Skip to main content

Quantum Measurement in UHM

Predicting a recorded outcome requires a preparation, an instrument with labelled outcomes and a probability rule. Here those inputs come from finite-dimensional quantum mechanics. A device that resolves two preparations defines a distinction the observer can actually use. For example, the controlled gate in the two-rebit experiment exposes a distinction invisible to all local record histories. Two-rebit experiment · Operational records

1. Statement of the Measurement Problem​

1.1 The Standard Problem​

In standard quantum mechanics, the measurement problem consists of three interrelated questions:

  1. The reduction problem: Why does the state ∣ψ⟩=∑kck∣ak⟩|\psi\rangle = \sum_k c_k |a_k\rangle "collapse" to a specific ∣ak⟩|a_k\rangle upon measurement?
  2. The probability problem (Born rule): Why is the probability of outcome kk equal to pk=∣ck∣2p_k = |c_k|^2?
  3. The preferred basis problem: What determines the basis {∣ak⟩}\{|a_k\rangle\} in which "reduction" occurs?

2. The specified measurement instrument​

A subobject S↪XS\hookrightarrow X in a topos has a characteristic arrow χS:X→Ω\chi_S:X\to\Omega. A Hilbert-space instrument is a family of completely positive operations whose sum preserves trace. Identifying an arrow with the operation Γ↦PkΓPk\Gamma\mapsto P_k\Gamma P_k requires a representation and a choice of instrument. Even a fixed observable allows different state updates.

Lüders instrument. Choose mutually orthogonal projectors with ∑kPk=I\sum_k P_k=I. Then the operation, probability and conditional state for the recorded outcome kk are

Ik(Γ)=PkΓPk,pk=Tr⁡(PkΓ),Γk=Ik(Γ)pk(pk>0).\mathcal I_k(\Gamma)=P_k\Gamma P_k,\qquad p_k=\operatorname{Tr}(P_k\Gamma),\qquad \Gamma_k=\frac{\mathcal I_k(\Gamma)}{p_k}\quad(p_k>0).

Cyclicity of the trace and Pk2=PkP_k^2=P_k give Tr Ik(Γ)=pk\mathrm{Tr}\,\mathcal I_k(\Gamma)=p_k. Replacing Ik\mathcal I_k by UkIk( ⋅ )Uk†U_k\mathcal I_k(\,\cdot\,)U_k^\dagger leaves these probabilities unchanged and can change the conditional states. Summing the operations describes discarding the outcome record.

3. The Born rule​

The trace pairing is the probability postulate used here. Positivity and completeness of the projectors give pk≥0p_k\ge0 and ∑kpk=1\sum_kp_k=1. For a pure state and Pk=∣ak⟩⟨ak∣P_k=|a_k\rangle\langle a_k| this gives ∣⟨ak∣ψ⟩∣2|\langle a_k|\psi\rangle|^2. Gleason’s theorem starts with a normalized, noncontextual, orthogonally additive probability assignment on the projectors of a complex Hilbert space of finite dimension at least three. Each such assignment has a unique representing density operator. Identifying the projectors with available tests and the representing operator with the preparation are physical premises. The topology of the state space supplies neither measurement frequencies nor the trace pairing.

4. Connection to Self-Observation (φ\varphi-Operator)​

Given Kraus operators with ∑mKm†Km=I\sum_m K_m^\dagger K_m=I, the channel φ(Γ)=∑mKmΓKm†\varphi(\Gamma)=\sum_m K_m\Gamma K_m^\dagger describes evolution averaged over the labels. Monitoring a specified Kraus realization yields an instrument. For example, discarding a fair random choice of II or ZZ gives the same qubit dephasing channel as measuring ZZ and discarding its outcome. The fair coin and the measurement have different records. Interpreting a monitored channel as self-observation requires a physically accessible record and a way to test what it reports. Self-observation

5. Decoherence​

Take a purely dissipative GKSL generator with real diagonal Lm=diag(ℓm1,…,ℓmN)L_m=\mathrm{diag}(\ell_{m1},\ldots,\ell_{mN}) and constant rates γm≥0\gamma_m\ge0. Direct substitution gives

Γij(t)=Γij(0)e−rijt,rij=12∑mγm(ℓmi−ℓmj)2.\Gamma_{ij}(t)=\Gamma_{ij}(0)e^{-r_{ij}t},\qquad r_{ij}=\tfrac12\sum_m\gamma_m(\ell_{mi}-\ell_{mj})^2.

