{"id":"9528deb6-9cde-45ca-b732-9dc49d2e5905","arxiv_id":"2505.22834","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Counterfactual reasoning yields a unitary model of constructor-based irreversibility, an incoherent homogenizer, new limits on work extraction from coherence, and a von Neumann entropy measure of locally inaccessible information.","lead":"This doctoral thesis uses counterfactuals, statements about which transformations are possible or impossible, to argue that quantum mechanics can consistently describe macroscopic irreversibility, measurement, and locality. It constructs a unitary toy model in which turning pure states into mixed states is possible in a cycle while the reverse is not, and it explores consequences for erasure and work extraction.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The pure/mixed asymmetry in Chapter 5 rests on Eq. 5.6's product-state approximation for reservoir robustness; unchecked reservoir-reservoir correlations for large N could flip the mixed-to-pure limit. Exact checks only reach 3x3.","rationale":"The reader's weakest assumption identifies exactly the product-state approximation in Eq. 5.6 and the reliance on a 3x3 exact check. This is the most load-bearing concern because the central claim of constructor-based irreversibility via the homogenizer depends on the limit behavior of the relative deterioration for large N, and that limit is only computed under the product approximation. Reservoir-reservoir correlations are physically expected in this collision model, so the approximation could substantially change the scaling of the robustness term. The thesis provides some support: the abstract Chapter 4 argument for compatibility of irreversibility with unitary dynamics is logically independent of the homogenizer example, and the Chapter 6 cswap analysis broadens the model class, but neither addresses the specific large-N joint fidelity issue. The experimental Chapter 7 is limited to four qubits and does not test reusability. Therefore the condition placed by the reader is appropriate. I do not propose changing the verdict because the claim is explicitly conditional in the thesis's own discussion (Section 5.5), and the requested check is feasible. If the exact test revealed that the asymmetry disappears for larger N, the verdict would need to move to REJECT for the strong claim, but the current evidence is insufficient to do so.","tokens_in":48115,"tokens_out":8514,"duration_ms":89700,"concrete_test":"Simulate exactly the mixed-to-pure homogenizer for n=3 cycles and reservoir sizes N=4,8,12,16 (state-vector or matrix-product-state simulation), with coupling eta=0.01. Compute the exact joint fidelity F(xi_Ntot, xi_0^{⊗N}) and the exact relative deterioration R_3^N, and compare with the product approximation of Eq. 5.6. If the exact R_3^N grows monotonically with N toward infinity, the product-state assumption is safe; if it saturates or decreases, the asymmetry is not established. Repeat at eta=0.005 to confirm the scaling.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the quantum homogenizer can be used in a cycle for the pure-to-mixed task but not for the mixed-to-pure task, demonstrating constructor-based irreversibility. The impossibility of the mixed-to-pure task in the large-N limit hinges on the relative deterioration R_N^n = epsilon/delta diverging, which requires the robustness delta (Eq. 5.6) to decay sufficiently fast with reservoir size N. Equation 5.6 approximates the joint reservoir fidelity as the product of single-qubit fidelities, implicitly assuming that errors in different reservoir qubits are independent. This is not justified: the reservoir qubits interact with a common sequence of system qubits, so their errors are correlated. For correlated states the joint fidelity with a product state is generally larger than the product of reduced-state fidelities, and in the extreme case of fully correlated errors it can decay linearly in N rather than exponentially. If the exact delta for mixed-to-pure decays only polynomially, the relative deterioration may fail to diverge, and the claimed impossibility is not established. The only exact check, in Section 5.4.3, is a 3x3 homogenizer, which cannot distinguish exponential from polynomial scaling. The thesis itself concedes in Section 5.4.1 that strong-coupling results are unreliable due to neglected entanglement; the weak-coupling argument based on von Neumann entropy (Fig. 5.6) only bounds local entropies, not joint fidelities. Thus