Logarithms

You know the inverse of many mathematical operations; we can undo an addition by 2 for
example by subtracting 2. But is there an inverse function for an exponential function?

Function Inverse

12
4 𝑥 ×3 𝑥÷
? 3 4
7
4 𝑥+ 3 𝑥−
? 3 4
16
4 𝑥
2
√ ?𝑥 4
1024
4 𝑥
5
√ ?𝑥
5
4
81
4 3
𝑥
log?3 𝑥 4
Such functions are known as logarithms, and exist in
order to provide an inverse to exponential functions.
Interchanging between exponential and log form
! (“said log base of ”) is equivalent to .
The log function outputs the missing power.

3 =9
2 log 3 9= 2
Here are two methods of interchanging between these forms.
Pick your favourite!
Method 2: Do same operation
Method 1: ‘Missing Power’
to each side of equation.
• Note first the base of the log must Since KS3 you’re used to the idea of doing the same
match the base of the exponential. thing to each side of the equation that ‘undoes’
• for example asks the question “2 to whatever you want to get rid of.
what power gives 8?”
We can imagine inserting the (−2 ) (−2 )
output of the log just after
the base. Click the button!
“log base ” undoes “ to the power of” and vice versa,

𝑙𝑜𝑔8¿ 3 ¿
3 Click to start
as they are inverse functions.

2 Fro-manimation (2□ ) (2□ )

Examples
log 5 25=𝟐 1
( )
Think: “5 to
? the power of log 3 =−?𝟑
what gives
27
log 3 81=𝟒
?
you 25?”

log 2 32=𝟓 ? log 2

1
16
=−𝟒?
( )
log 10 1000=𝟑 ?
log 𝑎 ( 𝑎 )=𝟑
3 ?
log 4 1=?𝟎! for all .
log 4 ( −1 ) =¿NOPE
?!!??
log 4 4 =𝟏 ?
While a log can output a negative number, we can’t log

()
1 negative numbers.

log 2 =−𝟏
? Strictly Just For Your Interest: However, if we were to expand the
range (i.e. output) of the log function to allow complex numbers

2 (known as the ‘complex logarithm’), then we in fact get . It’s

probably better if you purge these last few sentences from your
memory and move along…
With Your Calculator
There are three buttons on your calculator for computing logs:

I couldn’t think of a

log □ □ log 3 7=𝟏.𝟕𝟕𝟏𝟐𝟒…

?
? word that rhymed
with ‘ln’ so I recorded
it for you.
of

𝑙𝑛ln10=𝟐.𝟑𝟎𝟐𝟓𝟖…
?
? is the “natural log of ”,
meaning “log to the base ”,

𝑙𝑜𝑔
i.e. it the inverse of .
log 100=𝟐
?
We will use it more
extensively later this chapter.
Just like the symbol without a number is by default,
without a base is base 10 by default when used on
your calculator (although confusingly “” can mean “”
in mathematical papers)
Laws of Logs
! Three main laws:
The logs must have a
Special cases: consistent base.

i.e. You can move the

Not in syllabus (but in MAT/PAT): power to the front.

We often try to avoid

leaving fractions inside logs.
So if the answer was:

You should write your

answer as:
Reciprocating the input
negates the output.

This is known as changing the base. So

to get in terms of log base 3:
Examples
Write as a single logarithm: Write in terms of , and
a. a.
b. b.
c.
c. d.

a ? a
?
b ?
c ? b
?
d ?
c
?

d ?
 Anti Laws
These are NOT LAWS OF LOGS, but are mistakes students often make:

log 𝑎 ( 𝑏+ 𝑐 )=log 𝑎 𝑏+ log 𝑎 𝑐 There is no method to

?
simplify the log of a sum,
only the sum of two logs!
 FAIL
3
log 2 𝑥 ) =3 log 2 𝑥 The power must be on the input
?
(here the ), but here the power is
around the entire log.

 FAIL
Solving Equations with Logs
Solve the equation
This is a very common type of exam question.
The strategy is to combine the logs into one and isolate on one side.

2
We’ve used the laws of logs to combine

log 10 4+log10 𝑥 =2
them into one.

Use your favourite method of rearranging.

Either do “10 the power of each side” to

? “undo” the log, or the “insert the 2

between the 10 and ” method.

The subtle bit: You must check each value in the

original equation. If , then we’d have but we’re not
allowed to log a negative number.
Solving equations with exponential terms
Solve

Applying to each side of the equation: This is often said “Taking

logs of both sides…”
?

