{"id":"9c6523c6-19cb-47bb-a34e-57c6ea9d885a","arxiv_id":"2411.10999","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":3,"one_line_summary":"An explicit qubit-based quantum circuit for Maxwell's equations with PEC boundaries and time-dependent sources is constructed via Schrödingerization and autonomization, with gate-complexity analysis.","lead":"This paper designs explicit quantum circuits for simulating Maxwell's equations with conducting boundaries and time-dependent sources by converting the equations into a larger Schrödinger system. The claimed payoff is a polynomial speedup over classical solvers, but the complexity comparison and some accuracy claims are overstated.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The advertised 'almost log log 1/epsilon' qubit overhead is unsupported by the paper's own error analysis: Lemma 5.5 gives algebraic O(Delta_p^r + Delta_s^r) convergence with fixed r, which forces n_p, n_s = O(log(1/epsilon)), not O(log log(1/epsilon)).","rationale":"I read the paper as a genuinely constructive step: it provides explicit circuits for the Schrodingerized Maxwell system with PEC boundaries and sources, and the numerical experiments validate the recovery step (though not the full quantum circuit). The central advertised claim is the almost log log(1/epsilon) qubit increase from the extra dimensions, and this is where the argument is least secure. The paper's own Theorem 5.1 and Eq. (88) imply n_p = O(log(ln(1/epsilon)/epsilon^{1/r})) = O(log(1/epsilon)) for any fixed smoothness order r; the only explicit g and beta are C^2 and H^2, giving second-order convergence in the numerics. To reach log log(1/epsilon), one needs exponential or at least superalgebraic convergence with constants controlled as r grows, but no such estimate is given. This is more load-bearing than Assumption 4.1, because Assumption 4.1 is honestly stated as an assumption, whereas the log-log claim is presented as a consequence of the analysis. The classical baseline comparison also appears inflated (the paper cites FDTD cost as O(epsilon^{-1-d/2}) while the standard explicit FDTD cost is O(epsilon^{-(d+1)/2}), but the higher-order Trotter argument gives an independent speedup claim, so that issue is less decisive. My read agrees with the reader's CONDITIONAL verdict but identifies the unsupported spectral-convergence claim as the deepest unresolved premise. I recommend keeping the verdict at CONDITIONAL, requiring the authors to either prove exponential convergence with controlled constants or revise the abstract and conclusion to state an O(log(1/epsilon)) qubit overhead for the demonstrated constructions.","tokens_in":1123,"tokens_out":1398,"duration_ms":158097,"concrete_test":"Fix the explicit g(p) from Eq. (35) and beta from Eq. (34), and numerically measure the relative recovery error in Eq. (83) for the one-dimensional Maxwell test with T=1/2 over a range of N_p and N_s, e.g., from 2^8 to 2^16. For target errors from 1e-4 to 1e-8, record the required n_p and n_s. If they grow proportionally to log(1/epsilon) with the observed second-order rate, the log-log claim is falsified. As an analytic check, re-derive Lemma 5.5 tracking the H^r norm of the solution in the spectral error bound; unless that norm grows at most subexponentially in r, the 'essentially spectral accuracy' statement cannot support n_p = n_s = O(log log(1/epsilon)). A construction of g_r and beta_r for arbitrary r with explicit norm growth would settle the question.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract and conclusions claim that the two extra dimensions (from Schrodingerization and autonomization) increase the qubit count only by almost log log(1/epsilon). However, the proof route in Section 5.1 uses Lemma 5.5, which contains an algebraic error bound of order Delta_p^r + Delta_s^r with r fixed. The paper's only explicit constructions are g in C^2 (Eq. 35) and beta in H^2 (Eq. 34), and the numerical table shows second-order convergence. To obtain n_p = n_s = O(log log(1/epsilon)), one would need exponential or superalgebraic convergence with constants controlled as r grows, but no estimate for the H^r norm of the solution or for the smoothness of g and beta as r grows is provided. Equations (88)-(89) in the proof of Theorem 5.1 actually set n_p, n_s = O(log(ln(1/epsilon)/epsilon^{1/r})) = O(log(1/epsilon)) for a fixed r. Therefore the paper's advertised log-log overhead is not a consequence of its own estimates. This is a load-bearing gap because the claimed minor increase in qubits and the resulting complexity advantage over classical FDTD depend on it. If n_p and n_s are actually O(log(1/epsilon)), every gate count in Theorem 5.1 gains an extra log(1/epsilon) factor, weakening the advertised advantage. This concern is independent of Assumption 4.1, which is an explicit restriction on sources; the log-log claim is presented as a derived result without the necessary supporting estimates.