{"id":"65a5ca5a-bb95-4022-840b-85e0b2380014","arxiv_id":"2607.19156","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A route–truncate–repair pipeline converts quantum adaptive agents' entropic memory savings into a smaller physical memory dimension with a certified fidelity-divergence rate.","lead":"This paper shows how to shrink the memory dimension of quantum adaptive agents by routing a reference input stream through the agent, compressing the resulting matrix-product-state bond, and repairing the update to keep it physically valid. It certifies the accuracy-memory trade-off with a fidelity-divergence rate and demonstrates up to 128× dimension reduction on two benchmark tasks.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Practical dimension reduction depends on unproven spectral decay of the driven memory; the formal validity and certificate results are sound, so this is a scope gap rather than a refutation.","rationale":"The paper's formal core is careful and conditional: Theorem 1 proves the routed MPS construction, Theorem 2 proves validity of the polar-completed reduced agent and an exact fidelity-divergence certificate under explicit mixing and nonzero-overlap assumptions, and the supplement supplies detailed proofs. The numerical implementation is archived, validated with fine-grid convergence checks, and the authors explicitly flag the reference-relative nature of the certificate and the unproven discarded-weight guide. The reader's weakest assumption—spectral decay of the driven memory—is a real limitation on practical scope, but it is not an internal inconsistency or an unsupported formal step. The benchmarks demonstrate the phenomenon, not a universal law. Therefore no verdict change is warranted; the concern is a scope caveat that should be kept in mind when interpreting the abstract's general phrasing.","tokens_in":28628,"tokens_out":14486,"duration_ms":158222,"concrete_test":"Run the archived pipeline (scripts/reproduce.py) on a modified resettable-clock family with power-law survival probabilities, e.g. Φ(n) = (1 + n/N)^{-α} for α ~ 1–2, at N = 256 with the design reference pR(x=1)=0.04 and target δ* = 10^-2 bits/step. If the smallest retained dimension d⋆ grows like O(N) or N^β with β close to 1, the claimed route from entropic advantage to dimension reduction is not generic; if d⋆ remains small, the spectral-decay concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's practical promise—that entropic quantum memory advantages convert into dimension reduction—requires the stationary driven memory ρ_M^(R) to have a rapidly decaying eigenvalue spectrum. Corollary 1 and the truncation in SM.K.3 select the subspace by this spectrum, and the reported 128× and ~9.5× reductions rely on the first two eigenvalues dominating. The paper proves no general link between the entropic advantage C_μ − C_q and this spectral tail. A process with a genuine entropic advantage but a heavy-tailed or nearly flat memory spectrum would still satisfy Theorem 1 and Theorem 2, yet the truncation would retain a large fraction of the original dimension, so the headline dimension reduction would fail. The paper is transparent about this: the certificate is reference-relative, the discarded-weight guide is reported only as a numerical observation (SM.G, Numerical observation S1), and the two benchmarks do not establish generic spectral decay. Thus the central mathematical claims are not threatened, but the scope of the practical claim is narrower than the abstract's wording might suggest.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a route-truncate-repair procedure for reducing the Hilbert-space dimension of quantum adaptive agents while approximately preserving their input–output behaviour. A stationary finite-memory reference input process is routed through the agent, producing a uniform temporal matrix product state whose virtual bond is the joint reference–agent memory. The authors prove that this MPS is left-canonical and that its canonical bond state is the stationary fixed point of the driven memory channel. They then truncate the agent-local marginal spectrum and repair the projected Kraus operators stimulus-wise via a polar-type normalization, yielding a valid quantum instrument that accepts arbitrary input strings. The main quantitative result is an exact asymptotic fidelity-divergence certificate, R_F^(Q) = -1/2 log_2 μ, where μ is the spectral radius of a rectangular mixed transfer operator. Two benchmarks — a resettable renewal clock and an adaptive cyclic walk — show substantial