{"id":"e57df461-0226-45f9-8e26-43f80986de1c","arxiv_id":"2505.02465","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Near-axis diagnostics estimate field error, effective ripple, ripple-well onset, magnetic well stability, and beta limits for quasi-isodynamic stellarators, benchmarked against VMEC and NEO.","lead":"This paper builds a toolkit to design twisted-donut shaped fusion devices, called quasi-isodynamic stellarators, using a fast near-axis model instead of slow full 3D equilibrium calculations. It adds diagnostics for field error, particle leakage, ripple wells, stability, and pressure limits, and validates them against full numerical solvers.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"ε_eff estimate at A_ref=10 is used where the paper's own Aw diagnostic predicts ripple wells in all seven benchmark configurations.","rationale":"The reader's weakest assumption correctly identifies the single-well premise of the ε_eff derivation. My stress test sharpens this into an internal inconsistency: the measure ε_edge_eff is evaluated at A_ref=10, yet Table 2's own Aw values show that ripple wells appear at larger aspect ratios for all seven benchmark configurations. Thus the near-axis ε_eff omits a trapped-particle class that exists in the very field it describes, and Figure 2 shows the resulting departure growing precisely in the A≲10 regime where ε_edge_eff is defined. This is a loading-bearing concern for the central claim that near-axis measures can replace global neoclassical calculations for design decisions. The paper has genuine independent support: the δB correlation >0.99 against 1680 VMEC cases, the machine-precision check on the Shafranov-shift gradient, and the public pyQIC code. Those support other parts of the toolkit, but not the ε_eff measure at its stated reference aspect ratio. The conditional verdict remains appropriate rather than rejection, because the broad toolkit may still be useful if A_ref is adjusted or if the ripple-well contribution is demonstrated to be small; the proposed concrete test would settle that directly.","tokens_in":38141,"tokens_out":4702,"duration_ms":58679,"concrete_test":"Take configuration 4.1 (Aw=27.4) at A=10; using the provided pyQIC code, evaluate the full Nemov ε_eff (Eq. 4.1) on the second-order near-axis field including every trapping well per field line, not just the B0-defined well. Compare to the Section 4.4 estimate and to NEO on the VMEC equilibrium. If including secondary wells changes ε_eff by more than 20%, or if the Section 4.4 estimate misses the NEO value at A=10, then ε_edge_eff is not reliable at its own reference aspect ratio.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.1 builds the near-axis ε_eff on the assumption that B0(φ) defines a single trapping well per field period, so all bounce integrals run between B0-defined bounce points. Sections 4.4 and 5 concede that secondary wells and misaligned maxima are not captured. The internal tension is quantitative: Table 2 reports Aw, the largest aspect ratio at which ripple wells first appear, for the seven benchmark configurations: 27.4, 37.0, 12.9, 32.8, 24.4, 18.5, 16.5. The paper's own headline measure ε_edge_eff is defined at A_ref=10 (Eq. 4.13), and for every benchmark configuration Aw > A_ref. Hence at the very aspect ratio used to define the measure, the asymptotic field contains ripple wells that the ε_eff calculation omits. Figure 2 shows departures growing at A≲10, which is exactly where ε_edge_eff is evaluated. The excellent agreement at larger A does not validate the measure at its stated reference point; it only validates it where the single-well assumption happens to hold. Because ε_edge_eff is proposed as a design tool replacing global transport calculations, this omission can change design rankings, especially for configurations with large Aw (e.g., 4.1 and 4.2).","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a suite of near-axis measures for quasi-isodynamic (QI) stellarators, intended to make QI design independent of global-equilibrium and neoclassical calculations. The measures are: a near-axis estimate δB_ar of the field truncation error δB, a second-order near-axis evaluation of the effective ripple ε_eff (with a leading-order