Chapter IX · The logarithmic, exponential and circular functions of a real variable
Logarithmic tests of convergence
Series such as that comparison with cannot decide, settled by the integral test with and ; and Euler's constant .
Why this matters
Chapter VIII decided every series : convergent for , divergent for . But the scale of infinity has room between and every : for example and tend to faster than and more slowly than every . Comparison with powers of is powerless on them. §218 settles them with the integral test of §180 and the integrals of the scale-of-infinity lesson, and shows that the dividing line reappears, at exponent , on every rung of the logarithmic scale.
The idea
The integral test says that, for a positive decreasing , the series and the integral converge or diverge together, and their difference stays bounded. And the integrals we need are easy, because is :
So the question for is the same as the question for , the continuous version of , with in place of . The answer is the same: convergence exactly when . What changes is the speed. Where grows like , the series grows like .
Try this. Start with : the dots are exact partial sums of up to . Widen the view to and beyond, and read how large must be for the sum to pass . Then set and , and try the second family.
- exact partial sums (N ≤ 10⁶)
- the integral from the first term to N
- integral + constant (beyond 10⁶)
What to notice
With the sum after a million terms is only , and after terms about ; to pass you need about terms. Yet it diverges: the dashed curve is unbounded. With it levels off at a finite sum. For the family at , the growth is like , slower still.
Precise statements
Let and . The series and the integral
are convergent if and divergent if .
Let . If for large , with a constant and some , then converges. If for large , with , then diverges.
Write , , and so on. For each , the series and integral
converge if and diverge if ; here and must be large enough that all the logarithms involved are positive.
How large is “large enough”? needs ; needs ; needs ; the next needs .
Proofs
Apply the integral test to . Compute the integral exactly by the substitution , and read off when it stays bounded.
is positive and eventually decreasing
For , . Writing , the exponent has derivative , negative once . So decreases for large , which is all the integral test needs.
Compute the integral up to
With , :
Let
If , and the integral tends to : convergent. If , ; if , : divergent.
Transfer to the series
By the integral test (§180), converges exactly when the integral does. (The first few terms, before starts to decrease, do not affect convergence.)
Compare term by term with the series just decided.
Convergence
If for with , then the partial sums of are bounded by those of a convergent series, so they converge (Chapter VIII).
Divergence
If for and , the partial sums of exceed times those of the divergent .
The same substitution, one level deeper each time: the derivative of is exactly .
Differentiate
By the chain rule (and induction on ), .
Substitute
The integral becomes between and . Since , it converges exactly when .
Series
The integrand is positive and eventually decreasing (each factor in the denominator eventually increases), so the integral test applies as before.
Try this. Compare (blue) with (ochre) over terms, then look at the terms instead of the sums.
- sₙ for
- sₙ for
What to notice
Term by term the two series are hard to tell apart, and after terms the divergent one is not even far ahead. One converges (to about ) and the other diverges, but no finite number of terms can show it: only the theorem can.
Worked examples and exercises
Let . Show that converges and diverges, whatever the value of .
Hint
Solution
Convergence. By the scale of infinity, , so for large . Then , the term of a convergent series.
Divergence. Likewise for large , so for large (as ), and diverges.
Logarithmic factors matter only when the power of is exactly ; that is why a finer test is needed there.
Decide whether and converge.
Hint
Solution
The first is the case , of the finer test: divergent. (Its partial sums grow like .)
For the second, once , so from there on the term is at least , the case , : divergent by comparison. (Hardy notes that neither series can be decided by the test with alone, since their terms lie between and .)
Show that increases and tends to a limit with .
Hint
Solution
On , , so each bracket satisfies . Hence is a sum of positive terms, increasing in , and (the sum telescopes). An increasing bounded sequence has a limit , and . Its value is , Euler’s constant; whether it is rational is a famous open problem. Equivalently, .
Show that .
Hint
Solution
. By Euler’s limit, with , so . Since has the same limit, the series converges to . (The constant cancels: we never need its value.)
Show that diverges.
Hint
Solution
(from Euler’s limit, or from ). For , , so and . The comparison form of the test gives divergence. This is a special case of Examples XC, 6: if diverges, so does .
Check yourself
Which of these series converges?
At the level of , the exponent of decides: convergence exactly when it is greater than in the denominator. () and () diverge; is even larger than . Only converges, though extremely slowly.
After terms the partial sums of have reached only about and are growing very slowly, so the series converges to about .
False. The series diverges by the logarithmic test: its partial sums behave like , which is unbounded. It passes near and near . Numerical evidence cannot distinguish such slow divergence from convergence.
Euler’s constant , to four decimal places.
. With the difference is , already correct to six places, because the error is about .
Which substitution turns into ?
, so and the integral becomes , which converges. With one gets , the same problem one level up.
Hardy’s language
Conversely, the rate of increase of the higher logarithmic functions is extremely slow.
Summary
- and converge iff : the substitution turns them into .
- The same holds one level down for every iterated logarithm: converges iff .
- Divergence at is extremely slow: needs about terms to pass .
- Euler’s constant: ; hence .