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper gives an explicit qubit-level quantum circuit construction for Maxwell's equations with perfect electric conductor (PEC) boundary conditions and time-dependent source terms, based on Schrödingerization and autonomousization. The semi-discrete Maxwell system is converted into a source-free ODE by auxiliary variables, then into an autonomous Schrödinger-type Hamiltonian via the warped phase transformation, and the resulting time evolution is decomposed into three terms (HDs, HF, Hcurl) that are implemented with concrete Trotterized circuits, including multi-controlled rotations and QFT-based spectral discretizations. The paper also provides an error analysis (Lemmas 5.2 and 5.5), a complexity theorem (Theorem 5.1), a comparison with classical FDTD, and numerical experiments on a one-dimensional reduced Maxwell problem.","tokens_in":24249,"tokens_out":8437,"duration_ms":93284,"significance":"If the advertised claims hold, this is a valuable step beyond black-box Hamiltonian simulation for Schrödingerization: the circuit constructions for the source coupling and for the PEC boundary terms are explicit, the gate counts are concrete, and the numerical experiments verify the recovery and convergence mechanism. The paper's strongest contribution is the detailed circuit decomposition and the associated Trotter-error analysis, which are not merely existence arguments. However, the central advertised feature, namely that the two extra dimensions increase the qubit count by only almost log log(1/epsilon), is not supported by the paper's own estimates; and the treatment of 'time-dependent source terms' is restricted to a narrow structural assumption. These issues affect the framing and the main complexity comparison, so the paper needs revision before the claims can be accepted.","major_comments":[{"comment":"The claim that the extra p- and s-dimensions add only 'almost log log 1/epsilon' qubits is not supported by the manuscript's error analysis. Lemma 5.5 bounds the spectral and delta-approximation errors by algebraic quantities O(Delta_p^r + Delta_s^r) with a fixed r, and the proof of Theorem 5.1 explicitly chooses Np, Ns = O(ln(1/epsilon)/epsilon^{1/r}), giving n_p, n_s = O(log(1/epsilon)) for fixed r. The paper's explicit smoothness examples are g in C^2(R) and beta in H^2(R), and the numerical table shows second-order convergence, not spectral convergence. No estimates for the H^r norms of the solution or for the r-dependence of constants are provided, so the 'essentially spectral accuracy' statement in §3.4 cannot be used to replace the algebraic bound. This is load-bearing because the abstract, the introduction, and the conclusions all advertise the log-log overhead, and every gate count in Theorem 5.1 inherits an extra log(1/epsilon) factor if n_p and n_s are actually O(log(1/epsilon)).","section":"Abstract, §3.4, §5.1 proof of Theorem 5.1, Eqs. (88)–(89)"},{"comment":"The paper's abstract claims a quantum algorithm for Maxwell's equations with 'time-dependent source terms', but Assumption 4.1 restricts the source vectors at each time level s_l to take only two constant values on index subsets of total size O(m). The circuit for the source-coupling term U2, the decomposition in Eqs. (37)–(38), and the gate counts in Lemma 5.3 and Theorem 5.1 all depend on this two-valued structure. For a general smooth source J(x,t), rho(x,t), the source matrix F(s_l) would contain many independent entries, and the same circuit construction would require a different, potentially much more expensive decomposition. The paper should either state this restriction as an explicit scope limitation in the abstract and theorem statements, or provide a construction for general sources that supports the advertised claim.","section":"Section 4, Assumption 4.1 and Eqs. (37)–(38); Theorem 5.1"},{"comment":"The claimed improved complexity with a higher-order Trotter-Suzuki formula, ~O(epsilon^{-5/4} log(1/epsilon)) for d = O(1), is only sketched. Equation (91) gives an Nt scaling that mixes spatial, spectral, and source-size parameters, but no complete error analysis is given for the higher-order splitting applied to the full Hamiltonian H = HDs + HF + Hcurl, including the commutator structure and the number of terms k. Since this ~O(epsilon^{-5/4}) bound is presented as a significant further acceleration over the first-order result, it needs a derivation rather than a one-line assertion.","section":"Section 5.1.1, Eqs. (90)–(92)"}],"minor_comments":[{"comment":"The statement that 'for general r, this will lead to essentially the spectral accuracy' is imprecise and contradicts the algebraic convergence used in Lemma 5.5 and in the numerical experiments; it should be either proved or removed.","section":"Section 3.4, after