the core demonstration of irreversibility is conditional on an unverified approximation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This DPhil thesis uses constructor-theoretic counterfactuals to address irreversibility, measurement, and locality in quantum theory. Part I develops a quantum model of constructors, then analyzes the quantum homogenizer as an example of constructor-based irreversibility: the claim is that the homogenizer can be used in a cycle for the pure-to-mixed task but not for the mixed-to-pure task, yielding an additional cost to erasure beyond Landauer's principle. It also proposes an incoherent cswap homogenizer, reports an NMR implementation, and proves that a known coherent work-extraction protocol cannot be made universal. Part II studies quantum measurement paradoxes, giving a local account of Hardy's paradox and a von Neumann entropy-based measure of locally inaccessible information. The thesis is broad, and several chapters rely on earlier publications.","tokens_in":48455,"tokens_out":6553,"duration_ms":75157,"significance":"The cleanest contribution is Chapter 4, where the unitary, time-reversal-symmetric model of a constructor and the argument that constructor-based irreversibility is compatible with unitary dynamics are presented carefully and soundly. If the Chapter 5 asymmetry between pure-to-mixed and mixed-to-pure transformations could be established rigorously, it would be a notable result: exact, scale-independent irreversibility coexisting with reversible microscopic dynamics, plus a genuine additional cost to erasure. Chapter 8 also contains a clean and explicit demonstration, based on unitary invariance of fidelity, that the preprocessing step of the [15] coherent work-extraction protocol cannot be universal. The experimental and theoretical comparisons in Chapters 6 and 7 are useful supporting material. However, the headline claim of Chapter 5 depends on an uncontrolled product-state approximation, and the general impossibility claims in Chapter 8 depend on an imported theorem and a conjecture, so the manuscript as it stands does not fully establish its most ambitious conclusions.","major_comments":[{"comment":"The central asymmetry claim—mixed-to-pure is impossible while pure-to-mixed is possible—is based on the limits of relative deterioration in Eq. (5.4), with the robustness denominator computed using the product-state approximation in Eqs. (5.5)–(5.6). This approximation neglects reservoir-reservoir entanglement and is load-bearing for the claimed large-N behavior. Because the reservoir qubits interact with a common sequence of system qubits, their errors are correlated; in the extreme case of fully correlated errors, the joint fidelity with the original product state can decay linearly in N rather than exponentially. If the exact robustness decays only polynomially for the mixed-to-pure task, the relative deterioration need not diverge and the claimed impossibility is not established. The exact checks in §5.4.3 are limited to a 3×3 homogenizer, which cannot distinguish exponential from polynomial scaling, and the von Neumann entropy argument in Fig. 5.6 bounds only the sum of local entropies, not the joint fidelity appearing in Eq. (5.2). The thesis itself notes in §5.4.1 that strong-coupling results are unreliable due to neglected entanglement. The manuscript therefore needs either a rigorous bound or a substantially larger exact or tensor-network computation showing the required exponential decay of the robustness, or the conclusion should be explicitly qualified as conditional on the product-state approximation.","section":"§5.3.3 and §5.4.3, Eqs. (5.4)–(5.6)"},{"comment":"The general impossibility of a universal deterministic work extractor from coherence is presented as a consequence of Theorem 1 imported from constructor-theoretic distinguishability, and the explicit quantum demonstration for the [15] protocol is then tied to the conjecture that the mixed-to-pure task is constructor-theoretically impossible. The manuscript states in §8.4 that a universal work extractor using this protocol is possible if and only if the task {mixed state → pure state} is possible, and that this is 'ruled out by the conjecture.' Since the mixed-to-pure impossibility is only demonstrated for the quantum homogenizer and is conjectured to hold generally, the broad claim 'there cannot be a universal work extractor' is not established beyond the homogenizer-based argument. The unitary-invariance argument in §8.3.2 is sound