Calculator and technology skills

A calculator can evaluate, approximate and display mathematics. It cannot decide what a symbol means, choose a valid model, prove a claim or judge whether an answer is sensible. At A-level, effective technology use therefore has three stages:

1. Formulate: translate the problem into correct mathematics.
2. Compute: use a calculator or other permitted tool accurately.
3. Interpret: check the output and answer the question in context.

The mathematics must remain visible. Writing only a decimal copied from a screen gives no evidence of the method and may earn little credit even when the number is correct.

Prerequisites

You should be able to:

• use brackets, fractions, powers and roots;
• distinguish exact values from decimal approximations;
• round to significant figures and decimal places;
• read coordinates and basic function graphs.

Review calculator fluency for key entry techniques and standard form and accuracy for rounding, estimation and bounds.

Button names differ between models. Learn where your calculator stores its angle, equation, table and statistics settings, but learn the mathematics independently of any particular sequence of keys.

Write before you press

Record a formula, substitution or equation before calculating. This separates a mathematical error from an entry error and preserves method marks.

Worked example: retain the structure

The displacement of a particle is

$s=3t^2e^{-0.4t}.$

Find $s$ when $t=2.7$.

First substitute visibly:

$$s=3(2.7)^2e^{-0.4(2.7)}.$$

Then evaluate the complete expression:

$$s=7.430068\ldots\approx 7.43.$$

The line of substitution shows that the exponent is $-0.4t$, not merely $-0.4$, and that $t^2$ applies only to $t$.

Misconception: the screen is working

A calculator history is useful for checking entry, but it is not a mathematical explanation. For a quadratic equation, for example, write the equation and its solutions. If a question requires a particular method, such as completing the square or Newton’s method, a solver output cannot replace that method.

Settings are part of the mathematics

Before a calculation, check settings that affect its meaning.

Degrees and radians

The same number represents different angles in different modes:

$30^\circ=\frac{\pi}{6}\text{ radians}.$

Consequently,

$\sin 30^\circ=\frac12,$

but

$\sin(30\text{ radians})=-0.9880\ldots.$

Use degrees when angles carry a degree symbol. Use radians when angles involve $\pi$, when the interval is expressed in radians, or when calculus results such as

$\frac{d}{dx}(\sin x)=\cos x$

are being used. Those standard derivative formulae assume radians.

Display, distribution and statistics modes

Fixed decimal display can hide stored precision. Scientific display can disguise the ordinary size of a result. Statistics modes may retain old data. Distribution menus may ask for a lower tail, an upper tail or an interval. Always read the labels and clear old lists before starting a new data set.

Self-check 1

1. In which angle mode should you evaluate $\cos(2\pi/3)$?
2. Your calculator gives $\sin 45=0.8509\ldots$. What is the likely problem?
3. Why can a display of $2.00$ be misleading during a multi-step calculation?

Answers

1. Radian mode, because the angle is expressed using $\pi$.
2. The calculator is probably in radian mode. In degrees, $\sin45^\circ=\sqrt2/2\approx0.7071$.
3. Fixed display may be rounding a more precise stored value to two decimal places. Use the stored value, not the displayed rounded value, in later steps.

Exact values, stored precision and rounding

Keep exact values for as long as possible:

$\frac13,\qquad \sqrt5,\qquad \pi,\qquad \ln 2.$

These contain more information than finite decimal approximations. Convert to a decimal only when the question or context requires one.

Worked example: premature rounding changes the answer

Evaluate

$A=\frac{\pi r^2}{\sqrt3}$

for $r=6.48$, giving the answer to $3$ significant figures.

Using full stored precision,

$$\begin{aligned} A&=\frac{\pi(6.48)^2}{\sqrt3}\\ &=76.1663\ldots\\ &=76.2\quad\text{to }3\text{ significant figures}. \end{aligned}$$

If $\pi$ is replaced by $3.14$ and $\sqrt3$ by $1.73$ too early, the result is $76.205\ldots$. That happens to round to the same value here, but premature rounding can cross a rounding boundary in another question. The safe rule is simple: keep the calculator’s stored values until the final answer.

Use $\approx$ when a value has been rounded:

$\sqrt2\approx1.414,$

not $\sqrt2=1.414$.

