{"id":"e11e4444-fb1d-4a90-bf4a-bed32e3cb7d2","arxiv_id":"2411.19310","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For a 1+1 dimensional Vlasov model with Krook collisions, a Carleman linearization quantum algorithm has polynomially worse complexity than classical finite difference methods and requires unphysically large collision rates to provably converge.","lead":"This paper maps a discretized nonlinear Vlasov equation onto a quantum algorithm and shows the quantum approach needs polynomially more resources than standard classical codes. It also finds the algorithm only provably converges when collisions are far stronger than real plasmas have.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The paper's negative conclusions depend on treating R<1 as necessary; its own Sec. 9 caveat (citing Ref. [34] convergence at R≈44) shows this is the least secure condition, and relaxing it could eliminate both the Nv restrictions and the polynomial slowdown.","rationale":"The derivation is careful and self-consistent: the finite-difference discretization, the matrix maps, and the norm computations are explicit, and the authors flag the R<1 caveat in Section 9. I do not find an internal algebraic inconsistency in Eqs. (60)-(64) or (71)-(74). But the paper's own caveat identifies the single point on which the negative conclusions hinge. R<1 is a sufficient condition in Krovi's theorem, and the paper supplies no evidence that it is necessary for this system. The numerical result in Ref. [34] (R≈44 for Burgers) is a direct warning that the sufficient bound can be extremely loose. If R>1 convergence extends to the Vlasov two-stream system, then the ν0≥O(N_v^{3/2}) substitution used in Section 7 and Appendix D is not required. In that case γ need not approach ||u_in||, ∥ubar_in∥ can decay with N_v, NC can be logarithmic in N, and the source of the polynomial slowdown—polynomial NC combined with linear sparsity—can disappear. The paper's conclusion that the quantum route is polynomially slower would then rest on an unnecessarily restrictive sufficient condition. This is exactly the reader's weakest assumption; my read agrees. A targeted numerical Carleman-truncation study on the discretized Vlasov equations with R>1 would settle it. Consequently the verdict remains conditional: the quantitative negative claims should not be taken as established until this test is run or a necessity proof for R<1 is supplied.","tokens_in":32897,"tokens_out":10519,"duration_ms":92896,"concrete_test":"Run a numerical Carleman-linearization truncation study on the discretized Vlasov ODEs, Eqs. (11a-11d), with the two-stream initial condition Eq. (59), choosing parameters so that the R of Eq. (60) is approximately 10-50 while ν0 is kept at O(N_v) rather than O(N_v^{3/2}). Compute the Carleman error ∥z1(T)-u(T)∥ for NC=2,3,4,...; if the error decays exponentially with NC for several N_v and the required NC grows only logarithmically with N_v, then R<1 is not a necessary condition and the paper's Section 5.3 restrictions and Section 7 complexity comparison must be revisited.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the quantum algorithm is polynomially slower than classical and needs unphysical collision rates. Both conclusions are derived under the R<1 convergence criterion of Eq. (16), which forces ν0≥O(N_v^{3/2}) (Eq. 60). That criterion is a sufficient condition in Krovi's theorem, not a necessary one. The paper's own Section 9 cites Ref. [34], where Carleman linearization of Burgers' equation converges with R≈44, and 'anticipates' the same for Vlasov. If R>1 solutions converge, the bound ν0≥O(N_v^{3/2}) is not binding. Then the asymptotic analysis of Section D.3 changes: with ν0~N_v (still satisfying the discriminant condition of Eq. 20), γ~√(N_x N_v), ∥ubar_in∥→0, and NC becomes logarithmic in N instead of polynomial; the mechanism that makes the QLSA polylog dimension fail disappears, and Eq. (73) need not be polynomially worse than Eq. (74). Thus both the 'severe restrictions' (Eqs. 62-64) and the polynomial slowdown rest on an unproven necessity of R<1.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper maps the one-dimensional electrostatic Vlasov equation with a Krook collision operator, discretized on an Nx×Nv grid, onto the Carleman-linearization-based quantum ODE solver of Krovi. It constructs the explicit matrices F^(2), F^(1), F^(0), derives bounds for the convergence parameter R, the Carleman truncation level N_C, the norm of the linearized evolution matrix, and the resulting query and gate complexities. It concludes that the convergence criterion R<1 forces a collision frequency ν0≥O(Nv^{3/2}), which leads to physically unrealistic restrictions on the velocity-grid size for warm interstellar and inertial-confinement-fusion parameters, and that the quantum algorithm's complexity is polynomially worse than the classical