and proves non-universality of the particular protocol, but the chapter should clearly separate that established quantum-mechanical result from the stronger, conjecture-dependent constructor-theoretic claim.","section":"§8.4, §8.2, and §5.5"}],"minor_comments":[{"comment":"The introduction says the reconciliation of macroscopic irreversibility with unitary dynamics is done 'without the need for approximations,' but §5.3.3 introduces the product-state approximation in Eq. (5.5). Please qualify the claim so it does not contradict the later analysis.","section":"Chapter 1, p. 2"},{"comment":"The name Yngvason is misspelled as 'Yvnangson' in §2.2 and as 'Yvnangon' in §3.4; the references should be corrected.","section":"§2.3 and §3.4"},{"comment":"There is a typo: 'seeimingly paradoxical' should be 'seemingly paradoxical.'","section":"§2.3"},{"comment":"The word 'resed' in the first paragraph of §5.5 should be 'reused.'","section":"§5.5"},{"comment":"The notation 'CSW AP' is used in the chapter title and headings before being defined; please introduce the abbreviation explicitly as 'controlled-SWAP (cswap)' and use it consistently.","section":"§6.1.1"}],"recommendation":"major_revision","confidential_remarks":"This is a thesis-length manuscript with several independent contributions. The main technical gap is the unvalidated large-N product-state approximation in Chapter 5, which is load-bearing for the headline result. The author is honest about the conjecture in Chapter 8, but the presentation should not blur the line between proven quantum-mechanical results, imported constructor-theoretic theorems, and conjectures. I see no indication of misconduct; the issues are technical and could in principle be addressed with a more careful asymptotic analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nQuick take: this DPhil thesis is better than the usual compilation. The genuinely new parts are the relative-deterioration analysis of the quantum homogenizer, the cswap homogenizer, the no-go for universal work extraction from coherence, and the Heisenberg-picture quantification of locally inaccessible information. The cleanest single result is Chapter 4's argument that a task can be possible while its transpose is impossible under time-reversal-symmetric dynamics, because the reverse direction requires the right correlations between machine and input. That argument is general and, as far as I can tell, correct.\n\nChapter 5's demonstration with the homogenizer is the load-bearing example. The asymmetry between pure-to-mixed and mixed-to-pure tasks comes from comparing error and robustness via the relative deterioration. The robustness uses the product-state approximation of Eq. (5.6), neglecting reservoir-reservoir correlations. The thesis checks this against exact numerics only up to 3x3 and defends it with a von Neumann entropy argument. The stress-test's worry that the joint fidelity could decay polynomially rather than exponentially is legitimate, but I don't think it sinks the result: the entropy argument does bound the distance between the joint reservoir state and the product of its marginals, which in turn controls the error in Eq. (5.6). That bound isn't made explicit, and the infinite-N,n limit isn't rigorously established. So the asymmetry claim is plausible and probably true, but it falls short of the exact, scale-independent status that constructor-theoretic irreversibility is supposed to have.\n\nChapter 8's impossibility of a universal work extractor from coherence is interesting. The part that rules out generalising the [15] protocol is solid: it follows from unitary invariance of fidelity. The broader constructor-theoretic claim, however, leans on a conjecture that mixed-to-pure is impossible in general, not just for the homogenizer. The paper is honest about that.\n\nOverall: this is a serious piece of work, clearly written, with reproducible recurrence derivations and an actual NMR experiment. The citation pattern is fine; self-citations point to the published versions of the same results. The soft spots are in the asymptotic analysis and the conjectured generality, not in the core ideas.\n\nWho it's for: people working on constructor theory, quantum thermodynamics, and measurement-based foundations. A serious referee should engage with it; for a journal, the strongest chapters (4, 5, 8) could stand as separate papers, and the asymptotic approximation should be tightened before the broad claims are taken as established.