Solution
- Let be an element outside , that is, there exist and such that, for every , we have . In other words, there exists an interval of radius around into which no term of the sequence enters after some rank. But then, if we take for instance the interval of radius around , this interval is entirely contained in the complement of . Indeed, if belongs to this interval, then by the reverse triangle inequality, for , we have:
which shows that cannot belong to .
This shows that the complement of is open in , and therefore is closed.
- Let and belong to , and without loss of generality assume that . Let . We need to show that belongs to . Let therefore and . By assumption, there also exists such that, for every , the sequence moves in small successive steps:
Without loss of generality, assume that (otherwise, redefine ). Since is a subsequential limit, there exists such that
Likewise, since is a subsequential limit, there exists (!) such that .
- If , then we are done, since .
- If , then we are done, since .
- Otherwise, we have .
But then, between the indices and , since the sequence moves in small successive steps of size at most and eventually exceeds , there must necessarily exist an intermediate index between and such that . This is exactly what we wanted to prove!
For the second part of the question, consider the following sequence, given by the partial sums of the harmonic series:
This is a classical example of a sequence for which the difference between successive terms tends to , but which nevertheless diverges. Since the sequence is increasing and diverges, it has no cluster point. Therefore is empty in this case.

No messages yet.