In quantum mechanics, the state space is a separable complex Hilbert space.
By definition, a Hilbert space is a complete inner product space.
The term complete means that any Cauchy sequence of elements (vectors) belonging to the Hilbert space converges to an element which also belongs to the space. In other words, completeness means that the limits of convergent sequences of elements belonging to the space are also elements of the space. Intuitively, we can say that Hilbert spaces have no “holes”.
If the state space is infinite-dimensional, we implicitly invoke its completeness every time we expand a state in terms of a complete set of eigenstates, such as the energy eigenstates or the eigenstates of another observable, since an infinite series of eigenstates is meant as the limit of the sequence of the respective partial sums when the number of terms tends to infinity. The sequence of partial sums is then a Cauchy sequence converging to the initial state, which must belong to the space.
Qualitatively, considering a convergent sequence of physical states, we expect that it converges to a physical state too, because it would be unphysical, by means of such a sequence, to end up at an unphysical state. For instance, assume that we perform a series of small changes to the state of a quantum system and suddenly we reach an unphysical state. This would be physically unacceptable. Thus, from a physical perspective, the completeness of the state space seems unavoidable.
However, looking in some of the so-called standard textbooks of quantum mechanics, particularly in Sakurai’s, Merzbacher’s, Gasiorowicz’s, and Griffiths’s, this essential property is either overlooked or just mentioned, and it is not highlighted properly.