\n\nRegards.","headline":"A worthwhile compilation thesis: the clean no-go argument in Chapter 4 and the homogenizer toy model are real contributions, but the central asymptotic irreversibility claim is conditional on an approximation that deserves a more rigorous treatment.","tokens_in":48912,"tokens_out":6505,"would_cite":false,"duration_ms":71467,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A quantum homogenizer shows that exact irreversibility can coexist with time-reversal-symmetric quantum dynamics.","keywords":["counterfactuals","constructor theory","quantum homogenizer","irreversibility","Landauer erasure","Hardy's paradox","locality","von Neumann entropy"],"falsifier":"Compute the exact relative deterioration for a mixed-to-pure homogenizer with $N$ reservoir qubits and $n$ cycles without the product-state approximation, keeping all reservoir-reservoir correlations, for weak but nonzero coupling. If $R_n^N$ fails to diverge in the limit $N\\to\\infty$ followed by $n\\to\\infty$, the claimed impossibility of erasure in a cycle is refuted; the current exact check reaches only a 3-by-3 homogenizer, so this large-$N$ calculation is the decisive open test.","tokens_in":47932,"feed_emoji":"⚛️","tokens_out":6804,"duration_ms":75062,"temperature":0.7,"pith_summary":"This work argues that apparent macroscopic puzzles—irreversibility, measurement, and locality—can be consistently described by universal quantum theory when stated as counterfactuals about which transformations are possible or impossible in a cycle. Its central result is that a quantum homogenizer, a machine of reservoir qubits interacting via partial swaps, can in a cycle transform a pure qubit into a mixed one but cannot transform a mixed qubit into a pure one, even though every interaction is unitary and time-reversal-symmetric. This constructor-based irreversibility implies an additional cost to information erasure beyond the usual Landauer entropy bound. The same counterfactual method is then applied to measurement: apparent Bell-type non-locality is given a local account, and the locally inaccessible information in entanglement is shown to be quantified by von Neumann entropy.","feed_headline":"Quantum eraser works one way but not the reverse in a cycle","feed_subtitle":"A mixed-to-pure transformation cannot be repeated in a cycle, adding a cost to erasure beyond Landauer.","key_machinery":"The central object is the quantum homogenizer, in which a system qubit interacts one by one with $N$ reservoir qubits through the partial-swap unitary $U = \\cos\\eta\\,\\mathbf{1} + i\\sin\\eta\\,S$. The load-bearing quantity is the relative deterioration $R_n^N = \\epsilon_n^N / \\delta_n^N$, the ratio of the error in homogenizing a system qubit to the robustness of the reservoir after $n$ uses. The criterion is that a task is possible for the homogenizer when $R_n^N \\to 0$ in the limit of a large reservoir followed by many cycles, and impossible when it diverges. For the measurement part, the machinery is the Heisenberg-picture local account and the sharpest observable, which connects locally accessible and inaccessible information to the reduced density matrix.","core_discovery":"On the paper's own terms, the central discovery is that exact, scale-independent irreversibility is compatible with unitary quantum dynamics. The proof-model is the quantum homogenizer: the pure-to-mixed task has relative deterioration tending to zero in the double limit of many reservoir qubits and many cycles, so a constructor exists, while the mixed-to-pure erasure task has relative deterioration diverging, so no constructor exists. Because the asymmetry concerns the possibility of performing a task in a cycle rather than the reversibility of a single trajectory, it is not removed by time-reversal-symmetric dynamics. The work infers an additional, non-entropic cost to erasure relevant to Szilard engines, and extends the counterfactual method to measurement, showing that classical reasoning about non-commuting measurements produces apparent contradictions that disappear when the full quantum state of the measurement devices is included.","pith_inferences":["Beyond the paper, the relative-deterioration criterion could be applied to other repeated thermal machines, giving a quantitative test for when a finite reservoir can be treated as a constructor.","If the mixed-to-pure impossibility is a general information-theoretic constraint rather than a feature of the homogenizer, it would supply a scale-independent second law that constrains