The alternating harmonic series βn(β1)n+1β converges conditionally (Leibniz gives convergence; βn1β diverges, so no absolute convergence). For the positive terms pkβ=2kβ11β and negative magnitudes qkβ=2k1β, we have βpkβ=βqkβ=+β (a corollary of the previous problem, or directly via 2kβ11ββ₯2k1β and comparison with the harmonic series).
Both blocks recur forever.
If positives were taken only finitely often, then eventually only negatives are taken: the running sum drops by qjβ each step and βqjβ=β forces tβββ β eventually tβ€s, and the rule would take a positive: contradiction.
Symmetrically for negatives (otherwise tβ+β would force a negative take).
Since each block consumes the next unused terms in order, every piβ and every qjβ is eventually used exactly once β a genuine rearrangement.
Crossing lemma: crossing upward from tβ€s by adding p, the new gap is (t+p)βs=pβ(sβt)β€p (equality only when t=s).
Crossing downward from t>s by subtracting q, the new gap is sβ(tβq)=qβ(tβs)<q.
So every crossing overshoot/undershoot is at most its crossing term's magnitude.
Now look at one positive block (after the first crossing; finitely many initial steps don't affect convergence).
Starting from an undershoot gap Ξ΅ (with Ξ΅β€qlastβ, the last negative taken, by the lemma), the block increases monotonically, so before the crossing step sβSβ€Ξ΅, and at the crossing step Sβsβ€pcrossβ.
Hence throughout the block β£SNββsβ£β€max(qlastβ,pcrossβ).
Since the blocks recur forever (step 2), both term indices tend to β: qlastββ0 and pcrossββ0 (the crossing positives are a subsequence of the piβ).
Hence β£SNββsβ£β0, i.e. SNββs.
Negative blocks are symmetric.
This is the greedy proof of Riemann's rearrangement theorem for finite s.
(The Β±β cases lie outside this statement's scope and need a separate construction: for target +β, for instance, replace the fixed target s by stage targets mβ+β β at stage m, take positive terms until the partial sum exceeds m.)