Quantum Decoherence and the Density Matrix

No real system is closed. A qubit sits in a magnetic field but also scatters stray photons; a dust grain drifts in near-vacuum but is still peppered by microwave-background quanta; a measuring pointer is bombarded by air molecules. Each collision leaves a record of the system's state in the environment, and once those records are distinguishable, the system's characteristic quantum trait — the ability to interfere — is gone. Decoherence is the theory of that loss, and it needs nothing beyond ordinary unitary quantum mechanics applied to the system-plus-environment together with the partial trace. The mechanism is entanglement followed by ignorance. A system in a superposition $\sum_i c_i|s_i\rangle$ that interacts with an environment ends up in the entangled state $\sum_i c_i|s_i\rangle|E_i\rangle$ ; the global state is still pure and still evolving unitarily. But a local observer sees only the reduced density matrix $\rho_S=\mathrm{Tr}_E$ , and there the off-diagonal coherence between branches $i$ and $j$ is multiplied by the environmental overlap $\langle E_j|E_i\rangle$ . As the environment records become orthogonal, the coherences vanish and $\rho_S$ goes diagonal in the pointer basis — the system looks like a classical "either/or". Two things this article is careful to own and to withhold. It owns the mechanism (the tracing law, the master equation, the decoherence time $T_2$ , and the ruthless $\Delta x^2$ scaling that makes cats decohere unfathomably faster than they lose energy). It withholds the resolution: a diagonal reduced density matrix is an improper mixture — the whole is still pure, no single outcome has been selected. Decoherence explains why we never see interference of a live-and-dead cat; it does not say which one you find when you open the box. That residual question is the measurement problem, developed in The Quantum Measurement Problem by name.

Key takeaways

Decoherence = system–environment entanglement $\sum_i c_i|s_i\rangle|E_i\rangle$ plus the partial trace: the reduced coherence is scaled by the environment overlap, $\rho_{S,ij}(t)=\rho_{S,ij}(0)\,\langle E_j(t)|E_i(t)\rangle$ (derived).
Owned: as records orthogonalize, off-diagonals die and $\rho_S$ goes diagonal in the pointer basis — populations $|c_i|^2$ untouched. No new physics, no nonlinear collapse.
Master equation (owned): the closed-system von Neumann law gains a dissipator; pure dephasing $\dot\rho=-\tfrac{i}{\hbar}[\hat H,\rho]-\tfrac{\gamma}{2}[\sigma_z,[\sigma_z,\rho]]$ gives $\rho_{01}(t)=\rho_{01}(0)e^{-2\gamma t}$ , a decoherence time $T_2=1/2\gamma$ .
Punchline: the decoherence rate $\propto(\Delta x)^2$ ; a cat-sized separation decoheres $\sim10^{40}$ times faster than it relaxes — why macroscopic superpositions are never observed.
Honest scope: decoherence gives an improper mixture (global state still pure). It explains the appearance of collapse but does not select an outcome — that is the measurement problem (developed in The Quantum Measurement Problem by name).
Key idea: entanglement with an untracked environment, followed by tracing it out, drives the reduced density matrix toward diagonal form in the pointer basis — quantum mechanics manufactures classical-looking states without ever leaving unitary dynamics.

The core insight

A system that entangles with an environment, $\sum_i c_i|s_i\rangle|E_i\rangle$ , is described locally by $\rho_S=\mathrm{Tr}_E$ , whose off-diagonal elements carry the factor $\langle E_j|E_i\rangle$ . As environmental records become orthogonal these factors vanish and $\rho_S$ becomes diagonal in the pointer basis — the reversible microscopic dynamics produces an irreversible-looking classical mixture, without any modification of the Schrödinger equation.