successor theories of quantum mechanics as well.","The von Neumann-entropy measure of locally inaccessible information suggests that the entropy of a single reduced density matrix in a partially entangled pair could serve as a direct quantitative proxy for the degree of Bell-inequality violation."],"forward_implications":["Pure-to-mixed homogenization can be repeated indefinitely with enough reservoir qubits and weak coupling, while mixed-to-pure erasure cannot; this gives a concrete, exact form of irreversibility in quantum thermodynamics.","Because the asymmetry is about performing a task in a cycle rather than about entropy changes, it predicts an additional cost to erasure beyond Landauer's bound, relevant to Szilard engines and programmable nanomachines.","The same irreversibility appears in an incoherent controlled-swap homogenizer, so the effect does not depend on coherence between system and reservoir.","No universal deterministic work extractor can draw different amounts of work from non-orthogonal coherent inputs; the coherent catalyst in the work-from-coherence protocol must depend on the input state.","Apparent Bell non-locality in Hardy-type experiments is locally accounted for, and the locally inaccessible part of entanglement is measured by von Neumann entropy."],"supporting_citations":[{"why":"defines the quantum homogenizer and its convergence properties, the toy model on which the irreversibility analysis is built.","marker":"[61]"},{"why":"proposes the model of constructor-based irreversibility with unitary dynamics that the thesis generalizes and tests with the homogenizer.","marker":"[14]"},{"why":"is the published source of the relative-deterioration asymmetry between pure-to-mixed and mixed-to-pure tasks.","marker":"[2]"},{"why":"provides the NMR implementation whose results are compared with the theoretical homogenizer evolution.","marker":"[1]"},{"why":"supplies the constructor-theoretic theorem on deterministic work extraction from distinguishable states used to rule out universal work extractors from coherence.","marker":"[16]"},{"why":"is the coherent-work-extraction protocol whose generalization to arbitrary non-orthogonal inputs is proved impossible.","marker":"[15]"},{"why":"defines distinguishability and the constructor theory of information underlying the counterfactual approach.","marker":"[10]"},{"why":"formulates the constructor-theoretic thermodynamics that gives the exact, scale-independent second-law context.","marker":"[13]"}],"fun_headline_variants":["Irreversibility emerges from reversible quantum laws, no contradiction","Quantum erasure costs more than Landauer, says counterfactual analysis","Time-symmetric quantum laws still yield one-way irreversibility","Quantum homogenizer proves exact irreversibility without breaking unitarity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The asymptotic asymmetry between pure-to-mixed and mixed-to-pure tasks assumes the reservoir can be treated as a product state when computing robustness (Sections 5.3.3 and 5.4.3), neglecting reservoir-reservoir entanglement; exact checks are limited to a 3-by-3 homogenizer, so the large-$N$ divergence is not directly verified.","fun_headline_variants_meta":{"raw":{"variants":["Irreversibility emerges from reversible quantum laws, no contradiction","Quantum erasure costs more than Landauer, says counterfactual analysis","Time-symmetric quantum laws still yield one-way irreversibility","Quantum homogenizer proves exact irreversibility without breaking unitarity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001154,"raw_usage":{"total_tokens":4779,"prompt_tokens":939,"completion_tokens":3840,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":555,"completion_tokens_details":{"reasoning_tokens":3767}},"tokens_in":555,"tokens_out":3840,"duration_ms":31409,"temperature":1.0,"reasoning_tokens":3767,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T12:59:02.787688+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the exact relative deterioration for a mixed-to-pure homogenizer with $N$ reservoir qubits and $n$ cycles without the product-state approximation, keeping all reservoir-reservoir correlations, for weak but nonzero coupling. If $R_n^N$ fails to diverge in the limit $N\\to\\infty$ followed by $n\\to\\infty$, the claimed impossibility of erasure in a cycle is refuted; the current exact check reaches only a 3-by-3 homogenizer, so this large-$N$ calculation is the decisive open test.","supporting_citations":[],"review_version":1}