Chapter VIII · The convergence of infinite series and infinite integrals
Series of complex terms
A complex series is two real series side by side; it is absolutely convergent exactly when the series of moduli converges, and then it behaves like a series of positive terms.
Why this matters
Power series, the subject of the next lessons, are series of complex numbers: with anywhere in the plane. Before studying them, §197 settles what convergence and absolute convergence mean for complex series, and shows that nothing really new is needed. The key fact is that absolute convergence can be tested with a single real series of positive terms, , so every test from the first half of the chapter becomes a test for complex series.
The idea
A complex series with can be pictured as a walk in the Argand plane: start at , take the step , then , and so on. The partial sum is where you stand after steps. The series converges if you home in on a point. Since a complex sequence converges exactly when its real and imaginary parts do (§87), the series converges exactly when and both converge.
There are two quite different ways to home in on a point. Either the walk has finite total length, , in which case it cannot help settling down; or it has infinite length but keeps turning back on itself, spiralling in, as a conditionally convergent real series zigzags.
Try this. Start with at and vary . Compare the distance travelled with the length of the path in the read-out as you raise the number of terms. Then set to , and to . Finally switch to and increase .
- path of partial sums (each step is one term)
- s after 200 terms
- the sum
What to notice
For the path length grows like without bound, but the path curls up onto the sum: conditional convergence. For the length itself is finite: absolute convergence. At every step points the same way and the series diverges. The exponential series always has finite length, however large is: the path first swings far out, then the steps shrink factorially.
Precise statements
The series , with (, real), converges to if its partial sums tend to ; equivalently, if converges to and converges to .
The complex series is absolutely convergent if the real series and are both absolutely convergent.
is absolutely convergent if and only if converges, where .
An absolutely convergent complex series converges, and . Its terms may be rearranged without changing the sum. Every test for series of positive terms (comparison, ratio, root, …), applied to , is a test for absolute convergence of a complex series.
Today the theorem is usually taken as the definition: is absolutely convergent if converges. Hardy’s definition and the modern one say the same thing, by the theorem.
Proofs
Compare the three positive series , and , using two elementary inequalities between a complex number and its real and imaginary parts.
Two inequalities
For every : and (a side of a right-angled triangle is at most the hypotenuse), and (the hypotenuse is at most the sum of the other two sides).
If converges
Then and converge by comparison, with the first inequality. So and are absolutely convergent: is absolutely convergent by definition.
If is absolutely convergent
Then converges, and by the second inequality and comparison, so does .
Convergence
and are absolutely convergent real series, so they converge; hence converges.
The inequality
by the triangle inequality; let .
Rearrangement
A rearrangement of rearranges and in the same way, and neither of their sums changes (§192). So the real and imaginary parts of the sum do not change.
Dirichlet’s test also carries over: if the complex partial sums of are bounded and is real and tends steadily to , the proof of §196 works word for word, with moduli in place of absolute values. That is how the series in the figure above were shown to converge for .
Worked examples and exercises
Hardy leaves this to the reader: prove that an absolutely convergent complex series converges, using the general principle of convergence for complex sequences (§87) rather than real and imaginary parts.
Hint
Solution
Let . Since converges, there is such that whenever (Cauchy’s principle for the real series). By the triangle inequality in the complex plane,
So the complex partial sums satisfy Cauchy’s principle, which is sufficient for convergence of a complex sequence. In the picture: the stretch of path after step is shorter than , so its two ends are less than apart.
Show that is convergent but not absolutely convergent, and find its real and imaginary parts as real series.
Hint
Solution
, so diverges: not absolutely convergent. For odd , ; for even , . So the real part is and the imaginary part is . Both converge by the alternating series test, so converges. (Their sums turn out to be and ; in the figure, this is , .)
Show that is absolutely convergent for every , and that is absolutely convergent for every complex .
Hint
Solution
For : , and converges.
For any : with , , so converges by the ratio test; for there is nothing to prove. By the theorem both complex series converge absolutely. This is the first example of §198.
Show that if a complex series converges then . Deduce that does not converge when .
Solution
, and both and tend to the sum , so . If then for every , so and cannot converge. (For it converges absolutely, being dominated by the geometric series .)
Check yourself
A complex series converges if and only if its real part and its imaginary part both converge.
True: a complex sequence (here, of partial sums) converges if and only if its real and imaginary parts converge, and the partial sums of the real part are the real parts of the partial sums.
Which condition is equivalent to the absolute convergence of ?
That is the theorem. The first option is plain convergence, which allows conditional convergence of the parts. The third is again only convergence of the series. The fourth is necessary for convergence but far from sufficient.
If the partial sums of a complex series stay inside a bounded region of the plane, the path they trace has finite length.
False. For (with not a multiple of ) the partial sums converge, so they are bounded, but the length is . Finite length is absolute convergence, which is stronger.
Hardy’s language
Summary
- converges if and only if and both converge.
- It is absolutely convergent if and only if converges: the path of partial sums has finite length.
- Absolutely convergent complex series converge, satisfy , and may be rearranged freely.
- So every test for positive series, applied to , tests complex series; Dirichlet’s test also carries over.