We address two pressing questions in the theory of the Korteweg–de Vries
(KdV) equation. First, we show the uniqueness of solutions to KdV that are
merely bounded, without any further decay, regularity, periodicity, or almost
periodicity assumptions. The second question, emphasized by Deift, regards
whether almost periodic initial data leads to almost periodic solutions to KdV.
Building on the new observation that this is false for the Airy equation, we
construct an example of almost periodic initial data whose KdV evolution
remains bounded, but fails to be almost periodic at a later time. Our
uniqueness result ensures that the solution constructed is the unique
development of this initial data.
