Quantum Darwinism asks how a quantum state can produce many apparently objective classical records. Its answer is not that the environment merely destroys coherence; the environment can also copy selected information about a system into many disjoint fragments, allowing different observers to learn the same pointer-state facts without directly disturbing the system.
Guide 6 owns noise and decoherence channels. Guide 8 owns entropy and correlations. Guide 21 owns tomography. This guide owns the information-proliferation problem: pointer states, environment as witness, redundancy, mutual information plateaus, spectrum-broadcast structure and the distinction between decoherence and objective classical records.
System–environment coupling → decoherence selects stable pointer observables → environment fragments acquire records → many fragments carry the same accessible information → redundant records support operational objectivity.
1. Decoherence is necessary but not the whole story
Suppose a qubit system S interacts with an environment E. Decoherence can suppress off-diagonal terms in a preferred basis of S. After tracing over E, the reduced density matrix may become nearly diagonal.
That explains why interference between selected alternatives becomes hard to observe locally. It does not yet explain how several observers can independently measure different parts of E and agree on the same classical property of S.
Quantum Darwinism adds this second question: which information about S is redundantly proliferated into the environment?
2. The simplest imprinting model
Let the system pointer basis be |0⟩,|1⟩ and prepare
|ψ_S⟩=α|0⟩+β|1⟩.
Let N environment qubits begin in |0⟩. Apply controlled-NOT gates from S into each environment qubit. The joint state becomes
|Ψ⟩=α|0⟩|0…0⟩+β|1⟩|1…1⟩.
Every environment qubit now contains a perfectly distinguishable record of the system’s Z value.
3. Reduced system state
Tracing out the environment gives
ρ_S=|α|²|0⟩⟨0|+|β|²|1⟩⟨1|
because the environment record states |0…0⟩ and |1…1⟩ are orthogonal.
The original coherence has not vanished from the global pure state; it has become encoded in nonlocal correlations between S and E.
4. Environment as witness
An observer need not touch S. Measuring one environment qubit in the Z basis reveals the same pointer value.
A second observer can measure a different environment qubit and obtain the same information. In the ideal GHZ-like model, N separate fragments carry the same classical bit.
The environment is therefore not only a sink of phase coherence; it is a communication channel carrying selected information about the system.
5. Pointer states
Pointer states are states robust under the system–environment interaction in the sense that their defining classical alternatives are preferentially recorded rather than scrambled.
If the interaction Hamiltonian has the form
H_int=A_S⊗B_E,
then eigenstates of AS are natural candidates for pointer states because they imprint distinct environment phases or states without being superposed by the interaction itself.
Other parts of the system Hamiltonian can compete with this selection, so the precise pointer basis depends on relative timescales and coupling structure.
6. Predictability sieve
Zurek’s predictability-sieve idea ranks candidate states by their robustness under environmental monitoring.
Possible criteria include minimal entropy production, maximal purity retention or stable classical trajectories.
The selected pointer states depend on the physical interaction. They are not declared by measurement postulate alone.
7. Mutual information with an environment fragment
Let F be a fragment of the environment. Quantum mutual information is
I(S:F)=S(ρ_S)+S(ρ_F)−S(ρ_{SF}).
It measures total correlations—classical plus quantum—between S and F.
Quantum Darwinism often studies I(S:F) as fragment size grows.
8. Partial-information plot
Plot I(S:F) against the fraction f of the environment contained in F.
A Darwinistic pattern often shows:
- a rapid rise for small f;
- a broad plateau near the classical information about the pointer observable;
- a final rise when almost all of E is collected and global quantum correlations become accessible.
The plateau means many different small fragments each reveal essentially the same classical information.
9. Redundancy
Let H(S) denote the classical information scale to be recovered about the pointer observable. Choose tolerance δ and define fδ as the smallest environment fraction carrying at least (1−δ)H(S) of the desired information.
Then redundancy is
R_δ=1/f_δ.
If fδ=0.02, then roughly fifty disjoint fragments of that size can each carry almost all the selected classical information.
10. Worked ideal redundancy example
Take α=β=1/√2 in the N-environment-qubit copying model.
The pointer variable is one classical bit. Any single environment qubit has state I/2 and is perfectly correlated with S in Z.