Estimate before trusting an output

An estimate should predict the sign and order of magnitude. It need not be especially accurate.

Worked example: detect a missing bracket

Evaluate

$Q=\frac{(19.7)^2}{0.048+0.151}.$

Estimate first:

$Q\approx\frac{20^2}{0.2}=2000.$

Accurate calculation gives

$Q=1950.201\ldots.$

An output near $98$ would suggest that only part of the denominator was used. The estimate does not prove the exact value, but it rejects an impossible scale.

Useful plausibility tests include:

• probabilities must lie in $[0,1]$;
• a standard deviation cannot be negative;
• a length or time is normally positive;
• $|\sin x|\leq1$ and $|\cos x|\leq1$;
• a correlation coefficient satisfies $-1\leq r\leq1$;
• units must match the requested quantity.

Use tables to investigate functions

A table gives discrete samples, not a continuous graph. It is valuable for locating roots, exploring recurrence relations and checking how a function changes.

Worked example: bracket a root

Let

$f(x)=x^3+x-5.$

A table gives

$$\begin{array}{c|ccc} x&1.5&1.6&1.7\\ \hline f(x)&-0.125&0.696&1.613 \end{array}$$

Because $f$ is continuous and changes sign between $1.5$ and $1.6$, there is at least one root in

$1.5<x<1.6.$

A finer table gives

$$\begin{array}{c|cc} x&1.51&1.52\\ \hline f(x)&-0.047049&0.031808 \end{array}$$

so the root lies in $1.51<x<1.52$. A calculator solver gives $x=1.51598\ldots$, but the sign change supplies mathematical evidence for its location. Study change of sign for the conditions and limitations of this argument.

A table can miss behaviour

Sampling at $x=0,1,2,\ldots$ would give

$\sin(\pi x)=0$

at every sampled point, although the function is not identically zero. A smaller step size gives more information, but no finite table proves global behaviour.

Use graphs as evidence, not proof

Graphing technology is excellent for forming conjectures, locating intersections and checking algebra. Its picture depends on the window and sampling resolution.

Worked example: solve by intersection

To investigate

$e^{-x}=x,$

plot

$y=e^{-x}\qquad\text{and}\qquad y=x.$

Their intersection suggests

$x\approx0.567.$

Substitution checks the approximation:

$e^{-0.567}=0.56722\ldots,$

which is close to $0.567$. A numerical solver refines this to

$x=0.567143\ldots.$

The graph identifies a likely solution. It does not, by itself, prove uniqueness. A proof could note that

$h(x)=e^{-x}-x$

has

$h'(x)=-e^{-x}-1<0,$

so $h$ is strictly decreasing and can cross zero at most once.

Window choice matters

The graph of

$y=(x-100)^2-1$

has roots $99$ and $101$. A standard window such as $-10\leq x\leq10$ shows neither. Before concluding that no root exists, use algebra, inspect the domain and choose a window appropriate to the expected scale.

Graphing can also hide:

• narrow turning points;
• vertical asymptotes joined by an artificial line;
• repeated roots where the curve touches rather than crosses the axis;
• closely spaced intersections;
• oscillations between sampled pixels.

Solvers return candidates

Equation and polynomial solvers are useful for checking results and handling numerical equations. Their output must be tested against the original problem.

Worked example: reject an extraneous solution

Solve

$\sqrt{x+6}=x.$

Since the square root is non-negative, any solution must satisfy $x\geq0$. Squaring gives

$x+6=x^2,$

so

$x^2-x-6=0$

and hence

$x=3\quad\text{or}\quad x=-2.$

A polynomial solver correctly finds both roots of the squared equation. Checking the original equation gives

$x=3:\quad\sqrt9=3,$

whereas $x=-2$ violates $x\geq0$ and gives $\sqrt4\ne-2$. Therefore the original equation has only

$\boxed{x=3}.$

Initial values can affect numerical answers

For equations with several roots, a numerical solver may return the root nearest its starting value. For

$\sin x=0.4,\qquad 0\leq x<2\pi,$

one output is not the complete solution. The solutions are

$x=\sin^{-1}(0.4)=0.4115\ldots$

and

$x=\pi-0.4115\ldots=2.7301\ldots.$

Always combine technology with the stated interval and the structure of the function. See trigonometric equations.

