounding, bounds and one "find n" question to see what needs fixing. The tutor is matched to AI SL, practice is written out in full, and progress is tracked on past-paper style questions. Students approaching exams can also join our IB DP maths revision programme.
Home tuition is available in Gurgaon (Gurugram). Students elsewhere in India and abroad learn with us online. You can book a free demo class.
Written and checked by the Maths Bodhi team. Facts checked against official IB sources, November 2026.
Frequently asked questions
- Are sequences and series in IB Maths AI SL?
- Yes. AI SL covers arithmetic and geometric sequences and series, their use as models in real contexts, and their financial applications such as compound interest and depreciation.
- How do you calculate percentage error in IB maths?
- Use ε = |(approximate − exact) ÷ exact| × 100 and write the result as a percentage. The exact value always goes in the denominator.
- What are upper and lower bounds?
- They are the smallest and largest values a rounded number could have been. A length of 7.4 cm to 1 decimal place lies in 7.35 ≤ L < 7.45.
- How do I find n in a geometric sequence on the calculator?
- Set up the inequality, for example 5000 × 1.08^(n − 1) > 20 000, solve the equation on the GDC, then check the whole numbers either side before giving n.
- Is the sum to infinity in AI SL?
- Infinite geometric series are taught in AA. AI SL sequence questions use the nth term and the sum of a finite number of terms.
- How many significant figures should I give in IB maths?
- Give exact answers where possible; otherwise use 3 significant figures unless the question asks for a different accuracy. Money is usually given to 2 decimal places.