finite-difference solver. It also argues that coupling via Ampère's law makes dissipativity impossible because the electric-field block of F^(1) has zero columns.","tokens_in":33183,"tokens_out":8054,"duration_ms":72864,"significance":"If the conclusions were fully established, the paper would provide a valuable negative case study for quantum simulation of nonlinear kinetic plasma physics: it would show that current Carleman-QLSA methods, when applied to the Vlasov-Poisson system, require unphysically strong dissipation and do not offer an asymptotic speedup. The explicit construction of the F matrices, the norm estimates in Appendix C, and the block-structure analysis of F^(2) are substantive and reusable. The Ampère-law obstruction (zero columns in F^(1) leading to α(F^(1))≥0) is a clean structural argument. However, the paper's central negative conclusions rest on treating the R<1 condition of Eq. (16) as the operative convergence criterion, while Section 9 itself states that this condition is not strict and cites a numerical example with R≈44 that still converges. Because R<1 is a sufficient condition in Krovi's theorem, not a necessary one, both the severe-grid-restriction claim and the polynomial-slowdown claim are not as robust as the abstract and conclusion suggest.","major_comments":[{"comment":"The paper's two headline conclusions—the unphysical grid restrictions of Eqs. (62–64) and the polynomial slowdown of Eq. (73) versus Eq. (74)—are derived by treating R<1 from Eq. (16) as a binding constraint. This yields ν0≥O(Nv^{3/2}) in Eq. (60), which is then substituted into the complexity analysis in Section 7 and Appendix D.7. However, Section 9 explicitly acknowledges that the R<1 requirement is not strict in practice, citing Ref. [34], where Carleman linearization of Burgers' equation converges with R≈44, and anticipates that the same holds for Vlasov. Since R<1 is a sufficient condition in Krovi's theorem rather than a necessary one, the constraint ν0≥O(Nv^{3/2}) is not proven to be required for actual convergence. For example, with ν0~Nv the discriminant condition of Eq. (20) can still be satisfied, N_C becomes logarithmic, and the asymptotic slowdown of Eq. (73) need not follow. The authors should either provide evidence—analytical or numerical for the Vlasov system specifically—that R<1 is necessary, or substantially weaken the abstract and concluding claims so that they are explicitly conditional on the proven sufficient criterion.","section":"§5.3, §7, §9, Eq. (16)"},{"comment":"There is an inconsistency in the asymptotic norm of the two-beam initial condition. Eq. (115) gives ∥uin∥=N√Nx/(√2 xmax Δv); using Δv=2vmax/(Nv−1) this scales as O(Nv√Nx), not as O(√(NxNv)) as stated in Eq. (116). This is not merely typographical: Appendix D.2 uses Eq. (116) to obtain gu=O(Nv^{1/2}), whereas with the printed O(√(NxNv)) the ratio gu would be O(1), and the subsequent estimates of N_C and ∥A∥ in Appendices D.3–D.4 would differ. Please correct Eq. (116) and trace all downstream uses of this scaling.","section":"Eq. (116) and Appendix D.2"},{"comment":"The abstract and conclusion state flatly that the convergence criteria place severe restrictions on applications and that the quantum algorithm is polynomially less efficient. Section 9 says the opposite in a qualified way, namely that the R<1 restriction is not strict and that the plasma-parameter restrictions could be relaxed. This is a substantive tension rather than a presentation issue: a reader cannot tell whether the paper's intended contribution is a rigorous bound under Krovi's sufficient condition or a general statement about the infeasibility of Carleman-QLSA for Vlasov. Please make the conditional nature of the negative result explicit in the abstract and in Section 10, and clearly separate the rigorously proven statements from the anticipated ones.","section":"Section 9 and Abstract"}],"minor_comments":[{"comment":"The symbol N is used for two different quantities: the total number of grid points NxNv in Eq. (8) and the number of electrons per unit area in Eq. (12) and later in Eq. (115). This ambiguity is confusing in Appendix C; please introduce a distinct symbol, e.g., Np for the particle number.","section":"Notation, Eqs. (8) and (12)"},{"comment":"The caption of Figure 3 contains a run-on sentence and a typographical issue in the sentence beginning 'The small cells are row vectors with length N=12'; please rewrite the caption so that the block structure and the meaning of the colored cells are self-contained.","section":"Figure 3 caption"},{"comment":"The derivation of N_C in Eq. (122) relies on a Taylor expansion of log(1/∥u¯in∥) with respect to second-order corrections. The presentation would be clearer if the precise small parameter and the order at which terms are discarded were stated explicitly before Eq. (121).","section":"Appendix D.3"},{"comment":"The