Exponentials and logarithms
P3 Topic 3, spec point by spec point (Exponentials and logarithms). Each spec point has the method, worked examples written the way the mark scheme wants them, how the marks go and the mistakes that cost the most marks.
Spec3.1The function eˣ and its graph
Know \(y = e^x\) and its graph, including graphs of the form \(y = e^{ax + b} + c\).
Why e is special
Every exponential \(y = a^x\) has a gradient proportional to its height. For one base, \(e \approx 2.718\), the gradient equals the height:
\[\frac{d}{dx}e^x = e^x\]
That makes \(e\) the natural base for growth and decay: anything that grows at a rate proportional to its size follows \(e^{kx}\).
The graph of \(y = e^{ax + b} + c\)
- Horizontal asymptote \(y = c\) (the \(e\) part is always positive and tends to 0 on one side).
- \(y\)-intercept: \(e^b + c\).
- \(a > 0\): increasing. \(a < 0\): decreasing towards the asymptote.
Reading a model
In \(V = 18000e^{-0.2t} + 4000e^{-0.1t} + 1000\):
- the initial value is at \(t = 0\): \(V = 18000 + 4000 + 1000 = 23000\);
- the long-term value: as \(t \to \infty\), both exponentials \(\to 0\), so \(V \to 1000\);
- the range for \(t \geqslant 0\) is \(1000 < V \leqslant 23000\).
key features
Sketch \(y = 3 - 2e^{-x}\) and state the asymptote and intercept.
\(x = 0\): \(y = 3 - 2 = 1\). As \(x \to \infty\), \(e^{-x} \to 0\), so \(y \to 3\). As \(x \to -\infty\), \(y \to -\infty\).
Increasing curve through \((0, 1)\) rising towards the asymptote \(y = 3\). It crosses the \(x\)-axis where \(e^{-x} = \frac32\), i.e. \(x = -\ln 1.5\).
- Ranges in models: B1 for one end correct, B1 for the full range with the correct strict and non-strict signs (WMA13A Jan 2026 Q6a).
- Sketches: M1 for the shape, A1 for the correct position, starting on the axis and not crossing the asymptote (WMA13 Jan 2025 Q5b).
Often asked as: initial value (WMA13 Oct 2024 Q8a); long-term value; range; sketch with asymptote.
Including the long-term value in the range (it is never reached).
Drawing the curve crossing its asymptote.
Spec3.2The function ln x and solving exponential equations
Know \(\ln x\) as the inverse of \(e^x\) and its graph, and solve \(e^{ax + b} = p\) and \(\ln(ax + b) = q\).
The idea
\(\ln x\) means \(\log_e x\). It undoes \(e^x\):
\[\ln(e^x) = x \qquad e^{\ln x} = x \;\;(x > 0)\]
The graph of \(y = \ln x\) is the reflection of \(y = e^x\) in \(y = x\): it passes through \((1, 0)\), has the \(y\)-axis as an asymptote, and exists only for \(x > 0\). All the log laws from P2 apply.
Solving
- \(e^{ax + b} = p\): take \(\ln\) of both sides, \(ax + b = \ln p\).
- \(\ln(ax + b) = q\): take \(e\) of both sides, \(ax + b = e^q\).
- Two exponentials like \(e^{-0.2t}\) and \(e^{-0.1t}\): since \(e^{-0.2t} = (e^{-0.1t})^2\), it is a quadratic in \(X = e^{-0.1t}\). Reject negative \(X\).
a model
\(N = 125 - 93e^{-0.109t}\) (thousands of sales after \(t\) months). Find \(t\) when \(N = 100\).
\(93e^{-0.109t} = 25 \Rightarrow e^{-0.109t} = \frac{25}{93} \Rightarrow t = \dfrac{\ln\left(\frac{93}{25}\right)}{0.109} = 12.05\).
a hidden quadratic
\(V = 18000e^{-0.2t} + 4000e^{-0.1t} + 1000\). Find the exact \(t\) when \(V = 15000\).