buffer contribution and a second-order omnigeneity-breaking contribution), a ripple-well aspect ratio A_w, magnetic-well shaping gradients with minimal stabilising shaping, and a Shafranov-shift sensitivity/critical-β measure. The central quantitative claims are that δB_ar correlates above 0.99 with VMEC-based δB over 1680 configurations, and that the near-axis ε_eff reproduces NEO results over a large range of aspect ratios with the second-order contribution dominating. The paper also provides detailed appendices for the asymptotic expansions and shape-gradient calculations, and makes code and data openly available via Zenodo.","tokens_in":38390,"tokens_out":6692,"duration_ms":78574,"significance":"If valid, the proposed toolkit would be a significant methodological advance: it would allow QI design-space exploration and optimisation to proceed within the near-axis framework, dramatically reducing reliance on expensive global equilibrium and neoclassical evaluations. The paper is commendable in several respects: the measures are derived from near-axis theory rather than fitted to the quantities they predict; the δB_ar benchmark uses a large database of 1680 configurations; the ε_eff comparison to NEO covers multiple configurations and aspect ratios; the shape-gradient and adjoint treatments are carefully derived and checked to machine precision; and the code and data are openly available. However, the validity of ε_edge_eff at its defined reference aspect ratio is undermined by the paper's own A_w diagnostic, as detailed below, so the central claim that these measures can replace global calculations for design decisions is only partially established.","major_comments":[{"comment":"ε_edge_eff is defined at A_ref=10 in Eq. (4.13), but every benchmark configuration in Table 2 has A_w > 10 (values 27.4, 37.0, 12.9, 32.8, 24.4, 18.5, 16.5). By the definition in Eq. (5.1), A_w is the largest aspect ratio at which secondary wells appear in the asymptotic field, so at A=10 each configuration already contains ripple wells, the structures that the derivation in §4.1 explicitly assumes away and that §4.4 states the near-axis estimate cannot capture. The benchmark in Figure 2 therefore validates the ε_eff estimate only for A larger than these thresholds, not at the point A_ref=10 where ε_edge_eff is evaluated; the growing departures at A≲10 seen in Figure 2 are consistent with this. Because ε_edge_eff is proposed as a design measure replacing global neoclassical calculations, this omission can change the ranking of configurations, especially those with large A_w. Please redefine ε_edge_eff at an aspect ratio satisfying A_ref > max A_w (or otherwise ensure the single-well assumption holds), add a ripple-well correction, or explicitly restrict the measure's domain and quantify its error at A=10.","section":"§4.5, Eq. (4.13), and §5.1/Table 2"},{"comment":"The benchmark of the near-axis ε_eff uses only seven configurations, and while the agreement at large aspect ratio is good, it does not establish agreement at the reference point of ε_edge_eff. Figure 2 shows departures growing for A≲10, the regime in which ε_edge_eff is defined. The paper should report, for each of the seven configurations, the relative difference between the near-axis estimate and the NEO result at A=10, and ideally verify that the ranking by ε_edge_eff agrees with the ranking by NEO at A=10. Without this, the claim that ε_edge_eff is a useful design measure at A_ref=10 is not supported by the presented evidence.","section":"§4.4, Figure 2"}],"minor_comments":[{"comment":"The K-equation is written with two identical terms \"K4s sin 4χ + K4s sin 4χ\"; one of these should presumably be K4c cos 4χ (or a similar distinct harmonic). Please correct this typo.","section":"Appendix C, after Eq. (C8)"},{"comment":"The text refers to \"the measure ˆT and r_mhd_c\", but the table and the definitions in Section 6.1 use A_mhd_c; please unify the notation.","section":"§6.2 and Table 3"},{"comment":"\"In the vain of the near-axis expansion\" should be \"In the vein of the near-axis expansion\".","section":"Introduction"},{"comment":"The sentence beginning \"The latter is unable to capture local ripples...