Where this lives today

Decoherence is the central obstacle of quantum technology and, simultaneously, one of its sharpest diagnostics. Quantum computing lives and dies by $T_2$ : a qubit is useful only for the coherence time, and the entire enterprise of error correction, dynamical decoupling, and clean fabrication is a war against the tracing law derived above — engineers fight to keep $\langle E_j|E_i\rangle$ near $1$ by isolating the qubit from any environment that could record its state. Matter-wave interferometry runs the experiment in reverse: sending ever-larger molecules (thousands of atoms, and rising) through diffraction gratings while controlling the residual gas and thermal photons, groups measure decoherence turning on as the which-path information leaks out, mapping the quantum-to-classical boundary quantitatively. And cavity QED watches a single "cat state" of light — a superposition of two phases of a field mode — decohere in real time, confirming that the decoherence rate scales with the square of the separation between the components exactly as the $\Delta x^2$ law predicts. In every case the object measured is the off-diagonal element $\rho_{01}$ , and its decay is the environment doing arithmetic no one asked it to do.

Figure

The Δx² punchline: coherence decays first, energy relaxes last — the cat is off the chart

The physical picture

Put a particle in a superposition of "here" $|L\rangle$ and "there" $|R\rangle$ . On its own the amplitudes interfere. But each location scatters the surrounding air and light slightly differently, so the environment is dragged into a correlated state: $|L\rangle$ pushes it to $|E_L\rangle$ , $|R\rangle$ to $|E_R\rangle$ . The full state is $|L\rangle|E_L\rangle+|R\rangle|E_R\rangle$ — entangled, and still perfectly pure. To predict anything about the particle alone you trace the environment out, and the interference cross-term is weighted by $\langle E_L|E_R\rangle$ . After a single collision that overlap is already minuscule, the cross-term dies, and the particle behaves as a classical "here or there". The environment has, without anyone looking, measured which path was taken; and a which-path record is exactly what forbids interference. Why so fast and so irreversible? The environment has astronomically many degrees of freedom, and each carries away a sliver of which-path information. To recohere — to bring the interference back — you would have to reverse every one of those environmental correlations at once, which is as hopeless as unscrambling an egg. The quantum information is not destroyed (the global evolution is unitary and reversible in principle) but it is dispersed beyond any local reach. Decoherence is monitoring by the universe, and the universe never forgets. The environment does not decohere every basis equally. The pointer basis — the states that survive — is whatever the interaction can imprint on the environment without kicking the system around. For a position-coupling scattering environment those are localized wave packets, which is why chairs and dust grains have definite positions rather than living in position superpositions. The selection rule that picks this basis is developed in The Pointer Basis and Environment-Induced Superselection by name; here we own the mechanism that makes any such basis go diagonal.

Figure

Entanglement writes a which-path record; the trace turns it into decoherence

The tracing law

The core object is the reduced density matrix, obtained by tracing the environment out of the global state. The density-operator credentials ( $\rho^\dagger=\rho$ , $\mathrm{Tr}\,\rho=1$ , $\rho\succeq0$ ) and the partial trace itself are owned by The Density Operator and Mixed States and The Reduced Density Matrix and Partial Trace, recalled here by name; we put them to work.

Derivation ADecoherence as tracing — the coherence carries the environmental overlap

We start from a system superposition, let the interaction entangle it with the environment, and trace the environment out. The goal is to show that every off-diagonal element of the reduced density matrix is exactly its initial value times the overlap of the corresponding environmental records — so decoherence is nothing but the orthogonalization of those records, read through the partial trace. The populations, by contrast, will come out untouched, which pins the whole effect on phase information, not energy.

Step 1. Premeasurement. Let the interaction copy the system's pointer index into the environment without disturbing the pointer states, $\hat U|s_i\rangle|E_0\rangle=|s_i\rangle|E_i\rangle$ . A general initial system state $\sum_i c_i|s_i\rangle$ with the environment in its ready state $|E_0\rangle$ evolves to

|\Psi\rangle=\Big(\sum_i c_i|s_i\rangle\Big)|E_0\rangle\;\xrightarrow{\hat U}\;\sum_i c_i\,|s_i\rangle|E_i\rangle \qquad (1)

Why this is generic: any interaction of the form $\hat H_{\rm int}=\hat A_S\otimes\hat B_E$ with $\hat A_S|s_i\rangle=a_i|s_i\rangle$ leaves the $|s_i\rangle$ intact and merely evolves the environment conditionally on $i$ — the definition of a which-path measurement by the environment.

