{"id":"0b1fbcae-a5b8-473e-a687-ca96ef26c0bf","arxiv_id":"2608.06094","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For Lipschitz time-dependent Hamiltonians, the new algorithm uses O(alpha T + log(1/epsilon)/log(e + log(1/epsilon)/(alpha T))) HAM-T queries, matching the lower bound for time-independent simulation.","lead":"The authors give a quantum algorithm that simulates time-dependent Hamiltonians with the optimal number of queries to the Hamiltonian oracle. This settles the query complexity of a central open problem and shows time dependence adds no asymptotic query overhead.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified","rationale":"The paper's central claim is that the query complexity of time-dependent Hamiltonian simulation in the HAM-T model matches the known time-independent lower bound. I read the full proof. The key technical steps are the one-query Cayley transducer, the reuse construction, and the weighted combination giving factorial error decay. I verified the local Cayley identity algebra, the global transducer induction, the reuse error identity (UC - PN = C GN(A)Γ) including the telescoping sum, the lower-triangular form of A, the Schur-test bound leading to ∥Q(A)^q∥ ≤ (12eαT/q)^q, the coefficient normalization Λ = 2 - 2^{1-q}, and the robust amplification error bound 3η for η≤1/8. No step appears incorrect. The assumptions are stated in Theorem 9: Lipschitz continuity with known β affects the gate count but not the query count; the Hermitian-oracle requirement is addressed by a footnote that is terse but standard. The reader's weakest-assumption identification is reasonable but not load-bearing, since both assumptions are explicit and the reduction is fixable. The conclusion ACCEPT with MODERATE confidence is appropriate.","tokens_in":27385,"tokens_out":46402,"duration_ms":344523,"concrete_test":"Verify the Hermitian-oracle reduction with a 2x2 example: take a non-Hermitian unitary block-encoding U of a Hermitian H/α, set O' = H_F(|0><1|⊗U + |1><0|⊗U†)H_F, and check that O' is Hermitian, unitary, and (⟨0|_F⟨0|_A⊗I)O'(|0>_F|0>_A⊗I) = H/α. If the compression is correct, the footnote reduction is valid and the central query claim stands.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I find no load-bearing concern in the central query-complexity claim. The proof chain from the Cayley transducer (Lemma 2, Proposition 3) through the reuse error identity (Lemma 4), the lower-triangular structure and factorial decay (Lemma 5, Proposition 6), the coefficient combination (Proposition 7), and the robust amplification (Lemma 8) is internally consistent. I checked the algebra in the local Cayley identity, the telescoping identity in Lemma 4, the Schur-test bound in Proposition 6, the Λ=2-2^{1-q} computation, and the η≤1/8 amplification bound. The Lipschitz and Hermitian-oracle assumptions are explicit in Theorem 9; the latter is reduced by a terse footnote, but a standard Hadamard conjugation on the flag qubit makes the reduction rigorous without changing the query count. No circular dependence or omitted proof was found.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a query-optimal algorithm for simulating a general time-dependent Hamiltonian H(t) on [0,T] in the HAM-T oracle model, assuming H is Lipschitz continuous with norm bound α. The main result, Theorem 9, gives a quantum circuit that approximates the time-ordered propagator U_H(T) to error ε using 12q HAM-T queries, with q = O(αT + log(1/ε)/log(e + log(1/ε)/(αT))), matching the known lower bound for time-independent Hamiltonians. The proof constructs a one-query Cayley transducer with a catalyst state, then applies a weighted combination of reuse circuits so that the error from omitting the catalyst decays factorially. The same framework yields a query-optimal alternative to qubitization for time-independent Hamiltonians, and an appendix gives a continuous-time version of the transducer with no gate-efficiency claim.","tokens_in":27382,"tokens_out":44656,"duration_ms":349395,"significance":"If the result holds, it resolves a long-standing open problem by showing that time dependence incurs no asymptotic query overhead over the optimal time-independent Hamiltonian simulation bound. The proof is detailed and self-contained, with machine-checkable algebra in the Cayley transducer identity, the reuse error identity (Lemma 4), the factorial decay bound (Proposition 6), the coefficient normalization (Proposition 7), and the robust amplification lemma (Lemma 8). The construction is explicit, has no fitted parameters, and the claimed query bound is compared directly against an external lower bound. These are substantial strengths.","major_comments":[],"minor_comments":[{"comment":"The reduction that enforces Hermitian unitary block encodings with one flag qubit should state explicitly how the |+>-state compression interacts with the rest of the construction and whether the exact HAM-T query count is preserved; in the non-Hermitian case the displayed eU_j appears to require separate applications of U_j and U_j^†, which changes the reported 12q constant but not the asymptotic O(·) bound.","section":"Footnote 1"},{"comment":"The displayed identity immediately before the Schur-test application appears as \"RX r=0 ...