The Sequence That Must Have a Convergent Subsequence... or Does It?
Let be a sequence of functions defined by
Does have a subsequence that converges pointwise almost everywhere on ?
Surprisingly: yes. But can you find a subsequence that converges pointwise everywhere on ?
More precisely: prove that no subsequence converges at every point of .
Answer: The Sequence That Must Have a Convergent Subsequence... or Does It?
Key Idea / Intuition
The sequence is bounded in , and by the Riemann–Lebesgue lemma its -norm doesn't vanish — it stays at . By a soft functional-analysis argument, every subsequence converges weakly to zero in , but any pointwise-everywhere convergent subsequence would have to converge to a measurable limit, and then by dominated convergence the norms would have to be while the limit is zero — giving a contradiction. A more elementary route uses the Weyl equidistribution theorem: for any fixed irrational , the values are equidistributed and can't converge. One cannot avoid such irrationalities simultaneously for all .
Formal Proof / Solution
Step 1: Weak convergence to zero.
For any , by the Riemann–Lebesgue lemma:
So weakly in , and the same holds for any subsequence.
Step 2: -norm stays bounded away from zero.
So for any subsequence.
Step 3: Pointwise a.e. convergence implies convergence via DCT.
Suppose, for contradiction, that some subsequence pointwise everywhere on .
Since , dominated convergence gives:
But weak convergence forces a.e.: for any measurable set , by Riemann–Lebesgue (since ).
If for all measurable , then a.e., contradicting .
Conclusion.
No subsequence can converge pointwise everywhere on . (It can converge almost everywhere — that follows from the fact that in and by Banach–Saks or Komlós-type arguments a subsequence of Cesàro means converges a.e., but the full sequence itself already fails everywhere off a measure-zero set by Weyl equidistribution.)
Bonus elegant version via Weyl.
Fix any with irrational (a full-measure set). The sequence is equidistributed mod 1 for any lacunary subsequence, so visits both positive and negative values infinitely often — hence diverges. Since irrationals have full measure, no subsequence can converge everywhere.