Step 2. Form the global density operator and trace. With $\rho=|\Psi\rangle\langle\Psi|$ ,

\rho_S=\mathrm{Tr}_E\,\rho=\sum_{ij} c_i c_j^*\,\langle E_j|E_i\rangle\;|s_i\rangle\langle s_j| \qquad (2)

Why the overlap appears: the partial trace sums the environmental "diagonal" $\sum_k\langle E^{(k)}|\,\cdot\,|E^{(k)}\rangle$ , and for the cross term $|s_i\rangle\langle s_j|$ this collapses to the single number $\langle E_j|E_i\rangle$ — the inner product of the two records. The trace's basis-independence is owned by The Reduced Density Matrix and Partial Trace.

Step 3. Read off populations and coherences. The diagonal entries are $\rho_{S,ii}=|c_i|^2\,\langle E_i|E_i\rangle=|c_i|^2$ (records are normalized), independent of any overlap; the off-diagonals are

\rho_{S,ij}=c_i c_j^*\,\langle E_j|E_i\rangle,\qquad i\ne j \qquad (3)

Why this is the whole story: populations are frozen because a which-path record cannot change the probability of a path, only erase the interference between paths. As $\langle E_j|E_i\rangle\to0$ the coherences vanish and $\rho_S\to\sum_i|c_i|^2|s_i\rangle\langle s_i|$ — diagonal, a classical mixture. *Verified numerically: for $c=(0.6,0.8)$ and records with overlap $o=\langle E_1|E_0\rangle\in\{1,0.5,0.1,0\}$ , the traced $\rho_{S,01}$ equals $c_0c_1^*\,\langle E_1|E_0\rangle$ and $\rho_{S,00}=|c_0|^2=0.36$ for every $o$ ; at $o=0$ the reduced state is exactly diagonal.*

The master equation and the decoherence time

Continuous monitoring turns the discrete step above into a rate law. The closed-system dynamics is the von Neumann equation $i\hbar\dot\rho=[\hat H,\rho]$ , owned by Properties and Time Evolution of the Density Matrix; an open system adds a dissipator. That article states the general Lindblad form and hands its derivation to this one; we cash that bank at the pure-dephasing level, where the algebra is transparent and the decoherence time falls out.

Go deeper · Derivation B · 1 minThe dephasing master equation and the decoherence time T2

Derivation BThe dephasing master equation and the decoherence time T2

We take the Lindblad master equation with a single Hermitian jump operator proportional to the pointer observable $\sigma_z$ — the "pure dephasing" channel that decoheres without exchanging energy — and read its action on a qubit's off-diagonal element. The goal is the decay law $\rho_{01}(t)=\rho_{01}(0)e^{-2\gamma t}$ and the identification of the decoherence time $T_2$ , together with the fact that populations do not move at all under pure dephasing.

Step 1. Write the dephasing generator. With jump operator $L=\sqrt{\gamma}\,\sigma_z$ , the Lindblad dissipator $L\rho L^\dagger-\tfrac12\{L^\dagger L,\rho\}=\gamma(\sigma_z\rho\sigma_z-\rho)$ (using $\sigma_z^2=\mathbb 1$ ) gives, together with a Hamiltonian $\hat H=\tfrac12\hbar\omega\sigma_z$ ,

\dot\rho=-\tfrac{i}{\hbar}[\hat H,\rho]-\tfrac{\gamma}{2}\,[\sigma_z,[\sigma_z,\rho]] \qquad (4)

Why the double commutator: expanding $\tfrac12[\sigma_z,[\sigma_z,\rho]]=\rho-\sigma_z\rho\sigma_z$ reproduces the dissipator; the double-commutator form makes manifest that anything commuting with $\sigma_z$ (the populations) is annihilated.