\" and should be corrected to a summation over r from 0 to R, since the surrounding text is otherwise clear.","section":"§6.2, Proposition 6 proof"},{"comment":"The statement \"The circuit uses 12q queries\" should explicitly separate the αT ≤ ε case, where the identity circuit uses zero queries and q is not defined; currently the proof defines q only in the αT > ε branch.","section":"Theorem 9"},{"comment":"It would be helpful to state explicitly that each of the 4q loop layers contains a HAM-T query gate even when the work-qubit control is inactive for a particular selected N, so that SELECT uses exactly 4q oracle calls rather than N calls.","section":"§7.2, Algorithm 1"},{"comment":"The inequality log(e + L/(αT)) ≤ 2L for αT > ε and ε ≤ 1/2 is asserted without justification; a one-line proof would improve readability.","section":"Theorem 9 proof"}],"recommendation":"minor_revision","confidential_remarks":"The paper is technically strong and squarely within the scope of the journal. The minor comments are presentation-level only; none affects the central query-complexity claim. The AI-use disclosure is unusual but transparent and does not alter the scientific assessment."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague, the short version: this is the real thing. It closes the long-open gap between upper and lower bounds for time-dependent Hamiltonian simulation in the HAM-T model, and it does so with a genuinely new construction rather than a minor tweak. You should read it.\n\nWhat is new: the one-query Cayley transducer followed by weighted reuse with factorial error cancellation. The generic Belovs–Jeffery–Yolcu reuse gives O(1/epsilon^2) precision dependence; here they show that for their specific transducer, combining different reuse lengths makes the omitted-catalyst error decay factorially, yielding the optimal log(1/epsilon) dependence. They also obtain a query-optimal alternative to qubitization for time-independent Hamiltonians. The proof chain is detailed and internally consistent. I checked the local Cayley identity, the reuse error identity, the lower-triangular structure of the private block, the Schur-test factorial bound, the coefficient normalization with Lambda = 2 − 2^{1−q}, and the robust amplification lemma. Everything holds together. No fitted parameters, no circularity.\n\nSoft spots, mostly minor. The main theorem requires Lipschitz continuity with known beta because the gate-efficient discretization chooses J ~ max{q, beta T^2/epsilon, (alpha T)^{3/2}/sqrt(epsilon)}. That assumption is explicit, and it affects gate complexity, not the query bound. The Hermitian-oracle requirement is dispatched in a footnote via a flag qubit; the reduction is standard and correct but terse. The gate complexity is stated with suppressed polylog factors, which is acceptable when the headline result is query complexity. The continuous-time appendix is interesting but explicitly makes no gate claim. The paper also discloses that LLMs generated the central proof strategies, with the human authors stating they verified and rewrote them. I found the proofs coherent and checkable, not like generated filler. A machine-checked formalization would be the natural next step, but its absence does not undermine the result.\n\nWho this is for: anyone working on Hamiltonian simulation, quantum algorithms for differential equations, or transducer techniques. It deserves a serious referee. I would send it to review as is, expecting only minor revision on exposition—most notably moving the Hermitian-reduction footnote into the main text—and no substantive changes.","headline":"Genuine resolution of the optimal query complexity for time-dependent Hamiltonian simulation, with a new transducer construction and proofs that withstand close scrutiny.","tokens_in":28002,"tokens_out":1763,"would_cite":true,"duration_ms":14176,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68","68Q12"],"pacs":["03.67.Lx"],"model":"deepseek-v4-flash","headline":"Simulating time-dependent Hamiltonians needs no extra queries","keywords":["time-dependent Hamiltonian simulation","HAM-T model","query complexity","transducer","Cayley transform","oblivious amplitude amplification","block-encoding","quantum algorithms"],"falsifier":"A concrete check is to simulate a small system whose exact propagator can be computed numerically, such as $H(t) = a\\,\\sigma_x + b\\sin(\\omega t)\\,\\sigma_z$ on one qubit, choose $\\alpha$ and $\\beta$ from $a,b,\\omega$, and verify that the constructed circuit with $J = \\max\\{q, \\beta T^2/\\varepsilon, (\\alpha T)^{3/2}/\\sqrt{\\varepsilon}\\}$ and $q$ chosen by Theorem 9 achieves output error at most $\\varepsilon$; if any instance violates the promised error, the estimate (13) or the factorial bound (44) is wrong. A more direct refutation of the tightness claim would be a Lipschitz Hamiltonian in the HAM-T model whose query complexity is asymptotically larger than the stated bound.","tokens_in":27078,"feed_emoji":"⚛️","tokens_out":9163,"duration_ms":74120,"temperature":0.7,"pith_summary":"This paper claims to settle the query complexity of time-dependent Hamiltonian simulation in the standard black-box oracle model (HAM-T), where a query returns a controlled block-encoding of the Hamiltonian. For any Lipschitz-continuous Hamiltonian $H(t)$ with norm at most $\\alpha$, it constructs a circuit that approximates the time-ordered propagator $U_H(T)$ to error $\\varepsilon$ using $O(\\alpha T + \\log(1/\\varepsilon)/\\log(e+\\log(1/\\varepsilon)/(\\alpha T)))$ queries. This is the same asymptotic count as the known lower bound for time-independent Hamiltonians, so time dependence, in this model, costs no extra queries. The argument works by building a one-query transducer that implements a discretized evolution while leaving an auxiliary catalyst state untouched, then combining many reuse circuits with carefully chosen weights so that the error from omitting the catalyst decays factorially.","feed_headline":"Time-dependent simulation matches optimal query bound","feed_subtitle":"A transducer-based algorithm simulates Lipschitz Hamiltonians with the same HAM-T queries as time-independent ones.","key_machinery":"The load-bearing object is the one-query Cayley transducer: a unitary $S$ that, when supplied with an auxiliary catalyst state $\\Gamma|\\psi\\rangle$, uses a single HAM-T query to implement an approximation $U_C$ of $U_H(T)$ and returns the catalyst unchanged. The catalyst is a superposition over time-grid labels of local states, and its norm is bounded by $\\sqrt{\\alpha T}$. Removing the catalyst is governed by the identity $U_C - P_N = C\\,G_N(A)\\Gamma$, where $A$ is the private-to-private block; the key structural fact is $A = i(I - \\Delta)\\mathrm{HAM\\text{-}T}$, whose lower-triangular time-ordered form suppresses transitions that do not respect chronological order. The polynomial $Q(z) = (1+z^2)/2$ then cancels the identity part of $A^2$, so every term in $Q(A)^q$ contains at least $q$ time-ordered factors, yielding the factorial bound $\\|Q(A)^q\\| \\leq (12e\\alpha T/q)^q$. A weighted combination of reuse circuits with coefficients $\\lambda_N$ derived from $E_q(z)=Q(z)^q G_{2q}(z)$ keeps the LCU normalization below $2$, allowing one step of oblivious amplitude amplification to complete the simulation.","core_discovery":"The paper's central claim is Theorem 9: given HAM-T access to a Lipschitz-continuous Hamiltonian $H(t)$ on $[0,T]$ with $\\|H(t)\\| \\leq \\alpha$, there is a quantum circuit $U_{\\mathrm{sim}}$ such that $\\|U_{\\mathrm{sim}}(|0\\rangle |\\psi\\rangle) - |0\\rangle U_H(T)|\\psi\\rangle\\| \\leq \\varepsilon$ for every unit state, using $12q$ HAM-T queries with $q = O(\\alpha T + \\log(1/\\varepsilon)/\\log(e+\\log(1/\\varepsilon)/(\\alpha T)))$. Because the same expression is a lower bound for time-independent simulation, the result shows that general time dependence incurs no asymptotic query overhead. The construction first approximates the evolution by a product of Cayley transforms on a time grid, encodes the whole product into a single one-query transducer with a catalyst of norm $\\sqrt{\\alpha T}$, and then removes the catalyst by a weighted combination of reuse circuits. The chosen weights $E_q(z) = ((1+z^2)/2)^q G_{2q}(z)$ make the residual error drop faster than any polynomial, giving the logarithmic dependence on $1/\\varepsilon$. For time-independent $H$, the same method provides a query-optimal alternative to qubitization, the standard block-encoding technique for optimal Hamiltonian simulation.","pith_inferences":["If the stated query bound is tight, it gives a complete characterization of coherent-oracle HAM-T complexity for Lipschitz time-dependent Hamiltonians; the same transducer-reuse technique may extend to other coherent access models, where clock-embedding overhead could similarly be eliminated.","The continuous-time-register transducer in the appendix is an editorial step beyond the paper's main gate-complexity claim: if an efficient finite-dimensional circuit for that restricted transducer were found, it could remove the time-discretization error and potentially lower the gate count while retaining the optimal query count.","A natural testable extension is the low-energy subspace: because the catalyst norm and omitted-catalyst error are