Chapter IX · The logarithmic, exponential and circular functions of a real variable
as a limit:
The compound-interest limit , proved twice, once from and once by an explicit squeeze; and the identification of with the limit of Chapter IV.
Why this matters
In §73 Hardy proved that tends to a limit, and called it “provisionally”. Now we have a second , defined by . §215 shows they are the same number, and more: for every real . This is the formula behind compound interest, population growth and the probability of “no success in tries”, and it is often taken as the definition of . Hardy gives two proofs; the second gives explicit error bounds.
The idea
Taking logarithms turns the power into a product:
As , , and (the slope of at ). So the logarithm of tends to , and the power itself tends to .
The second proof is more concrete. The basic inequality of the first lesson, read through the exponential, says . Applied with and with it traps between two explicit expressions.
Try this. Start with and drag the point along . Then increase to , , and watch the green band between the two outer curves.
- the squeeze
What to notice
The lower curve always stays below , the upper curve always above, as long as . The band narrows roughly in proportion to : multiplying by divides the gap at by about . Convergence is slowest where is large: there is large.
Try this. With , find the needed for . Then set or .
- band l ± δ
- l
- n₀ (drag)
- n ≥ n₀, good
- n ≥ n₀, bad
What to notice
For the terms creep up to from below: you need around for , because the error is about . For larger the error is larger, roughly .
Precise statements
For every real ,
and more generally as the real variable or . In particular : the number of §73 is the number defined by .
If and , then
and if also , the difference between the outer terms is less than .
Proofs
Take logarithms, use the known limit , and return through the continuity of the exponential.
The key limit
as (first lesson, Examples LXXXIII, 3). For the theorem is trivial, so let .
Put
For we have , and
As , , so this tends to .
Exponentiate
by the definition of a general power. The exponent tends to and is continuous, so the power tends to .
Special cases
Letting run through the positive integers gives ; letting it run through the negative integers gives .
Use for , which is the inequality in exponential form, once with and once with . Then estimate the ratio of the outer terms with Bernoulli’s inequality.
for every
For , (first lesson); apply the increasing function . For the left side is .
Lower bound
With : . Since , the left side is positive, and raising both sides to the -th power keeps the order: .
Upper bound
With : , so . Taking reciprocals reverses the inequality: .
The ratio of the outer terms
Bernoulli’s inequality (§74), with and , gives , so the ratio is less than .
The gap
This tends to as , and lies between the two outer terms, so both tend to .
Hardy reaches the squeeze through the integral inequalities , obtained by comparing with ; with they are exactly the two bounds above. That route is the subject of the next lesson. He proves the case and leaves to the reader; the argument above covers both at once, because it only uses .
Worked examples and exercises
Use the squeeze with and to bound . How large must be for the upper bound to fall below ?
Hint
Solution
and , so . The book prints the upper bound as ; with the method gives only . The upper bound falls below first at : , while . Either way the width of the bracket shrinks only like : the error bound with is .
Show that if as , then .
Hint
Solution
Since , ; in particular for large . If ,
and the second factor tends to because . When the left side is . In all cases , and by continuity of the exponential . The theorem is the case .
Show that if then ; and if and then . Deduce that and .
Hint
Solution
First part. For large , . If , then , so . If , then . In either case , so .
Second part. (or when ), so , and .
Examples. with and , so . Directly: it equals . And has , , so it tends to .
Deduce from the inequality that tends to infinity faster than any power of .
Hint
Solution
Let be given and fix an integer . For , . So as , because .
A sum of money earns interest at per year, compounded times a year. Show that, however often the interest is compounded, the sum is multiplied by less than in a year, and that monthly compounding already gives more than times.
Hint
Solution
By the squeeze with , for every . Monthly: . Daily: , already within of , in line with the error estimate .
Check yourself
Since and , we have .
False. A limit of a power cannot be taken base first when the exponent also changes: the base is never exactly , and the small excess is raised to an ever larger power. The logarithm , not , so the limit is . Forms like "" are indeterminate.
Compute to four decimal places. (For comparison, .)
, below as the squeeze predicts. The gap is below the bound (that bound is for the whole bracket, so it is generous).
For a fixed , how do the sequences and approach ?
The squeeze puts below and above it for every . (In fact the first increases with and the second decreases, as for in §73.) This holds for negative too, as long as .
Roughly how does the error behave for large ?
, so . The error is about : halving it costs doubling . The squeeze’s bound is about twice the truth.
Hardy’s language
Summary
- and for every real ; with this identifies the of §73 with the of .
- Proof 1: , then continuity of .
- Proof 2: gives for , with gap below .
- More generally, whenever .