bound ∥A∥≤NC(δ)(∥F^(0)∥+∥F^(1)∥+∥F^(2)∥) in Eq. (45) is quite loose, and the subsequent complexity expressions inherit this looseness. Please state explicitly that all complexity claims are upper bounds based on this worst-case bound, rather than tight estimates.","section":"Section 7 and Eq. (45)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a serious and mostly careful analysis of a known quantum algorithm applied to a physically relevant nonlinear PDE. The main concern is that the authors' own Section 9 admits the key constraint R<1 is not necessary, which undermines the strength of the negative conclusions as presently worded. I would encourage the editor to send the paper back for a revision that either (a) provides evidence for the necessity of R<1 for the Vlasov system, or (b) reframes the paper as an analysis of the proven sufficient conditions and removes the unqualified claims from the abstract. The corrected Eq. (116) scaling should also be checked carefully, since it affects Appendix D."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nThe paper is worth a careful reading if you care about whether Carleman-linearization QLSAs can touch the nonlinear Vlasov equation. It does the field a service by being explicit: first concrete mapping of the nonlinear Vlasov equation with Krook collisions onto Krovi's Carleman framework, with exact F(2), F(1), F(0) entries in Appendix B and norm derivations in C-D. The asymptotic bounds in Eqs. 63–64 reproduce from Eq. 62; no fitted parameters anywhere. The Ampere's-law negative result is structural: E does not appear in the linear part of the evolution, so F(1) has at least Nx zero eigenvalues, forcing µ(F(1)) ≥ 0. That conclusion does not depend on the R<1 debate.\n\nThe soft spot is the R<1 criterion. It is sufficient in Krovi's theorem, not necessary. The paper's own Sec. 9 admits this, citing Ref. [34] where Carleman linearization converged at R≈44 for Burgers. The consequences are larger than the authors state. If R>1 works for Vlasov, ν0 does not need to scale as Nv^{3/2}; the Nv<1 restrictions in Eqs. 63–64 disappear, and the stress-test reasoning suggests the polynomial slowdown in Eqs. 71–73 may also weaken substantially, since NC could become logarithmic rather than polynomial. The paper presents the complexity comparison in Sec. 7 as unconditional when it is conditional on the very criterion its own Sec. 9 undercuts. That is the one place where I want the authors pushed: either state clearly that the complexity results assume R<1, or analyze the R>1 scenario.\n\nMinor: Eq. (116) says ||uin|| = O(sqrt(Nx Nv)), but Eq. (115) with dv = 2 vmax/(Nv−1) gives O(Nv sqrt(Nx)). The later scaling arguments use the correct form, so it's a typo.\n\nWho is this for? Quantum-computing skeptics and plasma simulators will both find useful, honest boundaries. It is not the last word—it rules out one linearization route, not all quantum approaches, and the classical comparison is to a simple finite-difference solver, which the authors state. But as a first explicit mapping with transparent errors and a clear negative result, it deserves a serious referee. I would send it out with a request to fix the typo and to engage directly with the R>1 caveat in the complexity section, then accept.","headline":"Careful, honest negative result for Carleman QLSAs on nonlinear Vlasov, with a load-bearing R<1 caveat that the authors themselves flag.","tokens_in":33746,"tokens_out":6089,"would_cite":true,"duration_ms":51872,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["52.65.-y","03.67.Ac"],"model":"deepseek-v4-flash","headline":"Quantum Vlasov algorithm needs impossible velocity grids","keywords":["Vlasov equation","Carleman linearization","quantum linear systems algorithm","nonlinear differential equations","plasma simulation","convergence criteria","computational complexity"],"falsifier":"Solve the Carleman-truncated system for the two-beam Vlasov–Gauss equations on a small but physical grid, e.g., N_v = 64, with warm interstellar medium parameters (T = 8000 K, x_max = 1000 km) and a physical collision frequency, so that R is orders of magnitude above 1; if the truncation error still decreases exponentially with N_C, the paper's central restriction fails. Alternatively, implement the QLSA on this small grid and compare the actual gate count against the classical solver, since the complexity bounds in Eqs. (71-72) are upper bounds and a concrete implementation could show whether the polynomial gap is realized.","tokens_in":32707,"feed_emoji":"⚛️","tokens_out":8163,"duration_ms":64245,"temperature":0.7,"pith_summary":"This paper asks whether a recently proposed quantum algorithm—one that linearizes quadratic nonlinearities by Carleman embedding and solves the resulting linear system with a quantum linear solver—can simulate the nonlinear Vlasov equation on a grid. The