Solve

Applying to each side of the equation:

?
Solving equations with exponential terms
Solve

Why can we not apply quite the same strategy here?

Because the exponential terms don’t have the same base, so we can’t apply the same log.
We ‘take logs of’/apply log to both sides, but we need not specify a base. on its own may
?
either mean (as per your calculator) or (in academic circles, as well as on sites like
WolframAlpha), but the point is, the base does not matter, provided that the base is
consistent on both sides.

Logs in general are great for solving equations when

𝑥 𝑥+1
the variable is in the power, because laws of logs

log3 =log2
allow us to move the power down.

This then becomes a GCSE-style ‘changing the

? Solution
subject’ type question. Just isolate on one side and
factorise out.

It doesn’t matter what base you use to get the final

answer as a decimal, provided that it’s consistent. You
may as well use the calculator’s ‘log’ (no base) key.
Test Your Understanding
1 Solve , giving your answer 3 Solve , giving your answer in
to 3dp. exact form.

2 𝑥−1=log 3 5 which could be simplified to:

? ?

2 Solve , giving your answer to 3dp.

𝑥+1 𝑥−1 𝑥 ( log4−log3) =log3+log4

log3 =log 4 ?
Natural Logarithms
We have previously seen that is the inverse function of .
We also saw that is “the” exponential function.
The inverse of is , but because of its special importance, it has its own
function name!

! The inverse of is
Since “ to the power of” and
𝑥 ?
ln 𝑒 =? 𝒙
“ of” are inverse functions,
they cancel each other out!

Solve Solve

ln 𝑥=2
“ln both sides”.
𝑥 =ln?5 On the LHS it
cancels out the “
? Do “ to the power
to the power of” of” each side. On
the LHS it cancels
out the ln.
Quadratics in
In previous chapters we’ve already dealt with quadratics in disguise, e.g.
“quadratic in sin”. We therefore just apply our usual approach: either make a
suitable substitution so the equation is then quadratic, or (strongly
recommended!) go straight for the factorisation.

Solve Solve

Note that therefore First write negative powers as

fractions:
Exponential functions are always
positive therefore:
? ?
Test Your Understanding

Solve Solve

2 2𝑥 𝑥
3 𝑥+1=𝑒 ?
𝑒 +5𝑒 −6=0 ?

Solve giving your answer as an exact value.

𝑥 𝑥+1
ln2 𝑒 =ln3 ?
 Graphs for Exponential Data
In Science and Economics, experimental data often has exponential growth, e.g.
bacteria in a sample, rabbit populations, energy produced by earthquakes, my Twitter
followers over time, etc.
Because exponential functions increase rapidly, it tends to look a bit rubbish if we
tried to draw a suitable graph:
But suppose we took the log of the number of transistors
for each computer. Suppose the number of transistors one
Take for example “Moore’s Law”, which hypothesised year was , then doubled 2 years later to get .
that the processing power of computers would When we log (base 2) these:
double every 2 years. Suppose we tried to plot this
for computers we sampled over time: The logged value only increased by 1! Thus taking the log
of the values turns exponential growth into linear growth
(because each time we would have doubled, we’re now
just adding 1), and the resulting graph is a straight line.
Number of transistors

If we tried to force
all the data onto the
graph, we would end

log(transistors)
up making most of
the data close to the
horizontal axis. This
is not ideal.

Year Year
1970 1980 1990 2000 1970 1980 1990 2000
Graphs for Exponential Data

Because the energy involved in earthquakes

decreases exponentially from the epicentre
of the earthquake, such energy values
recorded from different earthquakes would
vary wildly.

The Richter Scale is a logarithmic scale, and

takes the log (base 10) of the amplitude of
the waves, giving a more even spread of
values in a more sensible range.
(The largest recorded value on the Richter Scale is 9.5 in Chile in
1960, and 15 would destroy the Earth completely – evil scientists
take note)

The result is that an earthquake just 1

greater on the Richter scale would in fact be
10 times as powerful.
Richter Scale
Other Non-Linear Growth

𝑦 3 We would also have similar graphing

𝑦 =2 𝑥
problems if we tried to plot data that
followed some polynomial function such
as a quadratic or cubic.

