REVIEW 3 major objections 3 minor 7 references
Sudden death of entanglement, rebirth of magic
T0 review · 3 major / 3 minor · reviewed 2026-08-02 · deepseek-v4-flash
Pith's one-line read Under local amplitude damping, an n-qubit superposition α|0ⁿ⟩+β|1ⁿ⟩ loses magic at a finite noise strength, regains it at a later one, and the magic-rebirth threshold exactly mirrors the irreversible death of entanglement.
desk verdict The in-body GHZ magic death-and-rebirth result is exact and worth engaging; the abstract oversells it with unsupported thermal/unital claims. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central object is the stabilizer polytope S: the convex hull of all pure stabilizer states. The load-bearing identity is the reduction for states with one off-diagonal coherence: membership in S is equivalent to |c|≤min(p₀,p₁), a consequence of the affine-subspace, equal-modulus support structure of pure stabilizer states. Applied to amplitude damping, this turns the whole trajectory into two scalar inequalities, P₀≥c and Pₙ≥c, yielding the closed-form thresholds. The mirror symmetry comes from the complementary-channel relation, which sends damping strength γ to 1−γ, exchanging the entanglement-death condition with the magic-rebirth condition. Parity-syndrome projection onto span{|0ⁿ⟩,|
What would settle it
Construct or find one pure stabilizer state whose two computational-basis endpoint amplitudes have unequal modulus; such a state would violate the equal-modulus support property and could invalidate the membership criterion. Directly, prepare the two-qubit state α|00⟩+β|11⟩ with α=0.4, apply amplitude damping, and perform state tomography over the predicted stabilizer window [0.324,0.564]; if ρ(γ) inside that interval cannot be written as a convex combination of two-qubit stabilizer states, the central claim is false.
Extended reading notes
Core claim
On its own terms, the paper establishes an exact criterion for stabilizer-polytope membership of the damped trajectory: ρ_n(γ) is a stabilizer mixture exactly when both endpoint populations P₀ and Pₙ dominate the coherence c. From that criterion, the trajectory enters the stabilizer set at the unique root γ₋ of P₀=c and exits at γ₊=1−(α/β)^{2/n}, while bipartite entanglement dies at γ_e=(α/β)^{2/n}; the rebirth and entanglement-death thresholds are therefore reflected about γ=1/2. The paper further shows that the reborn magic is nonlocal in the sense that every proper marginal is stabilizer, and that a parity measurement followed by stabilizer decoding extracts it into a single-qubit magic s
Load-bearing premise
The central claim rests on the structural characterization of pure stabilizer states as having computational-basis support that is an affine subspace with equal-modulus amplitudes; if that characterization fails, the reduction of stabilizer membership to |c|≤min(p₀,p₁) and the identities γ_e+γ₊=1 do not follow. The proved results also assume zero-temperature, ground-state-preserving amplitude damping; the abstract's thermal 'magic island' and 'second sudden death' language re
Editorial extensions
If this is right
- For 0<α<β, the stabilizer window [γ₋,γ₊] is exact: inside it the state is a stabilizer mixture and classically simulable; outside it, the robustness of magic controls the classical-simulation overhead, so the overhead is non-monotone in damping strength.
- Because γ₊=1−γ_e, measuring the entanglement sudden-death time immediately predicts the magic-rebirth time, and conversely, for every n.
- The reflection survives inhomogeneous damping in geometric-mean coordinates: the death surface ∏γᵢ=r² and the rebirth surface ∏(1−γᵢ)=r² are exchanged by γᵢ→1−γᵢ; fixed-strength dephased amplitude damping preserves the reflection while shrinking the re-entrant domain.
- No pure stabilizer input can show finite death-and-rebirth under homogeneous damping: each stabilizer state either stays stabilizer forever (constant-Hamming-weight support) or leaves immediately and returns only at the endpoint γ=1.
- The extraction protocol is lossless and economical: the decoded single qubit is non-stabilizer exactly when the n-qubit state is, and the successful branch has per-register yield at least α²/2, with large-n behavior entering both standard single-qubit distillation windows.