authors map the electrostatic Vlasov equation with a Krook-type collision operator, in one space and one velocity dimension, into the exact input format the algorithm requires, and then evaluate the algorithm's own convergence and complexity criteria. They find that the algorithm's convergence parameter R can be below 1 only if the velocity grid has at most a fraction of a point for typical interstellar or inertial-confinement-fusion parameters, i.e., no physically meaningful grid satisfies the requirement. They also find that query and gate complexity upper bounds are polynomially worse than the classical finite-difference time complexity for the same scheme. The conclusion a sympathetic reader would draw is that, for this Carleman-linearization-based quantum approach, the nonlinear Vlasov equation is not a practical target.","feed_headline":"Quantum Vlasov algorithm needs impossible velocity grids","feed_subtitle":"For interstellar and fusion parameters, the allowed grid shrinks to less than one velocity point, so the quantum route collapses.","key_machinery":"The machinery is a chain: a finite-difference discretization on an $N_x \\times N_v$ grid; a row-major vectorization $u = \\mathrm{vec}(f)$; explicit matrices $F^{(2)}$ (quadratic, built from a trapezoidal-rule integration vector and a finite-difference velocity derivative stencil), $F^{(1)}$ (linear advection, background term, and diagonal Krook collision with log-norm $\\mu(F^{(1)}) \\le -\\nu_0$), and $F^{(0)}$ (Maxwellian source); the Carleman linearization that embeds the quadratic ODE into a block-tridiagonal linear system $\\frac{dz}{dt} = A z + b$ truncated at level $N_C$; and the convergence parameter $R$ of Eq. (16) together with the log-norm dissipativity condition $\\mu(F^{(1)}) < 0$. The argument's force comes from computing the asymptotic scaling of $\\|F^{(2)}\\|$, $\\|F^{(0)}\\|$, and $\\|u_{\\mathrm{in}}\\|$ and inserting them into $R$, producing the $N_v^{3/2}/\\nu_0$ scaling and the resulting physical grid-size restrictions.","core_discovery":"The paper establishes that when the discretized Vlasov–Gauss system is brought to the form $\\frac{du}{dt} = F^{(2)} u^{\\otimes 2} + F^{(1)} u + F^{(0)}$ and fed into the Carleman-linearization quantum ODE solver, the convergence criterion $R < 1$ (Eq. 16) becomes the binding constraint. With a two-beam initial condition and a physical Coulomb collision frequency, the asymptotic convergence parameter is $R = O(N_v^{3/2}/\\nu_0)$ (Eq. 60). Requiring $R < 1$ and inserting typical warm interstellar medium parameters gives $N_v \\lesssim 1.6 \\times 10^{-9}$, and inertial confinement fusion parameters give $N_v \\lesssim 2.24 \\times 10^{-5}$; conversely, a modest grid $N_v \\ge 100$ forces $x_{\\max} T \\lesssim 5.31 \\times 10^{-7}\\,\\mathrm{m\\,K}$. The query and gate complexity bounds (Eqs. 71–72) are polynomially larger than the classical time complexity $O(T^2/\\varepsilon_c)$ of the same finite-difference scheme. Coupling to Ampere's law instead of Gauss's law is shown to be worse: the field variables never enter the linear part of the evolution, so the log-norm of $F^{(1)}$ cannot be negative, violating the dissipativity condition outright.","pith_inferences":["Because the explicit $F$ matrices and their sparsity and norm scalings are derived independently of the specific quantum linear solver, the mapping itself is reusable: a future quantum nonlinear solver with milder convergence requirements or better dimension and sparsity scaling could inherit the mapping without modification.","The $R<1$ criterion used here is known from numerical experiments on a toy system (Burger's equation) to be conservative; if that finding carries over to Vlasov, the strict grid restriction is a property of the error-analysis bound rather than of Carleman linearization itself, and the practicality conclusion would need revisiting.","A direct test would be to run the truncated Carleman system classically for a small Vlasov grid with $R>1$ and check whether the truncation error decays exponentially in $N_C$; such a result would decouple the Carleman embedding from the QLSA convergence analysis.","The analysis assumes a single-species electron plasma in one dimension; the 3+3 dimensional Vlasov–Maxwell system, with magnetic fields and an explicitly evolved field, has a different linear structure, so the dissipativity failure found for Ampere's law may not persist."],"forward_implications":["For a Carleman-linearization quantum linear solver of the type analyzed here, the nonlinear Vlasov equation is not a practical quantum computing target: the convergence condition cannot be met with physically reasonable grids.","Even in the parameter regime where the quantum algorithm converges, its query and gate complexity upper bounds are polynomially larger than the classical time complexity, so no asymptotic quantum speedup is