Since the complex functions
can be represented as unit
vectors in the complex plane
, the partial sums of the series
, which are given by
can be represented very simply and elegantly on the complex plane, because
the direction of each vector
is given solely by
the unit-length basis element
, while its length is
given solely by
. The partial sum
is a sum of
such vectors, which can be represented by drawing the vectors
consecutively on the plane, starting from the origin, forming in this way
a chain of complex vectors, formed by segments connected end-to-end at
certain points, as illustrated in Figure 1. The complete
series is represented by an infinite chain of such vectors. The set of
vertex points along this infinite chain is identical to the set of partial
sums of the series. Therefore, the convergence of the series is translated
in this geometrical representation onto the fact that this sequence of
points has a limit, and approaches some fixed point of the complex plane.
It now becomes immediately clear that the absolute convergence of the series is equivalent to the complete chain of vectors having a finite total length. One can see this because the series is absolutely convergent if and only if the sum of the moduli of the coefficients is finite, since the basis elements have unit moduli. In short, absolute converge means that the sum
converges to a finite limit when . Since the modulus of each
coefficient gives the length of one of the vectors in the chain, this sum
is the total length of the chain. If the series is absolutely convergent,
then this sum is finite and so is the length of the chain. Likewise, if
the chain has a finite length then this sum is finite and the series is
absolutely convergent. If the series
is absolutely convergent,
then it is also convergent, and hence both the cosine and the sine series,
its real and imaginary parts, are convergent.
This relation is intuitively clear, since it is obvious that, if the chain has a finite total length, then it cannot extend to infinity, nor can it oscillate indefinitely between two points. It must extend a finite amount and hence stop at some point of the complex plane, which is its point of convergence. In this paper we will concern ourselves mostly with the case in which the series is convergent, but not absolutely convergent. In this case the chain has infinite total length, and then the series may fail to converge by extending to infinity or by oscillating indefinitely.
It is clear that the point presents a particular convergence
behavior, since for this value all the basis elements become equal to one,
and hence the chain extends only over the real axis of the complex plane.
It the series is not absolutely convergent, the chain has infinite length,
and may easily extend to infinity. This is not certain only due to the
possibility of successive exchanges of the signs of
, since the
cancellations thus introduced may make the series converge, with the chain
folding repeatedly over itself.
It is important to recall at this point that, if the series is not
absolutely convergent, then the order in which the sum is executed may
affect the final result. In this case, any statement of convergence must
state the order in which the summation is meant to proceed. When we draw
the chain in the complex plane, we are automatically adopting a particular
order, the natural order of the sequence, with increasing values of the
index . This order will always be assumed here, unless otherwise noted.
Next, we will prove a theorem concerning the convergence of the complex
series , that involves a criterion based on the behavior of the
real coefficients
.