Eq. (35)"},{"comment":"The piecewise definition of beta(ξ) uses |x| rather than |ξ|, and the function is only H^2-regular; calling this 'high-order' in later text should be qualified as fixed-order (r = 2) unless higher-regularity examples and their convergence orders are supplied.","section":"Eq. (34)"},{"comment":"The numerical experiments are for a one-dimensional reduced Maxwell system, not for the full three-dimensional PEC circuit or for the Assumption 4.1 source-coupling circuit; they validate the recovery and convergence mechanism but should not be read as a demonstration of the full circuit implementation.","section":"Section 6, Table 1 and Figure 8"},{"comment":"The notation ‖V − U‖ is overloaded: V and U first appear as Nt-step evolutions and then in the proof are related to one-step Trotter error via Eq. (86). Clarifying the distinction between the one-step and Nt-step operators would help the reader follow the error budget.","section":"Lemma 5.5, Eq. (83)"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern about the log-log overhead is valid and is the main obstacle to acceptance: the manuscript's own equations (88)–(89) give n_p, n_s = O(log(1/epsilon)), so the abstract and conclusions overstate the result. The paper is otherwise a solid construction paper with explicit circuits and a careful Trotter analysis, and the restrictive Assumption 4.1 is at least stated explicitly. I recommend major revision with a request to correct the complexity claims and to frame the source-term assumption as a scope limitation. The heavy self-citation is within the authors' own Schrödingerization program and is not, by itself, a problem."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the paper actually delivers something new: an explicit qubit circuit for the Schrödingerized Maxwell system with PEC boundaries and time-dependent sources, with detailed gate counts and an error analysis. Second, the advertised complexity wins don't survive the paper's own estimates. The abstract's 'almost log log 1/ε' qubit overhead for the extra dimensions is not supported by Lemma 5.5, which gives algebraic O(Δp^r + Δs^r) convergence with fixed r; equations (88)-(89) in the proof of Theorem 5.1 set n_p, n_s = O(log(1/ε)). You need exponential convergence with controlled constants to get log log, and nothing in the paper provides that. This is a load-bearing gap: every gate count in Theorem 5.1 picks up an extra log(1/ε), which weakens the claimed advantage over FDTD. And the FDTD baseline itself looks overstated: they quote O(ε^{-1-d/2}) for d=3, which appears to omit the number of time steps; standard second-order FDTD costs O(ε^{-(d+1)/2}) = O(ε^{-2}) in 3D. Against that baseline their O(ε^{-2} poly log) quantum cost is not a polynomial speedup. The higher-order Trotter version (ε^{-5/4}) depends on the same baseline and on commutator bounds I'd want to check.\n\nThat said, the construction is real work. The decomposition into U1, U2, U3, the use of Bell-basis decompositions for the curl terms, and the careful gate counting are all concrete and extend the prior circuit papers [16,26] to Maxwell with sources and boundaries. The error analysis in Lemma 5.2 and 5.5 is explicit enough to pin down where things go wrong. The numerical tests, though only 1D and only testing the Schrödingerization recovery rather than the full circuit, do confirm second-order convergence consistent with the algebraic error bounds.\n\nThe main soft spots, in order: (1) the log-log claim is unsupported and contradicted by the paper's own formulas; (2) the classical baseline is inflated; (3) Assumption 4.1 limits sources to a piecewise two-valued form on index subsets, which is more restrictive than the abstract's 'time-dependent source terms'; (4) the numerics don't exercise the constructed circuit. None of these makes the paper incoherent; they make the central complexity claims unreliable as stated.\n\nThis paper is for people working on Schrödingerization-based quantum PDE algorithms and circuit synthesis for Hamiltonian simulation. It deserves a serious referee: the construction is detailed, the errors are identifiable, and the fixes are tractable. I'd send it out with a clear request to revise the complexity claims, correct the FDTD comparison, and state Assumption 4.1 upfront.","headline":"Genuinely explicit circuit construction for Schrödingerized Maxwell, but the advertised log-log qubit overhead and the polynomial speedup over FDTD are not supported by the paper's own error bounds.","tokens_in":24720,"tokens_out":3670,"would_cite":false,"duration_ms":45118,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68","65M06","35Q61"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs an explicit qubit circuit that simulates Maxwell's equations with time-dependent sources and PEC boundaries, and proves it is polynomially faster than classical