TCE General Mathematics Mastery Pack

1. From a situation to a mathematical model

A mathematical model is a deliberately simplified representation of a real situation. Begin by naming the quantity to explain or predict, its units, the input variable and the period or population to which the information applies. Then separate supplied facts from assumptions. For a savings account, the opening balance and quoted rate may be facts, while a constant rate, no fees and equal time periods are modelling assumptions. This translation matters because an algebraically correct answer can still answer the wrong question.

A useful modelling cycle is: formulate the question; choose variables and assumptions; select a model; calculate; interpret; test reasonableness; and refine or report. The stages are connected. If a graph shows curved rather than constant growth, that evidence should change the proposed model. If a calculated vehicle value becomes negative, the context should change the useful domain even when the formula continues algebraically. Record these decisions so another reader can follow why the model was chosen.

For example, a club has 240 members now and expects 18 additional members each month. Let n be whole months after the count and M_n the membership. The recurrence M₀ = 240, M_n+1 = M_n + 18, or the rule M_n = 240 + 18n, represents constant absolute change. The domain is n = 0, 1, 2, … and outputs are member counts. At month 7, the model gives 366 members. A report must say “about 366 members after seven months under the constant-addition assumption”, rather than presenting 366 without its meaning or conditions.

One practical way to formulate the model is to make a small variable table before calculating. For the club example, n means elapsed whole months and M_n means the number of members at that time. The initial condition belongs at month zero, not month one. This notation prevents the common shift in which seven updates are accidentally labelled month eight. It also makes the output constraint explicit: although the formula is linear, a membership prediction should ultimately be reported as a whole-person estimate.

2. Choose a model from the pattern of change

Model choice should be justified by how the quantity changes. A linear or arithmetic model fits approximately constant first differences: the same amount is added in each equal interval. A geometric or exponential model fits approximately constant ratios or percentage changes: the same multiplier is applied each interval. A first-order recurrence can show either process and is especially useful when each new value depends on the preceding balance. A financial model may also require regular deposits, withdrawals or repayments, so a simple compound-growth expression may omit an essential cash flow.

Suppose a machine is valued at $18,000 and loses $2,400 each year. The first values are $18,000, $15,600 and $13,200; the difference is consistently −$2,400, so V_n = 18,000 − 2,400n is a flat-rate model. If instead it loses 16% of its current value each year, successive values are multiplied by 0.84 and V_n = 18,000(0.84)ⁿ is appropriate. At year 3 these give $10,800 and $10,668.67 respectively. Similar early answers do not make the models interchangeable: their assumptions and long-term behaviour differ.

For observed data, inspect a table, graph and relevant differences or ratios rather than choosing a familiar formula immediately. Allow for noise: real data rarely have perfectly equal differences. Compare whether departures from a proposed pattern are small and structureless or whether they curve systematically. State why the chosen feature supports the model, such as “the annual dollar decreases are approximately constant”, and also state a plausible limitation, such as a changing resale market. Model selection is an evidence-based decision, not a calculator menu choice.

A table can separate the competing patterns. Flat-rate values have first differences −2400, −2400, …, whereas reducing-balance values have changing dollar differences but ratios 0.84, 0.84, …. Checking only whether both lists decrease is insufficient, since many models share that broad feature. Use the feature that distinguishes them. If observed ratios drift and first differences also change, neither elementary model may be adequate over the full period, and a shorter fitted interval or a revised assumption may be required.

3. Work through finance with timing and units

Financial quantities depend on both the rate and the timing. Convert the annual rate to the rate per compounding period only when the problem’s convention supports that conversion, and make the number of periods match. Distinguish a present balance from a future balance, and distinguish growth of an untouched investment from an annuity with regular payments. A timeline marked 0, 1, 2, … often prevents an off-by-one error. Calculator settings such as payments at the beginning or end of a period are assumptions that must match the situation.

An untouched $5,000 investment earning 6% per year compounded annually for four years has A = 5000(1.06)⁴ = 6312.3848, so the balance is $6,312.38 to the nearest cent. A recurrence check gives A₀ = 5000 and A_n+1 = 1.06A_n; four updates produce the same unrounded value. The interest earned is $1,312.38, not $6,312.38. Keeping full precision until the final line prevents accumulated rounding error.

Reasonableness can be checked without repeating the entire calculation. Four years of simple interest at 6% would add $1,200, so a compound balance slightly above $6,200 is plausible; $63,123 is not. Also check that the balance exceeds the principal for a positive rate, the unit is dollars, and exactly four multiplications occurred. The final conclusion should identify the amount, date and assumptions: “With no deposits, withdrawals or fees and a constant 6% annual rate, the investment is modelled to be worth $6,312.38 after four years.”