Reading between the lines
- Editorial extension: if the exact thresholds hold, classical simulation of amplitude-damped circuits should become harder again after γ₊ than inside the stabilizer window; a concrete test is to compare simulation cost before, inside, and after the window on the same trajectory.
- Editorial extension: the paper links complementarity to reflection-symmetric coherence profiles, and shows a complex phase twist breaks the threshold reflection while leaving entanglement death unchanged; this suggests the mirror identity is a property of a class of non-unital channels, not of all Markovian noise.
- Editorial extension: the cat-state injection primitive at α=β approaches a standard magic state super-exponentially in n with bounded success probability; active recovery or feed-forward might trade that constant probability for higher yield, an avenue the paper leaves open.
- Editorial extension: the fully separable but magic-carrying branch implies separability is not a witness against quantum computational advantage; similar phenomena may exist in other resource theories whose free sets are not closed under the noise map.
Formalized claims in Lean
-
Claim #1: On its own terms, the paper establishes an exact criterion for stabilizer-polytope membership of the damped trajectory: ρ_n(γ) is a stabilizer mixture exactly when both endpoint populations P₀ and Pₙ dominate the coherence c. From that criterion, the trajectory enters the stabilizer set at the unique root γ₋ of P₀=c and exits at γ₊=1−(α/β)^{2/n}, while bipartite entanglement dies at γ_e=(α/β)^{2/n
/-- @claim 1 On its own terms, the paper establishes an exact criterion for stabilizer-polytope membership of the damped trajectory: ρ_n(γ) is a stabilizer mixture exactly when both endpoint populations P₀ and Pₙ dominate the coherence c. From that criterion, the trajectory enters the stabilizer set at the unique root γ₋ of P₀=c and exits at γ₊=1−(α/β)^{2/n}, while bipartite entanglement dies at γ_e=(α/β)^{2/n -/ def central_claim : Prop :=
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies how local amplitude damping acts on the n-qubit GHZ-type family |ψ_n⟩=α|0^n⟩+β|1^n⟩. It proves an exact stabilizer-polytope membership criterion for real GHZ-X states (SM Theorem S1), from which it derives magic death at γ_-, magic rebirth at γ_+, and entanglement sudden death at γ_e, with the complementarity γ_e+γ_+=1 for all n. It further shows that in a parameter regime the reborn magic lives in fully separable states with stabilizer marginals, extracts it losslessly onto a single qubit by parity-syndrome measurement (Theorem 2), and classifies pure stabilizer states under homogeneous amplitude damping into magic-generators and magic-insulators. The abstract additionally advertises a complete mapping of real phase-covariant Markovian semigroups, thermal 'magic islands', a condition T_2>T_1 for zero-temperature rebirth, and an optimal per-register yield α²/2.
Significance. If the central results are correct, this is a valuable and surprising contribution: an exactly solvable instance of mixed-state stabilizer-polytope membership under open-system dynamics, with magic death/rebirth thresholds computed in closed form and a sharp magic-entanglement complementarity. The mathematical core is strong: SM Theorem S1 gives an explicit convex decomposition (S26) and matching dual witnesses (S27); the thresholds γ_- , γ_+ , γ_e are derived, not fitted; and the extraction identity (11) is proved exactly. These constructive, parameter-free derivations are the paper's principal strength. The Bell-state splitting and the support-based generator/insulator classification are also clean and novel. However, the paper's advertised broader scope — full mapping of real phase-covariant Markovian semigroups, thermal excitation, unital dynamics, and an optimal yield — is not supported by the body or the SM, and this overreach is load-bearing for the abstract's claims.
major comments (3)
- [Abstract; End Matter A, Eq. (A1); App. S8] The abstract states that 'within real phase-covariant Markovian semigroups the phenomenon is mapped out in full', including zero-temperature rebirth iff T_2>T_1, no rebirth for unital dynamics, and a thermal 'magic island' ending in a second sudden death. The most general channel in End Matter A, Eq. (A1), is ground-state-preserving: E(|0⟩⟨0|)=|0⟩⟨0|, with no |0⟩→|1⟩ term, so it contains no thermal excitation. Appendix S8 treats only fixed-strength dephased amplitude damping with a constant profile η, not a thermal map. The sentence about T_2>T_1 appears only as an unproved consequence for λ(γ)=(1−γ)^a, a≥1/2. These abstract claims are precisely what a reader would cite for the broader phenomenon; they are unsupported as written. The authors should either provide the thermal/unital analysis or restrict the abstract.