Step 2. Act on the entries. Since $\sigma_z\rho\sigma_z$ flips the sign of the off-diagonal elements and leaves the diagonal alone, $(\sigma_z\rho\sigma_z-\rho)_{01}=-2\rho_{01}$ while $(\cdot)_{00}=0$ . The dissipative part of (4) therefore reads $\dot\rho_{01}\big|_{\rm diss}=-2\gamma\rho_{01}$ and $\dot\rho_{00}\big|_{\rm diss}=0$ . Combining with the Hamiltonian rotation $\rho_{01}\propto e^{-i\omega t}$ ,

\rho_{01}(t)=\rho_{01}(0)\,e^{-i\omega t}\,e^{-2\gamma t},\qquad \rho_{00}(t)=\rho_{00}(0) \qquad (5)

where: the phase factor $e^{-i\omega t}$ is the reversible Larmor rotation (owned by the evolution article) and $e^{-2\gamma t}$ is the irreversible environmental damping. Populations are frozen; only the coherence magnitude decays.

Step 3. Name the timescale. Writing $|\rho_{01}(t)|=|\rho_{01}(0)|\,e^{-t/T_2}$ identifies the decoherence (dephasing) time $T_2=1/2\gamma$ . Why $T_2$ and not $T_1$ : pure dephasing kills coherence ( $T_2$ ) without any population relaxation ( $T_1\to\infty$ here) — decoherence and dissipation are different clocks, and for macroscopic superpositions the decoherence clock runs vastly faster. *Verified symbolically: for $L=\sqrt{\gamma}\sigma_z$ the dissipator's $01$ entry is exactly $-2\gamma\rho_{01}$ and its $00$ entry is $0$ (sympy). Numerically integrating (4) with $\gamma=0.7$ from $|+\rangle\langle+|$ gives $\rho_{01}(1)=0.1233=\tfrac12 e^{-2\cdot0.7}$ to $10^{-3}$ .*

Limiting cases

case	reduced state	reading
environment records identical, $\langle E_j\\|E_i\rangle=1$	$\rho_S$ = the original coherent state	no monitoring, full interference survives
records orthogonal, $\langle E_j\\|E_i\rangle=0$	$\rho_S=\sum_i\\|c_i\\|^2\\|s_i\rangle\langle s_i\\|$	complete decoherence, diagonal mixture
single CNOT, $\\|+\rangle$ control	$\rho_S=\mathrm{diag}(\tfrac12,\tfrac12)$	one interaction can fully dephase
pure dephasing, rate $\gamma$	$\\|\rho_{01}\\|=\\|\rho_{01}(0)\\|e^{-t/T_2}$ , $T_2=1/2\gamma$	populations frozen, coherence decays
macroscopic separation $\Delta x$	rate $\propto(\Delta x)^2$ , $T_2$ astronomically short	superpositions never observed
whole system + environment	global state stays PURE	decoherence is local; no outcome selected

Common trapA diagonal reduced density matrix is not a definite outcome

The most seductive over-claim in this subject is that decoherence "solves the measurement problem" by turning a superposition into a mixture. It does not. What decoherence produces is an improper mixture: the global state $\sum_i c_i|s_i\rangle|E_i\rangle$ is still pure and coherent, and the diagonal $\rho_S=\sum_i|c_i|^2|s_i\rangle\langle s_i|$ only looks like the classical ensemble "the system is in $|s_i\rangle$ with probability $|c_i|^2$ ". A genuine (proper) mixture would mean a definite-but-unknown $|s_i\rangle$ was actually realized; decoherence never establishes that — the coherence is hidden in system–environment correlations, in principle recoverable. So decoherence explains why you never see interference of macroscopically distinct alternatives, and it picks out which alternatives are stable (the pointer basis), but it does not say which single one you find. The step from improper mixture to a single fact is the measurement problem, developed in The Quantum Measurement Problem by name. Trap to bank: "diagonal $\rho_S$ " $\ne$ "a definite outcome occurred."