Nominal descriptions must be read carefully. A quoted “6% per annum compounded monthly” is not the same timing as 6% compounded annually: under the usual nominal convention the monthly periodic rate is 0.06/12 and four years contains 48 periods. An effective annual rate is already a one-year multiplier and should not be divided by 12. The governing task or product definition decides the conversion, so state it rather than silently applying a remembered calculator routine.

4. Use technology, but expose the mathematics

Technology can generate tables, graphs, regression output and finance values efficiently, but the result must remain connected to the mathematical model. Define variables before entering expressions, show the equation or recurrence used, and retain enough working to reveal the inputs and timing. A screenshot or calculator number by itself does not explain whether the correct functionality, window, domain or payment convention was selected.

Verification uses a method that can detect an input or interpretation error. For the membership model, substitute n = 7 into 240 + 18n and also generate seven recurrence steps. For compound growth, compare an exponent calculation with a balance table. If graphing a model, choose a domain and range that display the meaningful data rather than a default window that hides intercepts or creates the illusion of a poor fit. Exact expressions can be kept during working even when the contextual answer is rounded.

Suppose software reports V(6) = 3,600 for V(n) = 18,000 − 2,400n. Substitution gives 18,000 − 14,400 = 3,600, confirming the computation. The model reaches zero at n = 7.5, but annual values use whole-number n and a physical asset cannot have a negative resale value. A sensible model domain may therefore end before the formula becomes negative. This check separates a technically valid software output from a defensible contextual result.

Graphical and numerical representations should agree. In a constant-addition recurrence, the generated points lie on the corresponding linear rule at integer inputs; the continuous line between them is a visual extension, not an extra set of monthly observations. In a finance table, each row should show the opening balance, interest or change, any transaction and closing balance in the correct order. That structure allows a reader to test one period locally and locate a timing error that a final-value check alone might miss.

5. Test assumptions, limits and sensitivity

Every model suppresses some detail. Assumptions state the conditions under which its calculations are intended to apply; limitations explain why actual outcomes may depart from them. Examples include a stable growth rate, equal time intervals, unchanged fees, reliable measurements, or a population that has enough capacity for continued growth. Write assumptions specifically. “The model is inaccurate” gives no basis for judging when or why it may fail.

A sensitivity check changes a plausible assumption and observes the effect. For the $5,000 investment, a 5.5% annual rate gives 5000(1.055)⁴ = $6,194.12, whereas 6.5% gives 5000(1.065)⁴ = $6,432.33. The range of about $238 shows how a one-percentage-point spread in the assumed rate affects the four-year prediction. This does not identify the true future rate, but it makes the dependence of the conclusion visible.

Check domain as well as numerical size. A linear membership rule may be reasonable for the next year but impossible indefinitely because staffing and building capacity are finite. An exponential depreciation model stays positive, yet may still become unrealistic once a scrap value dominates. When extending a model beyond observed data or its intended period, say that uncertainty increases. A strong evaluation links each limitation to the direction or nature of possible error rather than attaching a generic caution to an otherwise unexamined answer.

Limitations should be proportional to the decision. A $120 uncertainty may be immaterial when comparing investments that differ by $2,000, but decisive when their modelled values differ by $32.38. Compare the sensitivity range with the gap between alternatives. This turns evaluation into mathematical evidence: if plausible assumptions reverse the ranking, the conclusion should say the options cannot be separated confidently under the available information.

6. Draw a conclusion that answers the question

A conclusion is the final act of modelling, not a restatement of calculator output. It should answer the original question in words, use appropriate units and rounding, name the time or input value, and mention a decisive assumption or limitation. If alternatives were compared, state the criterion used: lowest total cost, greatest final value, closest fit to observed data, or another quantity that the problem actually asks about.

Imagine two four-year saving options for $5,000: Option A is the 6% compound investment, modelled at $6,312.38; Option B guarantees a maturity value of $6,280 after all fees. Under the stated figures, A is $32.38 higher. A defensible conclusion is therefore that A gives the larger modelled maturity amount, provided its 6% rate remains constant and no unlisted fees apply. Claiming that A is “always better” would exceed both the time domain and the evidence.

Before submitting, audit the chain: variables and units are defined; the model suits the pattern; substitutions and technology settings match the situation; intermediate precision is retained; the answer is checked; and the conclusion acknowledges material conditions. Criterion 3 rewards applying models to represent and analyse real situations, so the reasoning around a number is part of the mathematics. A result becomes useful only when a reader can see what it represents, why it is plausible and where its authority ends.

A final unit audit often reveals hidden errors. Money normally needs a dollar sign and cents, a population needs a sensible whole-number interpretation, a rate needs its time basis, and a predicted time needs the correct period label. If rounding changes whether a threshold is met, evaluate the unrounded value first and then report it. For example, $6,312.3848 exceeds a $6,300 target before it is displayed as $6,312.38; the decision is made from the unrounded result.