\" lacks a clear antecedent for \"the latter\"; please rephrase to identify the near-axis estimate explicitly.","section":"§4.4"},{"comment":"The caption states the plot is in log scale but does not state the normalization of ε_eff; please clarify that the ripple is normalized to B̄=1 T and R̄=1 m, as mentioned in the text.","section":"Figure 2 caption"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a good fit for J. Plasma Phys., and the central derivations are largely sound and carefully documented. The main issue is the conflict between the definition of ε_edge_eff at A_ref=10 and the paper's own A_w diagnostic, which shows ripple wells present at that aspect ratio in all benchmark configurations. This is fixable within the manuscript's scope by redefining the reference point, adding a correction, or clearly qualifying the measure's domain. I would not reject the paper; the toolkit is potentially valuable and the δB_ar benchmark is strong. A minor editorial concern is that the database of 1680 configurations relies on a forthcoming publication (Plunk et al. 2025), but since the data and code are available on Zenodo, this is not blocking."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this is a real contribution, and it deserves a serious referee. The one thing to flag before you read it is a mismatch between where the paper defines its headline ripple measure and where its own diagnostic says the assumptions break.\n\nWhat is actually new: the paper adds a coherent set of near-axis diagnostics that go beyond earlier work (mostly Landreman 2021 and the authors' own second-order QI construction). The field-error estimate δB_ar generalizes Landreman's construction to QI and validates beautifully against VMEC on 1680 configurations (correlation >0.99). The second-order effective ripple expansion is derived carefully in the appendices and tracks NEO well over a range of aspect ratios. The ripple-well onset measure Aw, the magnetic well shape gradient, and the Shafranov shift sensitivity are all genuinely new and useful for systematic design-space exploration. The code and data are on Zenodo, which makes the paper far more credible than most.\n\nSoft spots in proportion. The stress-test note is correct: Table 2 gives Aw values (12.9–37.0) that are all larger than the A=10 used to define ε_edge_eff in Eq. (4.13). The near-axis ε_eff calculation explicitly assumes a single trapping well per period, and the paper itself says that secondary wells and misaligned maxima are not captured. So at the reference point, the very assumptions behind the headline measure are violated. The paper is honest about this limitation, but that honesty does not resolve the tension: the measure is proposed as a design ranking tool, and at A=10 its ranking could be affected by the missing ripple-well physics. The benchmark also uses only seven configurations, all from the authors' own family; that is a small sample, though the VMEC-based δB benchmark is much stronger.\n\nWho this is for: anyone working on quasi-isodynamic stellarator design, near-axis theory, or stellarator optimisation. The paper is not for a general physics audience; it is dense but well-organized, and the appendices provide enough detail to follow the derivations.\n\nRecommendation: send it to peer review. The methods are novel, the implementation is open, and the weaknesses are identifiable and addressable (e.g., move A_ref to a safer value, or qualify ε_edge_eff with Aw). I would cite it if I did stellarator design.","headline":"A substantial near-axis toolkit for QI stellarators, with mostly solid benchmarks and one internal tension: the ε_edge_eff reference at A=10 sits where the paper's own ripple-well diagnostic says the single-well assumption fails.","tokens_in":38901,"tokens_out":2947,"would_cite":true,"duration_ms":39445,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["52.55.Hc"],"model":"deepseek-v4-flash","headline":"Second-order near-axis measures screen quasi-isodynamic stellarators without global equilibria","keywords":["quasi-isodynamic stellarators","near-axis expansion","second-order equilibrium","effective ripple","omnigeneity","magnetic well stability","Shafranov shift","stellarator optimisation"],"falsifier":"Take a family of near-axis QI configurations with