Thus one environment qubit suffices to recover the full pointer bit. If every fragment is one qubit, the ideal redundancy is N.
This model is deliberately extreme: real environments usually produce imperfect, noisy and overlapping records.
11. Imperfect records
Suppose one environment unit becomes |e₀⟩ when S=0 and |e₁⟩ when S=1, with overlap
c=⟨e₀|e₁⟩.
If |c|=1, the environment records nothing because the two conditional states are identical up to phase.
If c=0, the record is perfectly distinguishable.
Intermediate overlaps give partial records.
12. Decoherence factor and record distinguishability
For N independent identical environment units, the conditional environment states are tensor products and their overlap is
c^N.
The system off-diagonal coherence is multiplied by this same overlap factor.
Thus the environment’s ability to distinguish pointer alternatives and the system’s loss of local coherence are two sides of the same entangling process.
13. Decoherence can occur before high redundancy
The total environment may make two branches nearly orthogonal long before small fragments individually contain enough information to identify them.
Therefore strong decoherence of S does not automatically imply large Darwinistic redundancy.
This is a central conceptual distinction: suppressing interference and broadcasting accessible records are different tasks.
14. Holevo information
If the pointer observable defines classical variable Z with probabilities pz and fragment states ρF|z, the classically accessible information is bounded by the Holevo quantity
χ(Z:F)=S(Σ_z p_zρ_{F|z})−Σ_z p_z S(ρ_{F|z}).
Mutual information can include quantum correlations inaccessible through one measurement on F. Holevo information isolates how much classical information about Z can be extracted in principle.
15. Discord-like remainder
The difference between total mutual information and classically accessible information can be associated with discord-like quantum correlations for an appropriate measurement direction.
This connects naturally to Guide 82, where quantum discord is developed directly.
In Darwinistic regimes, the plateau seen by small fragments is expected to be dominated by classical pointer information, while uniquely quantum global correlations remain distributed nonlocally.
16. Spectrum broadcast structure
A particularly strong form of objectivity is represented by states of the form
ρ_{SF_1…F_R}=Σ_i p_i |i⟩⟨i|⊗ρ_i^{F_1}⊗…⊗ρ_i^{F_R},
with conditional fragment states for different i perfectly distinguishable within each fragment.
Each observer can learn i locally without disturbing the encoded classical variable, and many observers can agree.
17. Objectivity needs more than correlation
If two observers must coordinate a global measurement on the entire environment, the information is not redundantly objective in the Darwinistic sense.
Operational objectivity requires local accessibility from separate fragments, agreement among observers, and sufficiently nondisturbing readout of the selected classical information.
18. Environment fragments must be physically meaningful
The tensor-factor partition E=F₁⊗F₂⊗… should reflect physically separable or separately measurable environmental degrees of freedom.
An arbitrary mathematical basis rotation can turn locally accessible records into highly nonlocal encodings.
Redundancy is therefore defined relative to an operational decomposition of the environment.
19. Scattered photons as records
A macroscopic object’s position can imprint itself on many scattered photons. Different photons travel away and can be intercepted by different observers.
Because photons interact weakly after scattering, they can preserve records over long distances.
This makes photon environments a natural model for redundant classical information about position-like pointer observables.
20. Quantum Darwinism is not biological Darwinism
The analogy is selection and proliferation: environmentally stable pointer information is copied repeatedly, while fragile phase relations are not locally accessible.
There is no claim that quantum states reproduce biologically or undergo natural selection in the organismal sense.
21. The no-cloning theorem is not violated
Unknown arbitrary quantum states cannot be cloned perfectly.
Darwinistic environments do not copy the full state α|0⟩+β|1⟩ into many fragments. They copy selected classical information associated with approximately orthogonal pointer alternatives.
The phase coherence between alternatives remains encoded globally rather than being independently cloned into each fragment.
22. Redundancy depends on the observable
In the CNOT copying model, environment fragments redundantly encode Z.
They do not redundantly encode the X phase of α|0⟩+β|1⟩. Recovering that phase requires coherent access to large portions of the environment.
Objectivity is therefore basis selective.
23. Redundancy and quantum error correction look superficially opposite
Quantum error correction tries to hide logical quantum information from local errors by encoding it nonlocally.