controlled by the full Hamiltonian norm, one could try to replace $\\alpha T$ by the restricted evolution norm for states in a low-energy subspace, which would improve adiabatic state preparation costs.","The coefficient-construction idea here—choosing reuse lengths so that error terms cancel and only ordered transitions survive—may generalize to removing catalysts from other transducers, providing a general subroutine for state conversion with logarithmic overhead in $1/\\varepsilon$."],"forward_implications":["The HAM-T query complexity of time-dependent Hamiltonian simulation is $\\Theta(\\alpha T + \\log(1/\\varepsilon)/\\log(e+\\log(1/\\varepsilon)/(\\alpha T)))$ for Lipschitz Hamiltonians, matching the time-independent lower bound.","Time dependence itself causes no asymptotic query overhead; the Lipschitz constant only affects how finely the time grid must be sampled, i.e., gate and ancilla complexity.","For $H(t) \\equiv H$, the construction yields a query-optimal simulation algorithm that does not use qubitization, giving a different optimal method for the time-independent problem.","The result applies without assuming periodicity, locality, or any structural decomposition of $H(t)$, covering worst-case black-box access.","Because the reuse lengths are at most $4q$ and the circuit uses $12q$ queries, the precision dependence is polylogarithmic rather than the quadratic dependence of the generic transducer reuse bound."],"supporting_citations":[{"why":"It introduces the transducer abstraction and the finite-reuse construction whose generic precision bound the paper improves.","marker":"[BJY24]"},{"why":"It defines the HAM-T oracle model and gives the truncated Dyson-series algorithm that is the baseline to beat.","marker":"[L W18]"},{"why":"It establishes the time-independent query lower bound and qubitization framework that the result matches.","marker":"[LC19]"},{"why":"It supplies the block-encoding and linear-combination-of-unitaries machinery used to combine the reuse circuits.","marker":"[GSL W19]"},{"why":"It provides a truncated Dyson-series time-dependent simulation whose query complexity is compared and improved in the table.","marker":"[KSB19]"},{"why":"It gives the linear-combination-of-unitaries and oblivious amplitude-amplification primitives used in the final amplification step.","marker":"[BCC+15]"},{"why":"It supplies the discrete-clock embedding method whose precision overhead the new construction avoids.","marker":"[WWRL24]"},{"why":"It supplies the Floquet-Hilbert-space method that is optimal for periodic and multiperiodic Hamiltonians but not general Lipschitz ones.","marker":"[Miz23]"}],"fun_headline_variants":["Time-dependent simulation matches static query bound","Time dependence adds no query overhead","Optimal queries for time-dependent Hamiltonians","Transducer method yields optimal time-dependent simulation","Same query complexity for time-dependent and static cases"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that $H(t)$ is Lipschitz continuous with a known Lipschitz constant $\\beta$, because the time-grid size $J$ is chosen as $\\max\\{q, \\beta T^2/\\varepsilon, (\\alpha T)^{3/2}/\\sqrt{\\varepsilon}\\}$; without a known $\\beta$, the discretization error bound (13) and the finite circuit construction do not hold, even though the query bound might still be achievable.","fun_headline_variants_meta":{"raw":{"variants":["Time-dependent simulation matches static query bound","Time dependence adds no query overhead","Optimal queries for time-dependent Hamiltonians","Transducer method yields optimal time-dependent simulation","Same query complexity for time-dependent and static cases"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000242,"raw_usage":{"total_tokens":1573,"prompt_tokens":1042,"completion_tokens":531,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":658,"completion_tokens_details":{"reasoning_tokens":467}},"tokens_in":658,"tokens_out":531,"duration_ms":5030,"temperature":1.0,"reasoning_tokens":467,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:59:50.666055+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete check is to simulate a small system whose exact propagator can be computed numerically, such as $H(t) = a\\,\\sigma_x + b\\sin(\\omega t)\\,\\sigma_z$ on one qubit, choose $\\alpha$ and $\\beta$ from $a,b,\\omega$, and verify that the constructed circuit with $J = \\max\\{q, \\beta T^2/\\varepsilon, (\\alpha T)^{3/2}/\\sqrt{\\varepsilon}\\}$ and $q$ chosen by Theorem 9 achieves output error at most $\\varepsilon$; if any instance violates the promised error, the estimate (13) or the factorial bound (44) is wrong. A more direct refutation of the tightness claim would be a Lipschitz Hamiltonian in the HAM-T model whose query complexity is asymptotically larger than the stated bound.","supporting_citations":[],"review_version":1}