We will therefore look at the process to

convert a polynomial graph into a linear
one, as well as a exponential graph into a
linear one…
𝑥
Turning non-linear graphs into linear ones
Case 1: Polynomial Linear Case 2: Exponential Linear
Suppose our original model was a Suppose our original model was an exponential one:
polynomial one*:
Then taking logs of both sides:
Then taking logs of both sides:
Again we can compare this against a straight line:
We can compare this against a straight line:

 If , then the graph of against will be  If , then the graph of against will be
a straight line wih gradient and a straight line with gradient and
vertical intercept . vertical intercept .

log 𝑦 log 𝑦

log 𝑎 log 𝑎
log 𝑥 𝑥
* We could also allow non-integer ; the term would The key difference compared to Case 1 is that we’re only logging the values (e.g. number
then not strictly be polynomial, but we’d still say the of transistors), not the values (e.g. years elapsed). Note that you do not need to
function had “polynomial growth”. memorise the contents of these boxes and we will work out from scratch each time…

In summary, logging the -axis turns an exponential graph into a linear one.
Logging both the and -axis turns a polynomial graph into a linear one.
Example
[Textbook] The graph represents the growth of a population of log ⁡( 𝑃 )
bacteria, , over hours. The graph has a gradient of 0.6 and meets
the vertical axis at as shown.
A scientist suggest that this growth can be modelled by the
equation , where and are constants to be found.
a. Write down an equation for the line. 2
b. Using your answer to part (a) or otherwise, find the values of 𝑡
and , giving them to 3 sf where necessary.
c. Interpret the meaning of the constant in this model.

a log 𝑃 =0.6 𝑡 +2 Equation of straight line is where here:

?
b Just like on previous slide, start with the model then log it:

Comparing with our straight line equation in (a): Recall that means
?

c gives the initial size of the bacteria population.

?
Recall that the coefficient of an
exponential term gives the ‘initial value’.
Example
[Textbook] The table below gives the rank (by size) and population of the UK’s largest cities and districts
(London is number 1 but has been excluded as an outlier).
City B’ham Leeds Glasgow Sheffield Bradford
Rank, 2 3 4 5 6
Population, 1 000 000 730 000 620 000 530 000 480 000
Textbook Error: They use
The relationship between the rank and population can be modelled by the formula: but then plot against .
where and are constants.
a) Draw a table giving values of and to 2dp.
b) Plot a graph of against using the values from your table and draw the line of best fit.
c) Use your graph to estimate the values of and to two significant figures.

c First get equation of straight line:

a 0.30 0.48 0.60 0.70 0.78
6 5.86 ? 5.79 5.72 5.68 (reading from the graph)

b Let’s use these points on

As with previous example, let’s log the original
model so we can compare against our straight line:
log ⁡( 𝑃 ) the line of best fit to
determine the gradient. ?
6.4 ( 0.05 ,6.16 ) Comparing this with our straight line:
6.0 ( 0.77 ,5.68 )
5.6 ?
5.2

0.2 0.4 0.6 0.8

log 𝑅
 Test Your Understanding
Reflections: Consider what we’re
Dr Frost’s wants to predict his number of Twitter followers (@DrFrostMaths) doing in this whole process in case you
years from the start 2015. He predicts that his followers will increase don’t understand why we’re doing all
exponentially according to the model , where are constants that he wishes to of this:
1. We want to find the parameters
find.
of a model, e.g. that best fits
He records his followers at certain times. Here is the data: the data (in this case the
parameters we want to find are
Years after 2015: 0.7 1.3 2.2 and ).
Followers : 2353 3673 7162 2. If the data had a linear trend,
then this would be easy! We
a) Draw a table giving values of and (to 3dp).
know from KS3 that we’d just
b) A line of best fit is drawn for the data in your new table, and it happens to go plot the data, find the line of
through the first data point above (where ) and last (where ). best fit, then use the gradient
Determine the equation of this line of best fit. (The -intercept is 3.147) and -intercept to work out the
c) Hence, determine the values of and in the model. and in our linear model.
d) Estimate how many followers Dr Frost will have at the start of 2020 (when ). 3. But the original data wasn’t
linear, and it would be harder to
draw an ‘exponential curve of
best fit’.
4. We therefore log the model so
a 0.7 1.3 2.2 b
3.855− 3.372 that the plotted data then

3.372 ?
3.565 3.855 𝑚= ? =0.322 roughly forms a straight line, and

2.2− 0.7
then we can then draw a
(straight) line of best fit.
5. The gradient and -intercept of
this line then allows us to
c d estimate the parameters and in
𝑡
𝑃=𝑎 𝑏
the original model that best fit
? When ? the data.
The process of finding parameters in a
model, that best fits the data, is known
as regression.
The End