dimension reductions at a target fidelity-divergence rate of 10^-2 bits/step, with numerical residuals around 10^-14. Full proofs, numerical implementation, and reproducibility data are provided in the Supplemental Material.","tokens_in":28921,"tokens_out":7805,"duration_ms":87970,"significance":"If the results stand, the paper provides a concrete bridge between entropic quantum memory advantages and physically realizable dimension reductions, going beyond earlier MPS-truncation work for passive stochastic processes. The main theorems are carefully stated and supported by detailed proofs in the supplement. The exact spectral-radius certificate is a genuine strength, and the numerical validation is unusually thorough: completeness residuals below 1.2e-14, dominant-eigenpair residuals below 1e-14, and crossed-checked thresholds. The authors are also commendably transparent about scope: the discarded-weight guide is explicitly labelled as a numerical observation, not used for selection, and the certificate is reference-relative. The principal caveat is that the practical magnitude of compression depends on the spectral decay of the driven stationary memory state; no general theorem links the entropic advantage C_μ - C_q to that spectral tail. This does not threaten the validity or certificate theorems, but it narrows the universal reading of the abstract's promise. The paper includes a reproducibility package and openly available data, which further strengthens its contributio","major_comments":[],"minor_comments":[{"comment":"The abstract states that the procedure 'converts entropic quantum memory advantages into reductions in memory dimension.' This is stronger than what is proven: Theorems 1 and 2 guarantee a valid compressed agent and an exact certificate for any process, but the amount of dimension reduction is controlled by the eigenvalue decay of the driven stationary memory ρ_M^(R) (Corollary 1; SM.K.1, SM.K.3). A process with an entropic advantage but a flat or heavy-tailed memory spectrum would satisfy the theorems but yield little compression. I recommend tempering the wording, e.g. 'can convert' or 'as demonstrated on benchmark processes', and adding a one-sentence caveat in the abstract or introduction.","section":"Abstract / Outlook"},{"comment":"The figure labels appear garbled: panel (c) shows '± ⋆ = 10−2' and 'clock design (0:04)'; these should presumably be 'δ⋆ = 10−2' and 'p_R(x=1)=0.04'. Please correct the typesetting and ensure all symbols are defined in the caption.","section":"Fig. 3 / End Matter"},{"comment":"The paper is careful to state that the discarded-weight curve is not used to select retained dimensions and that the natural per-cut bound is false. This is good practice, but the observation is buried in the supplement. Since Fig. 3 visually invites comparison with the discarded-weight guide, I suggest adding a sentence near Eq. (5) or in the main text making clear that no general O(ε_M) law is claimed and that the exact rate is the only selection criterion.","section":"SM.G.2 / Numerical observation S1"},{"comment":"The proof of Theorem 2(ii) invokes 'mixing and nonzero-overlap conditions of Supplemental Material, Sec. SM.G' and the 'common Kraus-label alphabet.' These are crucial for the validity of Eq. (5). A short statement of the conditions in the main text, or at least a boxed definition of the common label convention, would make the theorem self-contained enough for a reader who does not immediately consult the supplement.","section":"Theorem 2(ii)"},{"comment":"Reference [48] is cited as a versioned Zenodo archive but no DOI or version identifier is given in the bibliography. Since the reproducibility claim rests on this archive, please include the full DOI or version string.","section":"References"}],"recommendation":"minor_revision","confidential_remarks":"The paper is technically sound and the supplement is exemplary. My only substantive request is to qualify the practical scope of the abstract: the conversion from entropic advantage to dimension reduction is demonstrated, not proven generically, and the paper itself is transparent about this in SM.G. With that wording adjusted, I would be happy to see it accepted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two quick takes. The route–truncate–repair construction is genuinely new: it turns branching input trees into a linear MPS by routing a reference process, gives a canonical bond to truncate, and repairs the projected updates into valid quantum instruments. That is a real step beyond earlier MPS compression for passive stochastic processes. The paper's formal core is sound and the presentation is unusually honest.