Statistics technology

Statistical functions reduce arithmetic, but the input convention and interpretation still matter.

Worked example: summary statistics from a list

For the data

$4,\ 6,\ 6,\ 9,\ 10,$

enter one value per observation. The mean is

$\bar x=7.$

The sum of squared deviations is

$$\sum(x-\bar x)^2 =(-3)^2+(-1)^2+(-1)^2+2^2+3^2 =24.$$

The population standard deviation is therefore

$\sigma=\sqrt{\frac{24}{5}}=2.19089\ldots,$

while the sample standard deviation is

$s=\sqrt{\frac{24}{4}}=\sqrt6=2.44949\ldots.$

These answer different questions. Use $\sigma$ when the entered data are the whole population being described, and $s$ when the data are a sample used to estimate population variation.

For a frequency table, enter frequencies in the frequency list. Entering each distinct value once calculates statistics for a different data set.

Distribution functions

Distinguish cumulative probability from point probability. For a binomial variable

$X\sim\operatorname{B}(20,0.3),$

the event $X\leq4$ is cumulative, while $X=4$ is a single value. For a continuous variable, by contrast,

$P(X=4)=0,$

and probabilities are areas over intervals. Write the required probability before choosing a calculator function.

Continue with statistics foundations and the relevant probability distribution lesson when those topics are introduced.

Spreadsheets and computer algebra

Spreadsheets are useful for repeated calculations, recurrences, simulations and large data sets. A formula should refer to cells consistently and units should appear in headings. Test a formula on a case you can calculate by hand before filling it down hundreds of rows.

Computer algebra systems can expand, factorise, differentiate and integrate symbolically. Their results may be written in an unfamiliar but equivalent form. For example,

$\frac12\ln(x^2+1)+C$

and

$\ln\sqrt{x^2+1}+C$

are equivalent over the real numbers because $x^2+1>0$. Verify an antiderivative by differentiating it. Do not submit unexplained software output where reasoning or a specified method is required.

Technology also inherits your assumptions. A regression model may calculate perfectly yet be unsuitable for extrapolation. A simulation may be too small to reveal a rare event. A symbolic simplifier may use domain assumptions you have not checked.

A reliable technology workflow

For every technology assisted calculation:

1. State the mathematics. Write the equation, formula, probability or function.
2. Predict the result. Estimate sign, size, number of solutions or graph shape.
3. Check settings and data. Confirm angle unit, mode, interval, lists and frequencies.
4. Compute with full precision. Use brackets and retain stored values.
5. Interrogate the output. Check domain, units, interval and plausibility.
6. Present the answer. Show enough method, round only at the end and conclude in context.

Self-check 2

1. A graph shows no intersection. Give two reasons why an intersection might still exist.
2. A solver gives $x=1.24$ for an equation on $0\leq x\leq10$. What three checks should follow?
3. For measured data, when might the sample standard deviation be more appropriate than the population standard deviation?
4. Why does a numerical check at $x=2$ not prove an identity?

Answers

1. The viewing window may exclude it, or the graphing resolution may miss it. An asymptote or poorly chosen scale can also obscure it.
2. Substitute $1.24$ into the original equation, confirm that it lies in the required interval and search systematically for any other solutions. Also check whether rounding meets the requested accuracy.
3. When the entered observations are a sample used to estimate the variability of a larger population.
4. An identity must hold for every value in its domain. One test can disprove a proposed identity if it fails, but cannot prove it if it passes.

Common mistakes

• Treating a decimal output as exact.
• Using degree mode for an angle expressed using $\pi$.
• Rounding intermediate values and then reusing them.
• Trusting a graph without checking its window and scale.
• Reporting one numerical root when the interval contains several.
• Accepting solver roots without substitution into the original equation.
• Confusing population and sample standard deviation.
• Giving calculator output instead of the method requested.
• Copying more decimal places than the data or question justify.

Next steps

Strengthen the checking stage with checking answers and communicating reasoning. Then apply these habits in mathematical problem solving, mathematical modelling and numerical methods.