Let \(X = e^{-0.1t}\): \(18000X^2 + 4000X - 14000 = 0 \Rightarrow 9X^2 + 2X - 7 = 0 \Rightarrow (9X - 7)(X + 1) = 0\).
\(X > 0\), so \(X = \frac79\). \(-0.1t = \ln\frac79 \Rightarrow t = 10\ln\frac97\).
- M1 for rearranging to \(Ae^{kt} = B\), dM1 for correct log work, A1 for the value. Using 100 000 when the model is in thousands scores nothing (WMA13 Oct 2023 Q4b).
- Quadratics: M1 for forming and solving, A1 with the negative root rejected, dM1 for the ln work, A1 for the exact simplified answer (WMA13A Jan 2026 Q6c).
Often asked as: "find \(T\)" in a model; where a curve like \(y = 6\sqrt{x}\ln(4x)\) crosses the axis (\(\ln(4x) = 0 \Rightarrow x = \frac14\)) (WMA13 Jan 2025 Q9a).
\(\ln(a - b) = \ln a - \ln b\). There is no such law.
Taking \(\ln\) of each term separately: \(\ln(125 - 93e^{kt})\) is not \(\ln 125 - \ldots\). Rearrange first.
Spec3.3Using logarithmic graphs to find parameters
Turn \(y = ax^n\) and \(y = kb^x\) into straight-line graphs with logs and find the constants from the gradient and intercept.
The idea
Curves are hard to read; straight lines are easy. Taking logs turns these two models into \(Y = mX + c\):
| Model | Take logs | Plot | Gradient | Intercept |
|---|---|---|---|---|
| \(y = ax^n\) | \(\log y = n\log x + \log a\) | \(\log y\) against \(\log x\) | \(n\) | \(\log a\) |
| \(y = kb^x\) | \(\log y = x\log b + \log k\) | \(\log y\) against \(x\) | \(\log b\) | \(\log k\) |
To get the constants back, undo the log: if \(\log_{10} a = 0.7\), then \(a = 10^{0.7}\).
How to tell which model: if the graph uses \(\log y\) against \(\log x\), it's a power law. If it uses \(\log y\) against plain \(x\) (or \(t\)), it's exponential.
an exponential model
The graph of \(\log_{10} N\) against \(t\) is a straight line through \((0, 1.3)\) and \((10, 3.4)\), where \(N = pq^t\). Find \(p\) and \(q\).
\(\log_{10} N = \log_{10} p + t\log_{10} q\).
Intercept: \(\log_{10} p = 1.3 \Rightarrow p = 10^{1.3} = 19.95\).
Gradient: \(\log_{10} q = \frac{3.4 - 1.3}{10} = 0.21 \Rightarrow q = 10^{0.21} = 1.622\).
Meaning: \(p\) is the value at \(t = 0\); \(N\) grows by about 62% per unit of time.
a power law
The graph of \(\log_{10} y\) against \(\log_{10} x\) is a line through \((1, 2.2)\) and \((3, 5.2)\). Find \(y\) in the form \(ax^n\).
Gradient \(n = \frac{3}{2} = 1.5\). Intercept: \(2.2 - 1.5(1) = 0.7\), so \(a = 10^{0.7} = 5.01\).
\(y = 5.01x^{1.5}\).
- M1 for a correct expression for one constant, A1 for its value, dM1 for both expressions, A1 for both (WMA13A Jun 2026 Q4a).
- "Find the equation of the line" must be in terms of \(\log_{10} N\) and \(t\); writing the model instead scores 0 for that part (WMA13 Jun 2022 Q4a).
- Using the model: M1 for substituting, A1 for the value with a conclusion such as "yes, it raises £350 000 within 12 months" (WMA13 Jan 2024 Q3c).
Often asked as: find the line, then the constants, then interpret \(p\) or \(q\), then predict; explain why the model may not hold long term.
Using \(\log b\) as \(b\).
Taking the gradient from points read off the axes when the question gives exact coordinates.