FDTD.","keywords":["Schrödingerisation","Maxwell's equations","quantum circuit construction","PEC boundary conditions","time-dependent sources","autonomization","FDTD comparison","Lie-Trotter-Suzuki splitting"],"falsifier":"Take a 3D Maxwell problem with a source that at one time level is a smooth function taking many distinct values across the whole grid, implement the paper's $U_2$ by generalizing Assumption 4.1, and check whether the CNOT count still fits Theorem 5.1's $O(|I||I_s|)$ formula and whether the Trotter error bound of Lemma 5.2 holds; if either fails, the claim that the method handles general time-dependent sources is refuted.","tokens_in":23665,"feed_emoji":"⚡","tokens_out":12792,"duration_ms":122445,"temperature":0.7,"pith_summary":"This paper closes the gap between Schrödingerisation as a black-box Hamiltonian-simulation method and a concrete circuit-level algorithm: it writes down, gate by gate, a quantum circuit that simulates Maxwell's equations in a cube with perfect electric conductor walls and time-dependent current and charge sources. The construction lifts the non-unitary Maxwell system into a Schrödinger-type system via the warped phase transformation, adds an autonomous time variable $s$ to remove time dependence, and discretizes on a Yee lattice so the resulting Hamiltonian is a sum of low-weight Pauli terms. For sources that at each sampled time take two constant values on index sets of total size $O(m)$, the paper proves explicit gate counts and argues the lifted dimensions add only about $O(\\log\\log(1/\\varepsilon))$ extra qubits. If correct, electromagnetic simulation becomes a concrete candidate for polynomial quantum speedup over the classical FDTD method rather than an abstract formalism.","feed_headline":"Explicit qubit circuit for Maxwell's equations beats FDTD","feed_subtitle":"Schrödingerization lifts Maxwell's equations into a Hamiltonian system; extra dimensions cost O(log log 1/ε) qubits.","key_machinery":"The load-bearing object is the autonomous Schrödinger-type Hamiltonian $H = I_n \\otimes P_s \\otimes I_{N_p} + \\sum_{l} (H_1(s_l) \\otimes |l\\rangle\\langle l| \\otimes D_p - H_2(s_l) \\otimes |l\\rangle\\langle l| \\otimes I_{N_p})$ obtained after Fourier-spectral discretization of the warped phase $p$ and delta-smeared initial data in the autonomous variable $s$. Its circuit-friendliness comes from the splitting $H = H_{Ds} + H_F + H_{\\mathrm{curl}}$: $H_{Ds}$ is a QFT-diagonal phase, $H_F$ is a sum of $X/Y$ rotations controlled on source and time index patterns, and $H_{\\mathrm{curl}}$ is a sum of shift operators that Lemma 4.1 rewrites with CNOTs and controlled-$R_Z$ gates. These blocks turn Schrödingerisation into a concrete gate sequence, and the complexity theorem is a count of those gates plus the Trotter steps needed for precision $\\varepsilon$.","core_discovery":"The central claim is that Schrödingerised Maxwell equations with PEC boundary conditions and time-dependent sources can be implemented by an explicit three-block circuit: $U_1$ applies the $p$-momentum evolution with a phase gate conjugated by the QFT, $U_2$ implements the source coupling with multi-controlled $R_X$/$R_Y$ rotations activated on binary-encoded index sets, and $U_3$ implements the Yee curl operator using a Bell-basis decomposition of shift matrices. Theorem 5.1 then bounds the total number of single-qubit and CNOT gates needed to prepare the field with precision $\\varepsilon$, and the comparison in Section 5.1.1 states the cost is $\\tilde{O}(\\varepsilon^{-5/4}\\log(1/\\varepsilon^{1/r}))$ with a higher-order Trotter formula, against $O(\\varepsilon^{-1-d/2})$ for classical FDTD. The paper further claims that the stretching transformation which homogenizes the source does not degrade the success probability, since $\\|u_f(0)\\|/\\|u_f(T)\\| = O(1)$ for bounded sources.","pith_inferences":["A practical reading of the 'almost $\\log\\log(1/\\varepsilon)$' claim: for a fixed smoothness order $r$, the $p$- and $s$-grid qubit counts from Equations (88)-(89) are $O(\\log(1/\\varepsilon))$; the advertised $\\log\\log$ scaling is realized only when $r$ grows with precision. This is our inference from Lemma 5.5, not stated by the paper.","If Assumption 4.1 is dropped in favor of arbitrary smooth sources approximated by $K$ two-valued patches, Theorem 5.1's complexity would gain a factor of $K$; the crossover precision at which the quantum algorithm beats FDTD would shift accordingly. This is a testable extension the paper does not discuss.","The Bell-basis shift decomposition used for the curl operator is generic for any discretized differential operator built from $S_+$ and $S_-$, so the same circuit pattern should extend to other hyperbolic systems such as elastodynamics or two-fluid plasma models, a connection the paper leaves implicit."],"forward_implications":["For sources satisfying Assumption 4.1, the