Quantum Darwinism describes selected classical information becoming locally available in many fragments.
The same global state can simultaneously protect quantum phase information nonlocally while broadcasting a classical pointer value locally.
24. Finite-temperature and noisy environments
If environmental fragments begin mixed or undergo their own noise, record distinguishability decreases.
Redundancy can still emerge if many fragments each retain enough signal about the pointer variable.
A realistic analysis should compare record signal with environmental entropy and detector noise rather than assuming pure environment ancillas.
25. Non-Markovian environments can recycle records
If an environment fragment later reinteracts with S, information can flow back and alter both decoherence and redundancy.
Guide 76 owns non-Markovian memory. Darwinistic reasoning is simplest when records leave the system and become approximately independent witnesses rather than repeatedly returning.
26. Common misconceptions
“Decoherence proves objective classical reality by itself.” Decoherence suppresses local interference; objectivity additionally requires redundant accessible records.
“The environment copies the entire quantum state.” No. It copies selected pointer information, consistent with no cloning.
“Any environmental partition demonstrates redundancy.” The fragments should correspond to independently accessible physical subsystems.
“A mutual-information plateau is the only possible objectivity criterion.” Holevo information, spectrum broadcast structure and operational disturbance tests refine the claim.
27. Worked synthesis problem
Let |Ψ⟩=(|0⟩|0000⟩+|1⟩|1111⟩)/√2.
- The system state alone is I/2 in the Z basis.
- Any one environment qubit reveals the system’s Z value perfectly.
- Four separate one-qubit fragments each carry the same pointer bit.
- No one fragment contains the relative phase between the two global branches.
- Recovering that phase requires coherent access to the global correlations.
This one state simultaneously exhibits decoherence of S, redundant classical records and globally stored quantum coherence.
28. Practice set
- What is the difference between decoherence and Darwinistic redundancy?
- What is a pointer state?
- Define I(S:F).
- What does a plateau in a partial-information plot indicate?
- Define redundancy R_δ.
- What does environment-state overlap c control?
- Why can decoherence occur before high redundancy?
- What is the Holevo quantity used for here?
- What is spectrum broadcast structure?
- Why is no cloning not violated?
Answers
- Decoherence suppresses local coherence; Darwinism asks whether selected information is copied into many accessible fragments.
- A state robustly selected/recorded by the environment under the relevant interaction.
S(S)+S(F)−S(SF).- Small fragments already contain almost all selected classical information.
1/f_δ, where f_δ is the minimum environment fraction carrying the chosen information tolerance.- Both record distinguishability and decoherence strength.
- The whole environment can distinguish branches even when each small fragment contains too little information individually.
- To bound/quantify classical information about the pointer variable accessible from a fragment.
- A state with one classical pointer variable encoded in many conditionally distinguishable fragments.
- Only selected orthogonal classical information is copied locally; arbitrary quantum coherence is not.
Sources and further study
[1] Wojciech H. Zurek, Environment-assisted invariance, entanglement, and probabilities in quantum physics and related work on decoherence and pointer states.
[2] Wojciech H. Zurek, Quantum Darwinism, Nature Physics 5, 181–188 (2009). Canonical overview of redundant environmental records.
[3] Robin Blume-Kohout and Wojciech H. Zurek, Quantum Darwinism: Entanglement, branches, and the emergent classicality of redundantly stored quantum information, Physical Review A 73, 062310 (2006).
[4] R. Horodecki, J. K. Korbicz and P. Horodecki, Quantum origins of objectivity, Physical Review A 91, 032122 (2015). Spectrum broadcast structure and objectivity.
[5] J. K. Korbicz, P. Horodecki and R. Horodecki, Objectivity in a noisy photonic environment through quantum state information broadcasting, Physical Review Letters 112, 120402 (2014).
Continue through Quantum Mathematics — Batch 21
Guide 82: Quantum Discord, Classical–Quantum States, Conditional Entropy and Measurement Disturbance develops nonclassical correlations beyond entanglement. Guide 83: Weak Measurements, Weak Values, Postselection and Pointer Shifts develops gentle pre/postselected measurement statistics. Guide 84: Leggett–Garg Inequalities, Macrorealism and Temporal Quantum Correlations develops temporal nonclassicality tests.