For the amplitude continue to k=iκ on the decaying physical sheet. Then ζ=q and D=cos(qa)+(κ2−q2)sin(qa)/(2κq). The double-angle identity factors its zero condition into the even and odd matching conditionsκ=qtan(qb) and κ=−qcot(qb). Thus physical upper-half-planeamplitudepoles are exactly the bound energies.
The literal statement for the probability requires correction. The displayed real-axis expression continued algebraically is 1/[D(k)D(−k)], and includes poles of the opposite-boundary-condition amplitude too. A concrete counterexampleuses ℏ2/(2m)=1, a=2, q=1, κ=cot1 and U0=1+cot21. Its continued probabilitydenominator vanishes at k=icot1, because D(−k)=0. But neither κ=tan1 nor κ=−cot1 holds, so E=−cot21 is not a bound energy. Keeping the amplitude and the decaying boundary condition removes this spurious identification.