The populations stay fixed. A coherence decays exactly when at least one positive-rate operator distinguishes its two indices. If every pair is distinguished, the limiting state is the diagonal mixture. A single outcome requires a recorded instrument. This dephasing semigroup is unital, so D(Γ(t)∥I/N)=log⁡N−S(Γ(t))D(\Gamma(t)\|I/N)=\log N-S(\Gamma(t)) cannot increase. A reset to a pure state shows why the entropy conclusion cannot be extended to all CPTP channels. Entropy and time

6. Preferred Basis Problem​

For the specified dephasing generator the fixed algebra consists of blocks indexed by the joint eigenvalue lists of the positive-rate LmL_m. Rank-one blocks select pointer rays up to phases and permutations. Conjugating every LmL_m rotates the fixed algebra. A noncommuting Hamiltonian can also move a diagonal state out of it. An experiment must calibrate which rays the actual coupling preserves.

7. Measurement and regeneration​

For a selected target σ\sigma and fixed rate, the replacement channel below gives a controlled example of feedback towards a target. Applying this description to an organism requires an independent measurement of the target, rate and response.

7.3 Self-consistency of a specified map​

Theorem 7.1: fixed replacement channel [T under fixed parameters]

Fix σ∈D7\sigma\in D_7 and 0<k≤10<k\le1 independently of the input. The linear operator map

Tk,σ(X)=(1−k)X+kTr⁡(X)σT_{k,\sigma}(X)=(1-k)X+k\operatorname{Tr}(X)\sigma

is CPTP and has the unique fixed state Γ∗=σ\Gamma^*=\sigma. On states, the fixed-point equation reduces to k(σ−Γ)=0k(\sigma-\Gamma)=0. Moreover

Tk,σn(Γ)=σ+(1−k)n(Γ−σ),∥Tk,σ(Γ)−Tk,σ(η)∥F=(1−k)∥Γ−η∥F.T_{k,\sigma}^n(\Gamma)=\sigma+(1-k)^n(\Gamma-\sigma),\qquad \|T_{k,\sigma}(\Gamma)-T_{k,\sigma}(\eta)\|_F=(1-k)\|\Gamma-\eta\|_F.

The fixed target σ\sigma and weight kk are independent of the input. If a controller instead estimates them from measurements on each preparation, its data, update and resource costs become part of the experiment.

8. No-Signaling​

Let a fixed local channel ΦA\Phi_A be CPTP and act as ΦA⊗idB\Phi_A\otimes{\rm id}_B on a jointly prepared state. Completeness of its Kraus operators gives the marginal identity

Tr⁡A[(ΦA⊗idB)(ΓAB)]=Tr⁡AΓAB.\operatorname{Tr}_A[(\Phi_A\otimes{\rm id}_B)(\Gamma_{AB})] =\operatorname{Tr}_A\Gamma_{AB}.

Multiplying the difference by a scalar depending on the current marginal leaves its partial trace zero. This tests the unconditioned marginal in the stipulated joint update. For a nonlinear map FF the preparations ∑kpkF(ρk)\sum_k p_kF(\rho_k) and F(∑kpkρk)F(\sum_k p_k\rho_k) can disagree. A physical extension must specify which records are retained, how mixed preparations compose, and what inputs the apparatus receives. The marginal identity alone does not prove no-signalling for arbitrary adaptive protocols. Joint update · Ensemble test

G4 gives a sharper resource statement for the canonical feedback ray with known target direction. For d≥3d\ge3 and fixed 0≤μ<10\le\mu<1, the exact output conditioned on success needs three or four independent preparations per attempt, depending on μ\mu. No deterministic exact implementation with a finite input bound exists on the whole ray. A computational-speedup claim would also have to count repreparations, success probability, precision and readout. Proof and assumptions

Related models and definitions