obtained from this route.","Coupling the Vlasov equation to Ampere's law instead of Gauss's law makes the linear part non-dissipative ($\\mu(F^{(1)}) \\ge 0$), so the Carleman-linearization algorithm cannot converge at all in that formulation.","The restriction acts like a CFL-type condition $R<1$ connecting the velocity grid size $N_v$, temperature $T$, and box size $x_{\\max}$: any practical grid requires either unphysically large collision rates or unphysically cold and small systems.","A relaxed convergence condition, as suggested by numerical evidence in a toy model, would substantially ease these restrictions; the paper explicitly anticipates that such a relaxation could carry over to the Vlasov equation."],"supporting_citations":[{"why":"Supplies the quantum linear ODE solver (Taylor-series time integration plus QLSA) and the theorems for error bounds, Carleman truncation level, condition number, and query and gate complexity that the paper evaluates.","marker":"[31]"},{"why":"Provides the convergence criteria (dissipativity and R<1), the rescaling choice for gamma, the state preparation for the linearized system, and numerical evidence that Carleman linearization can converge for R≈44 in a toy model.","marker":"[34]"},{"why":"Provides the explicit error bounds for Carleman linearization that justify the R<1 convergence condition used throughout the paper.","marker":"[18]"},{"why":"Establishes an analogous R formula and complexity analysis for the Carleman linearization of the Navier-Stokes equations, the framework the paper follows for the Vlasov case.","marker":"[23]"},{"why":"Supplies the plasma-physics textbook model for the Coulomb collision frequency and the typical inertial-confinement-fusion parameters used in the numerical restriction (Eqs. 61-65).","marker":"[9]"},{"why":"Supplies the warm interstellar medium temperature (T = 8000 K) used in the WIM numerical example.","marker":"[17]"},{"why":"Supplies the trapezoidal-rule discretization and the stability-condition viewpoint invoked in the discussion of the CFL-like restriction; it also underlies the classical finite-difference error analysis.","marker":"[42]"}],"fun_headline_variants":["Quantum Vlasov grid needs N_v < 10^-9","Vlasov solver: convergence forces grid below one point","Carleman Vlasov crushed by dissipation demands","Quantum Vlasov loses to classical on complexity","Plasma parameters kill quantum Vlasov grid"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the convergence criterion R<1 is necessary for useful accuracy: the paper's grid restriction comes from imposing Eq. (16), and if Carleman linearization still converges for R much larger than 1—as numerical experiments on a toy model suggest—the severe grid restriction would be substantially relaxed; a secondary premise is that the two-beam initial condition is representative, with the authors arguing other initial conditions only increase R.","fun_headline_variants_meta":{"raw":{"variants":["Quantum Vlasov grid needs N_v < 10^-9","Vlasov solver: convergence forces grid below one point","Carleman Vlasov crushed by dissipation demands","Quantum Vlasov loses to classical on complexity","Plasma parameters kill quantum Vlasov grid"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00023,"raw_usage":{"total_tokens":1508,"prompt_tokens":995,"completion_tokens":513,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":611,"completion_tokens_details":{"reasoning_tokens":435}},"tokens_in":611,"tokens_out":513,"duration_ms":5125,"temperature":1.0,"reasoning_tokens":435,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:19:55.012991+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Solve the Carleman-truncated system for the two-beam Vlasov–Gauss equations on a small but physical grid, e.g., N_v = 64, with warm interstellar medium parameters (T = 8000 K, x_max = 1000 km) and a physical collision frequency, so that R is orders of magnitude above 1; if the truncation error still decreases exponentially with N_C, the paper's central restriction fails. Alternatively, implement the QLSA on this small grid and compare the actual gate count against the classical solver, since the complexity bounds in Eqs. (71-72) are upper bounds and a concrete implementation could show whether the polynomial gap is realized.","supporting_citations":[{"cited_title":"Explicit Error Bounds for Carleman Linearization","cited_arxiv_id":null,"evidence_quote":"Provides the explicit error bounds for Carleman linearization that justify the R<1 convergence condition used throughout the paper."},{"cited_title":"Quantum Carleman linearisation efficiency in nonlinear fluid dynamics","cited_arxiv_id":"2410.23057","evidence_quote":"Establishes an analogous R formula and complexity analysis for the Carleman linearization of the Navier-Stokes equations, the framework the paper follows for the Vlasov case."}],"review_version":1}