- [Abstract; Theorem 2, Eq. (11); App. S4] The abstract's 'optimal per-register yield α²/2' is nowhere derived. Theorem 2 proves the lossless expected-robustness identity (P0+Pn)(R(ρ̃)−1)=R(ρ_n)−1 and the postselection probability P0+Pn ≥ α², and App. S4 gives a cat-state injection approaching |H⟩ with success probability tending to 2−√2, but no argument identifies α²/2 as an optimal yield. If the yield statement is a separate result, it needs a proof and a precise definition; otherwise it should be removed.
- [End Matter A, Proposition 1] Proposition 1 does not 'map out in full' the real phase-covariant Markovian semigroups. It assumes the existence of unique thresholds and then characterizes the reflected identity γ_e+γ_+=1 via S(γ_e)=S(1−γ_e). No classification of unital or thermal phase-covariant semigroups is given, and no theorem states 'zero-temperature rebirth occurs iff T_2>T_1' for a general Markovian semigroup. The abstract's universality claim goes beyond what End Matter A proves; the manuscript should either supply the missing analysis or narrow the stated scope to the ground-state-preserving class actually treated.
minor comments (3)
- [Eq. (7)] The notation ⟨ZI⟩ and ⟨XX⟩ in Eq. (7) is not defined before use. It presumably denotes expectations of Z⊗I and X⊗X on the two-qubit X-slice; please define explicitly.
- [Fig. 2(b)] The label 'C(1−γ)' in the pointwise mirror identity (8) refers to the concurrence of the Stinespring-mirrored state at damping 1−γ, not to a function evaluated at 1−γ in the original trajectory. A sentence clarifying this would prevent confusion.
- [SM Theorem S1 proof] The proof relies on the affine-support/equal-modulus structure of pure stabilizer states, cited to Ref. [29]. Since this structure is load-bearing for the entire membership criterion, a self-contained statement of the needed lemma (even without full proof) inside the SM would improve readability.
Circularity Check
No significant circularity: thresholds are computed from an external stabilizer-support theorem, not from fitted inputs or self-citations.
full rationale
The central derivation is self-contained conditional on the standard external theorem of Dehaene–De Moor [29] that pure stabilizer states have affine computational-basis support and equal-modulus amplitudes. Theorem S1 derives the real GHZ-X criterion |c| <= min(p0,p1) by exhibiting an explicit stabilizer decomposition and matching dual witnesses, rather than assuming that criterion. Corollary S1 then obtains gamma_- and gamma_+ by solving P0=c and Pn=c, and gamma_e = r^{2/n} is obtained from a separate partial-transpose 2x2 determinant in Proposition 2/End Matter. The complementarity gamma_e + gamma_+ = 1 follows from these independent formulas, not from a parameter fit or from circular re-use of the claimed conclusion. The lossless extraction identity (Eq. 11) is derived algebraically from the robustness formula and the decode Bloch coordinates. No fitted parameter is renamed as a prediction, and no load-bearing self-citation appears. I additionally flag, as a correctness gap rather than a circularity, that the abstract's claims about thermal excitation, T1/T2 conditions, unital dynamics, and optimal per-register yield alpha^2/2 are not derived in the body or SM: End Matter A treats only ground-state-preserving phase-covariant channels, and no thermal/unital/yield derivation appears. This omission does not make the stated GHZ-amplitude-damping derivation circular.
