= Solution
An apparent collapse is the transition from predicting several possible recorded outcomes to describing a particular recorded result. Whether that transition is a physical discontinuity, a conditional state update or an effective description depends on what the <wave function> is taken to represent. The practical success of collapse-based calculation does not by itself settle that interpretive question.
The difficulty is exposed by <unitary time evolution>. An ideal apparatus that reliably measures a basis has
$$
|0\rangle|A_{\mathrm{ready}}\rangle\longmapsto|0\rangle|A_0\rangle,\qquad
|1\rangle|A_{\mathrm{ready}}\rangle\longmapsto|1\rangle|A_1\rangle.
$$
Linearity then makes a superposed input evolve as
$$
(a|0\rangle+b|1\rangle)|A_{\mathrm{ready}}\rangle
\longmapsto a|0\rangle|A_0\rangle+b|1\rangle|A_1\rangle.
$$
This is an entangled state, not automatically one pointer packet or the other. A separate projection rule would instead select one recorded outcome, with <Born rule> weights, and assign its corresponding conditional state.
<Quantum decoherence> explains much of the practical appearance of a classical record. Interaction with an environment correlates the packets with nearly orthogonal environment states. The subsystem's <reduced density matrix> then has very small interference terms between the pointer alternatives. Local experiments on that subsystem approximately agree with a diagonal mixture. But this is an <improper mixture>: the total state can remain a pure superposition. Taking a <partial trace> has not selected one term of the total state. Suppression of observable interference and the occurrence of a unique absolute outcome are distinct claims.
There is a <measurement problem> if one simultaneously insists that the wave function is the complete physical description, that its evolution is universally unitary, and that a measurement has exactly one absolute outcome of the ordinary recorded kind. The ideal model shows why simply applying the same unitary law to a larger apparatus does not make that combination work. Nor does calling collapse a change in knowledge solve the problem for someone who has already taken the wave function to be a complete ontic state.
Different interpretations revise different parts of this package. <Bohmian mechanics> adds an actual configuration and obtains effective subsystem collapse while the total wave evolves unitarily. The <many-worlds interpretation> retains the total superposition and treats each record as definite relative to its decohering branch. A stochastic collapse theory instead modifies the physical dynamics. Everett's original proposal explicitly separates his relative-state account from an additional collapse postulate.<Relative State Formulation of Quantum Mechanics by Hugh Everett III>[1957, introduction and measurement discussion: universal wave mechanics and correlations between apparatus and observed system]{ref}
Thus collapse is not problematic merely because conditional probabilities change when a record is learned. The problem is the status of that record and of the dynamical rule used to describe it. A defensible interpretation must explain when its updating rule is applicable, what makes a record definite in its own ontology, and why the ordinary measurement statistics follow. No appeal to human consciousness is required by the displayed unitary measurement model.
Back to article page
