P3.3 - Exponentials and logarithms

Syllabus
2019
Topic
P3.3
Level
A2

Learning objectives

Read every feature of a transformed exponential graph

The graph of y=exy=e^x is always positive, passes through (0,1)(0,1) and approaches the horizontal asymptote y=0y=0 as xx\to-\infty. Its domain is R\mathbb R and its range is (0,)(0,\infty).

For y=eax+b+cy=e^{ax+b}+c, with a0a\ne0:

Feature Result
horizontal asymptote y=cy=c
domain and range xRx\in\mathbb R; y>cy>c
yy-intercept (0,eb+c)(0,e^b+c)
direction increasing if a>0a>0; decreasing if a<0a<0

Writing ax+b=a(x+b/a)ax+b=a(x+b/a) shows that the inside change moves and horizontally scales the base graph; adding cc moves every output vertically and therefore moves the asymptote from y=0y=0 to y=cy=c.

For y=e2x+1+3,y=e^{-2x+1}+3, the asymptote is y=3y=3, the range is y>3y>3, and the curve decreases because the coefficient of xx is negative. At x=0x=0, y=e+3y=e+3, so (0,e+3)(0,e+3) fixes its vertical position.

The curve approaches its asymptote but never reaches it because eax+b>0e^{ax+b}>0. Do not set the asymptote to y=0y=0 after an outside shift, and do not use bb alone as a horizontal translation without first factoring aa.

Use natural logarithms to undo exponentials

The natural logarithm is the inverse of the exponential function: ln(ex)=x\ln(e^x)=x and elnx=xe^{\ln x}=x for x>0x>0. Therefore the graphs of y=exy=e^x and y=lnxy=\ln x are reflections in y=xy=x.

Feature of y=lnxy=\ln x Value
domain x>0x>0
range R\mathbb R
intercept (1,0)(1,0)
vertical asymptote x=0x=0

eax+b=px=lnpba(p>0),ln(ax+b)=qx=eqba(ax+b>0),a0e^{ax+b}=p\Rightarrow x=\frac{\ln p-b}{a}\quad(p>0),\qquad \ln(ax+b)=q\Rightarrow x=\frac{e^q-b}{a}\quad(ax+b>0),\qquad a\ne0

For e3x1=7e^{3x-1}=7, take natural logs to get 3x1=ln73x-1=\ln7, so x=(1+ln7)/3x=(1+\ln7)/3. For ln(52x)=1\ln(5-2x)=1, exponentiate to obtain 52x=e5-2x=e, hence x=(5e)/2x=(5-e)/2; this satisfies the required domain 52x>05-2x>0.

A logarithm accepts only a positive argument. Do not write ln(u+v)=lnu+lnv\ln(u+v)=\ln u+\ln v, and do not take lnp\ln p when an equation has p0p\le0; in that case the real exponential equation has no solution.

Linearise a model to estimate its parameters

A logarithmic plot turns a nonlinear relationship into a straight line. The chosen horizontal axis determines which model is being tested, while the gradient and intercept recover the original parameters.

Original model Straight-line form Plot Gradient Intercept
y=axny=ax^n logy=nlogx+loga\log y=n\log x+\log a logy\log y against logx\log x nn loga\log a
y=kbxy=kb^x logy=xlogb+logk\log y=x\log b+\log k logy\log y against xx logb\log b logk\log k

Using base-10 logarithms, an intercept cc means a=10ca=10^c or k=10ck=10^c. For the exponential plot, a gradient mm means b=10mb=10^m. With natural logarithms, use ece^c and eme^m instead; keep one base throughout.

Suppose a graph of log10y\log_{10}y against log10x\log_{10}x has gradient 12-\frac12 and intercept 22. Then log10y=12log10x+2,\log_{10}y=-\frac12\log_{10}x+2, so y=102x1/2=100x1/2.y=10^2x^{-1/2}=100x^{-1/2}.

Estimate the gradient from two well-separated points on the straight line of best fit, not automatically from two raw data points. Read the vertical intercept at horizontal coordinate zero, then convert the logged parameters back to the original scale.

Both logarithmic axes require positive logged values. Plotting logy\log y against xx tests an exponential form; plotting it against logx\log x tests a power form. Interchanging these axes changes the model and the meaning of the gradient.