Worked examples

Example 1 — one interaction can fully decohere a qubit (confidence builder). A system qubit is prepared in $|+\rangle=(|0\rangle+|1\rangle)/\sqrt2$ , an environment qubit in $|0\rangle$ . They interact through a controlled-NOT (the environment flips iff the system is $|1\rangle$ — a perfect which-path record). Find the system's reduced state.

Step 1. Entangle. $\mathrm{CNOT}\,|+\rangle|0\rangle=\tfrac{1}{\sqrt2}(|0\rangle|0\rangle+|1\rangle|1\rangle)$ — the Bell state $|\Phi^+\rangle$ . The which-path record is now perfect: environment $|0\rangle$ means system $|0\rangle$ , environment $|1\rangle$ means system $|1\rangle$ , and the two records are orthogonal, $\langle0|1\rangle=0$ .

Step 2. Trace out the environment. Using equation (3) with orthogonal records, every off-diagonal element dies: $\rho_S=\mathrm{Tr}_E|\Phi^+\rangle\langle\Phi^+|=\boxed{\mathrm{diag}(\tfrac12,\tfrac12)}$ (machine-verified). The initially pure, fully coherent $|+\rangle$ has become the maximally mixed state — purity $\mathrm{Tr}\,\rho_S^2=\tfrac12$ (machine-verified).

Step 3. Note what did not happen. The global two-qubit state is still the pure $|\Phi^+\rangle$ (global purity $=1$ , machine-verified). No outcome was selected; the coherence moved into the system–environment correlation, where a joint measurement could still retrieve it. A single, perfectly-recording interaction is enough to decohere the system completely — this is why macroscopic pointers, which record continuously, are classical to spectacular precision.

Example 2 — how many environment particles it takes (exam level). A system couples to $n$ identical environment qubits, each rotated by $+\theta/2$ if the system is $|0\rangle$ and $-\theta/2$ if it is $|1\rangle$ , with $\theta=0.3$ rad. The environmental records are $|E_0\rangle=|q_+\rangle^{\otimes n}$ and $|E_1\rangle=|q_-\rangle^{\otimes n}$ . Find how the coherence decays and when it drops below $10^{-6}$ .

Step 1. Overlap per particle. A single environment qubit gives $\langle q_-|q_+\rangle=\cos\theta$ (each records the system's state imperfectly, since $\theta\ne\pi$ ). For $n$ independent qubits the total overlap multiplies: $\langle E_1|E_0\rangle=(\cos\theta)^n$ , so $|\langle E_1|E_0\rangle|=|\cos\theta|^n$ (machine-verified for $n=1,5,20$ ).

Step 2. Coherence decay. By equation (3), $|\rho_{S,01}(n)|=|c_0c_1^*|\,|\cos\theta|^n$ — an exponential decay in the number of environmental scatterings, $|\cos\theta|^n=e^{n\ln\cos\theta}$ . With $\theta=0.3$ , $\cos\theta=0.9553$ , so each collision multiplies the coherence by $0.9553$ .

Step 3. Count to $10^{-6}$ . Solve $(\cos0.3)^n<10^{-6}$ : $n>\ln(10^{-6})/\ln(0.9553)=302.6$ , so $\boxed{n=303}$ collisions suffice (machine-verified: $\cos(0.3)^{303}=8.9\times10^{-7}$ , $\cos(0.3)^{302}=9.3\times10^{-7}$ ). A modest $\theta=0.3$ needs a few hundred scatterings; a real dust grain in air suffers $\sim10^{20}$ collisions per second, so its coherence is annihilated in far less than a nanosecond — the microscopic origin of the $\Delta x^2$ scaling that makes the classical world classical.

