Localization sequences are the main tool for computing continuous K-theory, so we need to understand how dualizability behaves in them. One direction is easy: dualizability passes from the middle term of a strongly continuous localization sequence to the two outer terms. The converse fails; the gluing criterion and a counterexample are developed in Section 5.4, after the compact-morphism criteria needed for their proofs.
Proposition 3.19. Let \(K \xrightarrow {i} D \xrightarrow {p} E\) be a strongly continuous localization sequence in \(\PrL _{\st }\) (Definition 2.11). If \(D\) is dualizable, then so are \(K\) and \(E\), and the sequence is a sequence in \(\Pr ^{\dual }\).
Proof.
Since \(i\) and \(p^R\) are fully faithful, we have \(i^Ri \simeq \id _K\) and \(pp^R \simeq \id _E\). By strong continuity, \(i^R\) and \(p^R\) preserve colimits, so \(K\) and \(E\) are retracts of \(D\) in \(\PrL _{\st }\), hence dualizable. The functors \(i\) and \(p\) are strongly continuous by assumption. □