\n\nWhat it does well: Theorem 1 is clean — the routed Kraus operators are left-canonical by construction, and the bond state is the driven memory. Corollary 1 is a standard Ky Fan truncation, correctly scoped as retained stationary weight, not behavioral optimality. Theorem 2 supplies an exact fidelity-divergence rate via the mixed transfer; the supplement gives the mixing/injectivity assumptions and full proofs. The numerics are credible: the archive is versioned, the reported residuals are tiny, quadrature refinements are checked, and the selection of d is made on the exact rate, not on a fitted curve. The self-citations are legitimate; this is a direct continuation of [8] and [30,31].\n\nThe soft spot is exactly what the stress test flags: the headline 128× and 9.5× reductions depend on the driven memory's spectrum decaying fast. The theorems prove validity and certify accuracy for any truncation; they do not prove that a small d captures almost all stationary weight. That is a scope gap, not a flaw — the paper says so, explicitly keeps the discarded-weight guide as an unproven numerical observation, and does not use it to select dimensions. But the abstract's \"establish a route\" is stronger than what is demonstrated: the route is conditional on spectral decay and on a well-chosen reference. A process with a genuine entropic advantage and a flat memory spectrum would satisfy the theorems and fail the headline reduction. The reference-relative nature of the certificate is also worth remembering: validity holds for arbitrary inputs, but the accuracy guarantee is tied to the reference process; the robustness test covers only the clock.\n\nThis paper deserves a serious referee. The flaws are not load-bearing: they are honest boundary conditions on a useful construction. I would send it out, expect the authors to soften the abstract or add a caveat, and cite it in my own work.","headline":"Solid formal core with an honest scope gap: route–truncate–repair is new and correct, but the headline dimension reductions depend on spectral decay of the driven memory that is demonstrated, not proven.","tokens_in":29353,"tokens_out":1999,"would_cite":true,"duration_ms":22504,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Routing an adaptive agent through a reference input process turns its memory into a tensor-network bond that can be cut and repaired, producing smaller quantum agents with a certified accuracy trade-off.","keywords":["quantum adaptive agents","memory dimension reduction","matrix product states","tensor network truncation","quantum instruments","fidelity divergence","input-output processes","reference routing"],"falsifier":"Compute the stationary memory state ρ(R)_M for a concrete agent and reference process, then check the tail weight Σ_{i>d} λ_i at increasing retained dimensions d; if this tail stays large (say ≥ 1/2) at every d up to the full dimension, no small truncation can reach a target fidelity rate, directly contradicting the claimed practical reduction.","tokens_in":28544,"feed_emoji":"⚛️","tokens_out":2361,"duration_ms":24729,"temperature":0.7,"pith_summary":"The paper claims that the memory of a quantum adaptive agent can be compressed in dimension, not just in entropy, by a route–truncate–repair procedure. Routing a stationary reference input process through the agent produces a uniform matrix product state whose bond encodes the agent's memory; truncating that bond to its most-occupied subspace and locally repairing each stimulus-conditioned update yields a smaller, physically valid agent. The paper proves the repair yields a valid quantum instrument and provides an exact asymptotic fidelity-divergence rate that certifies the trade-off between accuracy and memory size. Benchmarks on a resettable clock and a cyclic walk show large dimension reductions while preserving behaviour under a specified fidelity threshold. If correct, this converts entropic memory advantages that previously did not reach hardware into practical memory-dimension reductions.","feed_headline":"Truncating memory shrinks quantum agents up to 128-fold","feed_subtitle":"Route-truncate-repair converts entropic savings into smaller physical memories with a certified fidelity bound.","key_machinery":"The routed isometry W = Σ|x⟩⟨x|⊗V_x, with a stationary reference input process R contracted onto the stimulus rail, gives joint Kraus operators L_ω = √R |c′⟩⟨c|⊗K that automatically satisfy Σ L_ω†L_ω = 1, placing the