Chapter IX · The logarithmic, exponential and circular functions of a real variable
The logarithmic series
Integrating the geometric series term by term gives for , and the remainder is an area we can see and bound.
Why this matters
The logarithm was defined as an area, . To compute with it we want a power series, as we now have for . But itself has no expansion in powers of (it tends to at ), so §220 expands instead. The result, , is valid only on the interval , and seeing exactly why is the main lesson: the proof is a model for §221 and for many later estimates.
The idea
Shift the integral: . For the integrand is a geometric series, . Integrating term by term would give at once. But “integrate an infinite series term by term” is exactly the kind of step this book does not take on trust.
Hardy’s trick is to stop the geometric series after finitely many terms and keep the exact remainder. For any and any ,
which you can check by multiplying through by . Only finite sums are integrated, and the whole question becomes: how big is the integral of the last piece?
Try this. Start with and , then add terms. Next set close to , then try negative near , and finally push past .
- the first terms of the geometric series
- area between them over = error
What to notice
The red area between the curve and its truncated geometric series is exactly the error of the partial sum. For it shrinks steadily as grows, slowly when . For near it shrinks too, but needs many terms. For the polynomial swings wildly beyond and the area explodes: no number of terms helps.
Precise statements
If then
If or the series does not converge. In particular,
The identity behind it is exact for every and every :
Proofs
Integrate the finite geometric sum with its exact remainder from to , then show that the remainder integral tends to , separately for and for . Outside the interval, show that the terms do not tend to .
The exact identity
Integrate the identity for from to ; this is legitimate for , since then on the whole range. The integral of is , which gives the identity above with .
:
On we have , so . Integrating, , which tends to .
: put with
Substitute : . On we have , so
The factor is large when is near , but it is fixed while .
Outside the series diverges
If , then (powers beat ), so the terms do not tend to and the series cannot converge. If , the series is , the harmonic series, which diverges to . (It must: as .)
At the bound is honest about how slow the series is: the error after terms really is about . Compare it with a point inside the interval.
Try this. This compares the errors of the series at (blue) and at (ochre) on a log scale. It starts with 30 terms; set Terms to .
- |s − sₙ| for
- |s − sₙ| for
What to notice
At the error falls in a straight line on the log scale, like a geometric series: each term gains a fixed number of digits. At it creeps down like : a hundred terms give only two decimals of . Hardy calls the series at practically useless for computation, and Exercise 5 below shows a better way.
Worked examples and exercises
Show that for .
Hint
Solution
. The theorem applies to exactly when , that is, , and gives .
Show that lies between the sum of the first and the sum of the first terms of .
Hint
Solution
At the identity reads with . For this gives ; for it gives . (This is also the alternating series test of §195, but here it falls out of the formula.)
Show that, for , . Why is this useful even for large ?
Hint
Solution
With we have and , so . By Exercise 1 this is . Unlike the original series, this one converges for every (indeed for every , where ), though slowly when is large, since then is close to .
Find .
Hint
Solution
With and , the logarithmic series gives . Also , so by the binomial series and the denominator is . The ratio is .
Show that, for ,
and use it with to find to three decimals.
Hint
Solution
For , lies in and . Both and are given by the logarithmic series, and subtracting, the even powers cancel:
For , : . The neglected terms are positive and less than , so and (in fact ). Four terms here beat a thousand terms of .
Check yourself
For which does converge to ?
At the remainder is at most , so the series converges (to ). At it is minus the harmonic series and diverges. For the function is perfectly well defined, but the terms do not tend to , so the series cannot represent it there: a function and its power series are different objects.
At Hardy’s bound is . What is the smallest for which this bound is less than ?
means , so . (For the bound equals exactly.) A thousand terms for three decimals: this is why the series is not used to compute .
The series for converges absolutely.
False. The series of absolute values is the harmonic series, which diverges. The logarithmic series converges absolutely for but only conditionally at ; that is why its sum can be changed by rearranging it (Misc. Examples IX, 19, on the problems page).
In the proof for , the bound for contains the factor , which is huge when is close to . Why does the proof still work?
For a fixed the factor is just a (possibly large) constant; it does not grow with . So the bound is a constant times , which tends to . The largeness of the constant only means that more terms are needed near , as the widget showed.
Hardy’s language
It is natural to expect that log (1 + x) will be equal, when −1 < x < 1, to the series obtained by integrating each term of the series 1 − t + t² − … from t = 0 to t = x.
Summary
- exactly when ; at it gives .
- Method: integrate a finite geometric sum with its exact remainder; the error is , an area.
- Bounds: for , and for .
- Convergence is geometric inside the interval but only like at ; recombining series, as in , gives fast methods.