Assumptions & free parameters
assumptions (8)
- standard math Dehaene–De Moor theorem: every pure n-qubit stabilizer state has computational-basis support that is an affine subspace, equal-modulus amplitudes, and Clifford (affine-quadratic) phases.
- standard math Endpoint specialization: for a pure stabilizer state, endpoint amplitudes on {0^n, 1^n} are both zero, one zero, or equal-modulus; the only qualifying real GHZ-X states are (|0^n⟩ ± |1^n⟩)/√2.
- standard math Entanglement diagnostics: vanishing of the 2×2 partial-transpose block determinant detects bipartite negativity death; the X-state concurrence uses the pairing |ρ14| − √(ρ22ρ33).
- standard math Robustness duality: R(ρ) = 1 ⟺ ρ ∈ S for the signed-decomposition robustness, with single-qubit formula R = 1 + max{0, |⟨X⟩| + |⟨Y⟩| + |⟨Z⟩| − 1} (SM Eq. S64).
- standard math Bravyi–Kitaev distillation benchmarks: 15-to-1 |H⟩-type protocol distills for ϵ < 0.141 (x + |z| > 1.015); 5-to-1 |T⟩-type distills for t > √(3/7); improved |H⟩ protocols [17] extend to the edge.
- standard math Haar-random amplitude facts: amplitudes are almost surely all nonzero, moduli are exchangeable, and P(|a_{1^n}| is the largest modulus) = 2^(−n).
- domain assumption Zero-temperature phase-covariant channel model: ground-state-preserving with |0⟩ fixed (End Matter Eq. A1; SM S10); thermal fixed points are outside the model.
- standard math Closure properties: stabilizer-polytope membership is preserved under nonzero-probability stabilizer postselection and under tracing out computational-basis ancillas; local channels preserve full separability.
Cite this review
Pith. "Pith review of Sudden death of entanglement, rebirth of magic." pith.science (2026). https://pith.science/paper/SWWHDPZF
@misc{pith2026260522603,
author = {Pith},
title = {Pith review of: Sudden death of entanglement, rebirth of magic},
year = {2026},
howpublished = {\url{https://pith.science/paper/SWWHDPZF}},
note = {Machine review of arXiv:2605.22603}
}
abstract
Local Markovian noise cannot bring entanglement back, but it can bring magic back. Unlike separability, stabilizer membership is not preserved by local channels, allowing dissipation to push states out of the stabilizer polytope as well as in. Under local amplitude damping, the $n$-qubit GHZ family $\alpha|0^n\rangle+\beta|1^n\rangle$ ($0<\alpha<\beta$) loses its magic at a lower damping strength $\gamma_-$ and regains it at a higher one $\gamma_+$, while entanglement is irreversibly lost at $\gamma_e$. This magic--entanglement complementarity, $\gamma_e+\gamma_+=1$ for every $n$, reflects a system--environment duality of amplitude damping. Within real phase-covariant Markovian semigroups the phenomenon is mapped out in full: zero-temperature rebirth occurs if and only if $T_2>T_1$, unital dynamics produce no rebirth, and sufficiently weak thermal excitation confines rebirth to a finite magic island ending in a second sudden death. For small $\alpha$, the reborn magic resides in a fully separable state with all proper marginals stabilizer, yet parity-syndrome extraction concentrates it onto a single qubit for magic-state distillation, without loss of expected robustness and with optimal per-register yield $\alpha^2/2$. Local dissipation further divides pure stabilizer states into magic-generators and magic-insulators: at two qubits, the Bell state $|\Phi^+\rangle$ generates magic immediately, while its Bell-state partner $|\Psi^+\rangle$ remains stabilizer. Together, magic and entanglement reveal a symmetry invisible to either alone.