history in left-canonical MPS form. The identity carrying the argument is the polar repair: for each stimulus, projected Kraus operators are normalised by G_x^{-1/2} so that the per-stimulus completeness relation Σ ̃K†̃K = P_M holds, making the compressed agent a valid quantum instrument. The mixed transfer operator Z ↦ Σ ̃L_ω Z L_ω† supplies the exact asymptotic fidelity-divergence rate through its spectral radius.","core_discovery":"The central discovery is that a driven quantum adaptive agent has a uniform matrix product state representation whose canonical bond is the stationary state of the agent's memory, and that agent-local spectral truncation of this bond followed by per-stimulus polar repair produces a trace-preserving quantum instrument with a computable fidelity-divergence certificate. The asymptotic fidelity loss rate equals minus half the log of the spectral radius of the mixed transfer operator between the original and repaired histories. This provides a direct bridge from information-theoretic memory savings to physical Hilbert-space dimension savings for adaptive agents.","pith_inferences":["The certificate is stated for classical, diagonal reference processes; extending it to coherent or temporally correlated input testers, which the paper itself identifies as open, would determine whether the trade-off holds for genuinely quantum stimuli.","If the stationary memory spectrum is flat, the dimension reduction collapses even though the validity and certificate theorems remain true; designing reference processes to shape the spectrum could be a practical strategy for harder agents.","The reference input process acts like a training distribution, so an agent could plausibly be recompressed online as the operating environment changes, with a fresh certificate each time.","The same routing construction naturally connects to stationary quantum combs and process tensors, suggesting that dimension reduction for temporal quantum information processors may follow from similar truncation-and-repair ideas."],"forward_implications":["Entropic quantum memory advantages can be converted into physical memory-dimension reductions, not just information-cost reductions.","Compressed agents remain physically valid instruments and can respond to arbitrary input sequences, not only the reference process used for compression.","The fidelity-divergence certificate gives an exact asymptotic rate quantifying the trade-off between retained memory dimension and behavioural fidelity.","In the benchmarks, the resettable clock compresses from dimension 256 to 2 (a 128-fold reduction) and the cyclic walk from 256 to 27 (roughly 9.5-fold), both at a per-step rate below 10⁻² bits.","The routed MPS representation opens the door to variational and more sophisticated tensor-network compression methods for adaptive agents."],"fun_headline_variants":["Quantum agents shrink via memory truncation","Paring memory downsizes quantum agents","Truncate memory, shrink quantum agent size","Memory pruning shrinks quantum adaptive agents","Cut quantum memory, reduce agent dimension"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The practical dimension reductions rest on the stationary memory state's eigenvalues decaying fast enough that a small retained dimension captures almost all stationary weight; a process whose memory spectrum is flat would keep the validity and certificate theorems intact but destroy the headline savings.","fun_headline_variants_meta":{"raw":{"variants":["Quantum agents shrink via memory truncation","Paring memory downsizes quantum agents","Truncate memory, shrink quantum agent size","Memory pruning shrinks quantum adaptive agents","Cut quantum memory, reduce agent dimension"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000126,"raw_usage":{"total_tokens":903,"prompt_tokens":652,"completion_tokens":251,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":396,"completion_tokens_details":{"reasoning_tokens":188}},"tokens_in":396,"tokens_out":251,"duration_ms":3413,"temperature":1.0,"reasoning_tokens":188,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T13:16:16.727972+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the stationary memory state ρ(R)_M for a concrete agent and reference process, then check the tail weight Σ_{i>d} λ_i at increasing retained dimensions d; if this tail stays large (say ≥ 1/2) at every d up to the full dimension, no small truncation can reach a target fidelity rate, directly contradicting the claimed practical reduction.","supporting_citations":[],"review_version":1}