paper's circuit explicitly prepares the electromagnetic field state with the stated gate counts, giving a qubit-level blueprint for Schrödingerised Maxwell simulation.","The added $p$ and $s$ dimensions cost only about $O(\\log\\log(1/\\varepsilon))$ extra qubits, so the overhead of unitarization and autonomization does not dominate at high precision.","With a second-order Trotter-Suzuki formula the total cost is $\\tilde{O}(\\varepsilon^{-5/4}\\log(1/\\varepsilon^{1/r}))$ in fixed dimension, compared with $O(\\varepsilon^{-1-d/2})$ for classical FDTD, so the quantum algorithm is polynomially faster.","The homogenization of the source via auxiliary variables keeps the success probability bounded, because $\\|u_f(0)\\|/\\|u_f(T)\\|=O(1)$ for $J,\\rho \\in L^\\infty(0,T;L^2(\\Omega))$.","The same circuit decomposition (QFT phase, multi-controlled rotations, Bell-basis shifts) gives a template for other Schrödingerised PDEs with boundary or source terms."],"supporting_citations":[{"why":"Introduces Schrödingerisation via the warped phase transformation that converts non-unitary linear dynamics into Schrödinger-type systems; the whole construction starts from this method.","marker":"[24,25]"},{"why":"Supplies the autonomization technique that adds the variable s and makes the non-autonomous Maxwell system time-independent.","marker":"[10]"},{"why":"Provides the stretching transformation that homogenizes the source-driven ODE system and the recovery formula used in Theorem 3.2.","marker":"[21]"},{"why":"Gives the Yee-lattice discretization of Maxwell's equations and its Schrödingerisation, the discretization used throughout the paper.","marker":"[22]"},{"why":"Supplies the smooth compactly supported initial data g(p) and the spectral-accuracy estimates used in Lemma 5.5 to control the p-dimension error.","marker":"[23]"},{"why":"Establishes the quantum-circuit pattern for Schrödingerised PDEs (QFT phase, multi-controlled rotations) that the explicit circuits extend to sources and boundaries.","marker":"[16]"},{"why":"Provides the Hamiltonian form in Equation (65) and the approximation lemma (Lemma 5.1) used to bound the curl-block Trotter error.","marker":"[34]"},{"why":"Higher-order Trotter-Suzuki formulas used in Section 5.1.1 to reduce the gate complexity to O(epsilon^{-5/4} log(1/epsilon^{1/r})).","marker":"[12]"},{"why":"Supplies the gate-count formula for decomposing multi-controlled RZ gates, used in Lemma 5.3 to count CNOT gates.","marker":"[47]"}],"fun_headline_variants":["Explicit Schrödingerized circuit for Maxwell's equations","Quantum circuit for Maxwell with time-dependent sources","Polynomial speedup over FDTD via Schrödingerization","Log-log qubit overhead for Maxwell's quantum simulation","Maxwell's equations on qubits with time-dependent sources"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"All circuit blocks and the complexity bounds assume that at each sampled time level the discretized source takes only two constant values, on index sets of total size $O(m)$; without this two-valued structure the source-coupling circuit $U_2$ and Theorem 5.1's gate counts do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Explicit Schrödingerized circuit for Maxwell's equations","Quantum circuit for Maxwell with time-dependent sources","Polynomial speedup over FDTD via Schrödingerization","Log-log qubit overhead for Maxwell's quantum simulation","Maxwell's equations on qubits with time-dependent sources"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000248,"raw_usage":{"total_tokens":1576,"prompt_tokens":1005,"completion_tokens":571,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":621,"completion_tokens_details":{"reasoning_tokens":494}},"tokens_in":621,"tokens_out":571,"duration_ms":6068,"temperature":1.0,"reasoning_tokens":494,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T19:04:52.006249+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a 3D Maxwell problem with a source that at one time level is a smooth function taking many distinct values across the whole grid, implement the paper's $U_2$ by generalizing Assumption 4.1, and check whether the CNOT count still fits Theorem 5.1's $O(|I||I_s|)$ formula and whether the Trotter error bound of Lemma 5.2 holds; if either fails, the claim that the method handles general time-dependent sources is refuted.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Yee-lattice discretization of Maxwell's equations and its Schrödingerisation, the discretization used throughout the paper."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Higher-order Trotter-Suzuki formulas used in Section 5.1.1 to reduce the gate complexity to O(epsilon^{-5/4} log(1/epsilon^{1/r}))."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the gate-count formula for decomposing multi-controlled RZ gates, used in Lemma 5.3 to count CNOT gates."}],"review_version":1}