Figures
Figures from the paper (1 more)
Reference graph
Works this paper leans on
-
[2]
Of these, only GHZ± n lie in the real GHZ-Xmanifold. 3 p00.0 0.5 1.0 p1 0.0 0.5 1.0 c °0.5 0.0 0.5 |c|> min( p0,p1) ∞ =0 |0n i p00.0 0.5 1.0 p1 0.0 0.5 1.0 c °0.5 0.0 0.5 |c|∑ min( p0,p1) ∞° ∞+ ∞ =0 |0n i p00.0 0.5 1.0 p1 0.0 0.5 1.0 c °0.5 0.0 0.5 |c|∑ min( p0,p1) ∞° ∞+ ∞ =0 |0n i (a) (b) (c) (d) (e) (f) 0 0.25 0.5 0.75 1 ∞ 0 0.1 0.2 0.3 c, m c(∞) m(∞) 0...
-
[3]
The robustness excessR(ρ)−1vanishes if and only if a state lies inS(Eqs
Thus the complex GHZ endpoint coherence is controlled by theℓ1 Clifford diamond norm, not by|c|, which is why phase twists can shift magic thresholds while leaving entanglement thresholds unchanged. The robustness excessR(ρ)−1vanishes if and only if a state lies inS(Eqs. (S14)–(S15)), giving an exact stabilizer- membership diagnostic on the mixed-state GH...
-
[4]
The same coordinate determines when the decoded state enters the standard Bravyi–Kitaev distillation windows
The single-qubit stabilizer octahedron|X|+|Z| ≤1hasH-axis half-width1/ √ 2, so ˜ρ(γ)/∈ O ⇐ ⇒ |x|+|z|>1⇐ ⇒h(γ)>1/ √ 2,(S67) matching Theorem 2. The same coordinate determines when the decoded state enters the standard Bravyi–Kitaev distillation windows. The Hadamard twirl maps the sign-corrected Bloch vector(|x|,0,|z|)to the positiveHaxis with polarization...
-
[5]
The ideal decoded state therefore approaches|H⟩super-exponentially in the cat length, while the postselection probability tends to the constant2− √ 2≃0.586
The success probability is psucc =P 0 +P n = 1 +a 2 +ϵ n 2 = 2− √ 2 + ϵn 2 .(S80) 11 Moreover γ⋆ n = 2 lna−1 n +O(n −2), ϵ n =O 2 lna−1 n n ,(S81) witha −1 = √ 2 + 1and2 ln √ 2 + 1 ≃1.763. The ideal decoded state therefore approaches|H⟩super-exponentially in the cat length, while the postselection probability tends to the constant2− √ 2≃0.586. This realiz...
-
[6]
Its value on the Family-B trajectory is Tr(W ρ) = 1 + 2B12 −2B 11 = 1 + 2t.(S104) This matches the primal upper bound and proves Eq
In the last case, 1 + 2 Re(u∗v)−2|u| 2 ∈[ 1−4a 2,1 ]⊆[−1,1].(S103) ThusWis dual feasible. Its value on the Family-B trajectory is Tr(W ρ) = 1 + 2B12 −2B 11 = 1 + 2t.(S104) This matches the primal upper bound and proves Eq. (S96). The prefactorαβ−min(α 2, β2)is non-negative and vanishes if and only ifα=βorαβ= 0. Atα=β= 1/ √ 2, under homogeneous AD, the two...
-
[7]
Therefore Heven = 4µ− √ 3g − √ 3g0 , E even = 2µ− p 4µ2 + 3g2,(S171) with ground eigenvector ratio as in Eq. (S167). In the symmetric odd-parity subspacespan{ D1 3 ,|111⟩}the lowest energy is Eodd =µ− √ 3g. The two even states orthogonal to D2 3 in the nonsymmetric block have energy0, and the two odd states orthogonal to D1 3 have energyµ. SinceE even <0<...
-
[8]
Bothr (n) 1 andr (n) 2 share the leading exponential rate2 −n/2, with r(n) 1 r(n) 2 = 2 1 + 22/n n/2 − →2−1/2 (n→ ∞).(S214) Thus, for any fixed0< r <1, sufficiently largenplaces the trajectory in Regime III. Two-qubit dephased amplitude damping.Amplitude damping composed with a fixed-strength phase-flip dephasing layer Dp(ρ) = (1−p)ρ+pZρZrescales the GHZ ...
Reviewed August 2, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.