From Zeh's insight to a quantitative science

Origins · 1 minquantum decoherence and how this was discovered

quantum decoherence

1970–1991 · Heidelberg, Los Alamos

Decoherence was named and framed by a small group who insisted that the environment is not a nuisance to be idealized away but the source of classical appearances. The founding paper is H. Dieter Zeh's 1970 argument (Foundations of Physics 1, 69) that macroscopic systems can never be treated as isolated, so their apparent classicality must come from unavoidable coupling to the outside — a heretical claim at the time, when the measurement problem was thought purely interpretive. Through the 1980s Wojciech Zurek sharpened the mechanism into the pointer-basis/einselection picture (Physical Review D 24, 1516, 1981; 26, 1862, 1982) and, in a widely read 1991 Physics Today article, brought "decoherence" to a broad audience. The quantitative theory of how fast — the scattering-induced decoherence rate and its ferocious $\Delta x^2$ scaling — is due to Erich Joos and Zeh (Zeitschrift für Physik B 59, 223, 1985), whose table of decoherence times for dust grains and larger objects made the abstract idea unignorable. Zurek's person and the einselection story are developed in The Pointer Basis and Environment-Induced Superselection by name; here we credit the founders and own the mechanism.

History

1970

— Zeh: macroscopic systems are never isolated; classicality is environment-induced (Found. Phys. 1, 69)

1981

— Zurek: the pointer basis and the commutivity criterion (Phys. Rev. D 24, 1516)

1985

— Joos and Zeh: scattering decoherence and the Δx² rate; the decoherence-time table (Z. Phys. B 59, 223)

1991

— Zurek's Physics Today article popularizes "decoherence" for a wide audience

2003

— Zurek's review consolidates decoherence, einselection, and quantum Darwinism (Rev. Mod. Phys. 75, 715)

Going further

References. Zurek, Rev. Mod. Phys. 75, 715 (2003) · Zeh, Found. Phys. 1, 69 (1970) · Joos & Zeh, Z. Phys. B 59, 223 (1985) · Schlosshauer, Decoherence and the Quantum-to-Classical Transition (Springer, 2007) · Breuer & Petruccione, The Theory of Open Quantum Systems, Ch. 3–4 · Nielsen & Chuang, §8 (quantum operations) · Joos, Zeh, Kiefer, Giulini, Kupsch & Stamatescu, Decoherence and the Appearance of a Classical World in Quantum Theory, 2nd ed. (2003).

Provenance. The base article's physics is standard and was preserved. The density-operator credentials and the partial trace are owned by The Density Operator and Mixed States and The Reduced Density Matrix and Partial Trace; the closed-system von Neumann equation and the general Lindblad form are owned by Properties and Time Evolution of the Density Matrix (which states the Lindblad equation and hands its derivation here), all recalled by name. Owned and derived here: (1) the tracing law $\rho_{S,ij}=c_ic_j^*\langle E_j|E_i\rangle$ from the entangled state (equations 1–3), machine-verified for records of overlap $\{1,0.5,0.1,0\}$ ; (2) the pure-dephasing master equation $\dot\rho=-\tfrac{i}{\hbar}[\hat H,\rho]-\tfrac{\gamma}{2}[\sigma_z,[\sigma_z,\rho]]$ giving $\rho_{01}(t)=\rho_{01}(0)e^{-i\omega t}e^{-2\gamma t}$ and $T_2=1/2\gamma$ , verified symbolically and by integration ( $\rho_{01}(1)=0.1233$ at $\gamma=0.7$ ); (3) the improper-mixture / honest-scope statement. Worked numbers recomputed: $\mathrm{CNOT}\,|+\rangle|0\rangle\to\rho_S=\mathrm{diag}(\tfrac12,\tfrac12)$ , purity $\tfrac12$ , global purity $1$ (Example 1); $|\rho_{S,01}|=|\cos\theta|^n$ with $n=303$ collisions to reach $10^{-6}$ at $\theta=0.3$ (Example 2). The pointer-basis selection is owned by The Pointer Basis and Environment-Induced Superselection; the emergence of the classical world by Emergence of Classicality from Quantum Mechanics; the measurement problem by The Quantum Measurement Problem; POVM/Kraus channels by Generalized Measurements and POVMs; the Zeno suppression of decoherence by The Quantum Zeno and Anti-Zeno Effects — all by name.

Builds on: The Density Operator and Mixed States · The Reduced Density Matrix and Partial Trace · Quantum Entanglement and Non-Separability · Pure versus Mixed States and the Purity Criterion · Properties and Time Evolution of the Density Matrix

Leads to: The Pointer Basis and Environment-Induced Superselection · Emergence of Classicality from Quantum Mechanics · Quantum Darwinism · The Quantum Measurement Problem

If the global state stays pure, in what sense has anything "decohered"?

Decoherence is a statement about the local description available to someone who can access only the system. The global system+environment state $\sum_i c_i|s_i\rangle|E_i\rangle$ is pure and coherent, but the object that predicts every local measurement is $\rho_S=\mathrm{Tr}_E$ , and there the interference terms are gone. So "decohered" means: no measurement on the system alone can reveal the coherence — it has been relocated into system–environment correlations that a local observer cannot undo. Nothing is destroyed globally; the accessibility of the coherence is what changed.

What is the difference between the decoherence time T₂ and the relaxation time T₁?

$T_1$ is the timescale for energy exchange with the environment (populations relaxing toward thermal equilibrium); $T_2$ is the timescale for phase coherence loss (off-diagonal decay). Pure dephasing, the channel derived above, has finite $T_2$ but infinite $T_1$ — coherence dies while populations sit still. In general $T_2\le2T_1$ , and for macroscopic superpositions $T_2$ can be dozens of orders of magnitude shorter than $T_1$ . That separation — coherence annihilated essentially instantly while energy barely changes — is exactly why we never catch a macroscopic object mid-superposition.

Does decoherence require a measurement or a conscious observer?

No. The "measurement" is done by the environment — photons, air molecules, phonons — entirely automatically, whether or not any physicist is watching. All that is needed is an interaction that correlates distinct system states with distinguishable environmental states. Consciousness plays no role in the mechanism derived here; the partial trace is just the statement that the observer happens not to track the environment. This is precisely why decoherence is often said to demystify measurement without solving it.

Can decoherence ever be reversed?

In principle yes, in practice essentially never. Because the global evolution is unitary, the information in the coherences is not destroyed, only spread into environmental correlations; reversing it would require re-focusing every one of the astronomically many degrees of freedom simultaneously — a spontaneous "recoherence" of overwhelming improbability. In tiny, well-controlled environments (a handful of qubits, a cavity mode) recoherence can be engineered, and spin-echo sequences deliberately reverse dephasing. It is the scale of a real environment, not any law, that makes macroscopic decoherence irreversible.

Why does the pointer basis end up being position for everyday objects?

Because the everyday environment couples to position: photons and air molecules scatter off an object at a rate that depends on where it is, so the interaction Hamiltonian is (approximately) a function of position. The observable the interaction couples to is the one whose eigenstates are recorded without being disturbed, and those become the stable pointer states. For scattering environments that observable is (coarse-grained) position, which is why chairs have definite locations rather than definite momenta or energies. The selection rule that formalizes this is developed in The Pointer Basis and Environment-Induced Superselection by name.

Is the Lindblad equation exact, or an approximation?

The Lindblad (Gorini–Kossakowski–Sudarshan–Lindblad) form is the most general generator of a Markovian, completely positive, trace-preserving evolution — exact within those assumptions. Real environments have memory (non-Markovian effects) and the derivation from a microscopic system–bath model requires the Born–Markov and secular approximations, valid when the environment relaxes much faster than the system. When those hold — as they typically do for a macroscopic bath — the Lindblad equation is quantitatively accurate. The general Lindblad structure is owned by Properties and Time Evolution of the Density Matrix; here we used its pure-dephasing special case, where it is transparent.

Does frequent measurement speed up or slow down decoherence?

Both are possible, and it is a genuine effect: continuous or rapid observation can freeze a system's evolution (the quantum Zeno effect) or, in other regimes, accelerate transitions (anti-Zeno). In the decoherence language, coupling more strongly or more often to a monitoring environment generally speeds the loss of coherence, but repeatedly projecting onto a pointer state can inhibit the leakage that would otherwise decohere a different observable. The competition between measurement-induced freezing and environment-induced decoherence is developed in The Quantum Zeno and Anti-Zeno Effects by name.