deliberately flattened $B_0$ minima, construct global equilibria at $A = 6, 8, 10$, and compare the predicted $\\epsilon_{\\mathrm{eff}}^{3/2} = \\epsilon_{\\mathrm{eff}}^{3/2,(0)} + r^2 \\epsilon_{\\mathrm{eff}}^{3/2,(2)}$ and the predicted first-ripple aspect ratio $A_w$ with direct neoclassical and field-line calculations; if the effective ripple deviates by more than a factor of two while $A_w$ still predicts no well, or if a new well appears at $A > A_w$, the central claim fails.","tokens_in":37923,"feed_emoji":"🧲","tokens_out":9699,"duration_ms":110233,"temperature":0.7,"pith_summary":"This paper argues that the second-order near-axis expansion of quasi-isodynamic stellarators can serve as a complete evaluation platform: candidates can be screened for model reliability, neoclassical transport, ripple-well formation, MHD stability, and pressure sensitivity using only axis and shaping functions, without solving a global equilibrium. The authors define explicit measures, including a field-error estimate $\\delta B_{\\mathrm{ar}}$, a near-axis effective ripple $\\epsilon_{\\mathrm{eff}}^{3/2} \\approx \\epsilon_{\\mathrm{eff}}^{3/2,(0)} + r^2 \\epsilon_{\\mathrm{eff}}^{3/2,(2)}$, a ripple-well aspect ratio $A_w$, a magnetic-well shape gradient, and a Shafranov-shift $\\beta$-sensitivity gradient, and benchmark them against global equilibrium and neoclassical calculations. The payoff is practical: the near-axis construction stops being merely descriptive and becomes a fast optimisation and trade-off engine for the large quasi-isodynamic design space.","feed_headline":"Near-axis formulas predict stellarator behaviour without global solves","feed_subtitle":"A field-error estimate with above 0.99 correlation and a second-order effective ripple put transport and stability in the designer's hands.","key_machinery":"The object doing the work is the second-order near-axis inverse-coordinate equilibrium: a Taylor-Fourier expansion of the field strength $B = B_0(\\varphi) + r B_1(\\chi,\\varphi) + r^2 B_2(\\chi,\\varphi)$ in Boozer coordinates, with all flux-surface geometry expressed through the same near-axis coefficients. Every proposed measure is a functional of these coefficients: $\\delta B_{\\mathrm{ar}}$ uses the second-order fields from the finite-aspect-ratio matching construction; $\\epsilon_{\\mathrm{eff}}$ is obtained by expanding the bounce integrals and using equidistribution over the field-line label $\\alpha$; $A_w$ is the smallest radius at which $\\partial_\\varphi|_\\alpha B = 0$ and $\\partial^2_\\varphi|_\\alpha B = 0$ occur together; magnetic-well and Shafranov-shift sensitivities are computed by adjoint solutions of the same second-order linear operator that determines $X_{20}$ and $Y_{20}$. Because all evaluations reduce to one near-axis solve, the toolkit is optimisable.","core_discovery":"The central claim is that the second-order near-axis equilibrium description carries enough information to quantify the properties that matter for a quasi-isodynamic stellarator. The paper constructs and benchmarks five kinds of measures: the truncation error $\\delta B_{\\mathrm{ar}} = a^2 \\sqrt{\\frac{1}{2\\pi}\\int_0^{2\\pi}\\left[(\\tilde B^{(0)}_{20} + \\tilde B^{(2)}_0)^2 + \\frac12(\\tilde B^{(0)}_{2c})^2 + \\frac12(\\tilde B^{(0)}_{2s})^2\\right]d\\varphi}$, which predicts the true field error $\\delta B$ with correlation above 0.99 on a database of 1680 configurations; the leading and second-order effective ripple, whose sum reproduces global $1/\\nu$ transport calculations over a wide range of aspect ratio with the second-order term dominant; the aspect ratio $A_w$ at which secondary trapping wells first appear; the magnetic-well margin $W$ and its shape gradient, which yields a constructive prescription for minimally shaped, marginally stable fields; and the Shafranov-shift gradient, which gives critical-$\\beta$ estimates for pressure-driven surface touching. Together these show that second-order near-axis data, not just first-order axis shape, control neoclassical and stability behaviour, and that these behaviours can be predicted and steered before any global calculation is done.","pith_inferences":["A natural extension the authors do not pursue is to use $\\delta B_{\\mathrm{ar}}$ directly as an inexpensive objective in large parameter scans; its monotonic relation with $\\delta B$, even where it overestimates, makes it a valid ranking surrogate.","The same adjoint-gradient construction that yields the magnetic-well sensitivity can be applied to other geometry functionals, such as ballooning stability, turbulence proxies, or coil-complexity measures, turning the near-axis space into a general optimisation geometry.","The trade-off illustrated for one configuration, where omnigeneising shaping reduces ripple but worsens the critical aspect ratio $A_c$, suggests a Pareto front between neoclassical quality and compactness that the new measures make measurable; testing this across the large database would be a direct follow-up.","The structured deviations in the $\\delta B_{\\mathrm{ar}}$-versus-$\\delta B$ scatter indicate that a calibrated mapping could turn the near-axis error estimate into an even more quantitative predictor, a calibration the paper only hints at."],"forward_implications":["A designer can rank quasi-isodynamic candidates by predicted neoclassical transport using only near-axis data, because $\\epsilon_{\\mathrm{eff}}^{3/2} \\approx \\epsilon_{\\mathrm{eff}}^{3/2,(0)} + r^2 \\epsilon_{\\mathrm{eff}}^{3/2,(2)}$ tracks the global effective ripple until the aspect ratio approaches the ripple-well limit.","Magnetic-well stability can be built into a configuration at construction time: the shape gradient gives a unique minimal second-order shaping that makes $W$ vanish, so candidate fields are born marginally MHD-stable in the interchange sense.","The Shafranov-shift gradient yields an estimate of the critical plasma $\\beta$ at which flux surfaces touch ($\\beta_\\Delta$ and $\\beta_c$) before any finite-$\\beta$ equilibrium is computed, flagging pressure-sensitive designs.","The ripple-well measure $A_w$ defines the range of aspect ratios over which the near-axis description itself is trustworthy, so the other measures carry a built-in validity check.","Because all measures are scalars or gradients evaluated from one near-axis solve, they can serve as objectives or constraints in automated optimisation over axis shape, elongation, triangularity, and pressure gradient, making trade-offs like omnigeneity versus shaping cost quantitatively explorable."],"supporting_citations":[{"why":"Supplies the finite-aspect-ratio matching construction from which $\\delta B_{\\mathrm{ar}}$ and the critical radius $A_c$ are derived.","marker":"Landreman 2021"},{"why":"Provides the inverse-coordinate near-axis equilibrium equations and the second-order solve used throughout.","marker":"Landreman & Sengupta 2019"},{"why":"Introduces the direct construction of omnigenous fields near the magnetic axis and the buffer-region notion behind $\\epsilon_{\\mathrm{eff}}^{3/2,(0)}$.","marker":"Plunk et al. 2019"},{"why":"Derives the second-order omnigeneity condition $\\Delta B_{2c}^{\\mathrm{QI}}$ that drives the ripple expression.","marker":"Rodríguez & Plunk 2023"},{"why":"Supplies the second-order QI construction, the parallel-adiabatic-invariant expansions, and the benchmark configurations used in Sections 4-7.","marker":"Rodriguez et al. 2024"},{"why":"Defines the effective ripple $\\epsilon_{\\mathrm{eff}}$ that the paper expands asymptotically in $r$.","marker":"Nemov et al. 1999b"},{"why":"Gives the near-axis magnetic-well formula $V''$ used for the stability measures.","marker":"Landreman & Jorge 2020"},{"why":"Previously applied the field-error goodness criterion to first-order QI fields and motivates its second-order generalisation.","marker":"Camacho Mata et al. 2022"},{"why":"Provides examples of the half-helicity axis family from which some benchmark configurations are drawn.","marker":"Plunk et al. 2024"}],"fun_headline_variants":["Second-order near-axis terms dictate stellarator trade-offs","Measure stellarator transport and stability from near-axis data","Near-axis toolkit predicts stellarator behaviour before global solves","Second-order expansion captures stellarator confinement and stability"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The effective-ripple calculation assumes that $B_0(\\varphi)$ defines exactly one trapping well per field period and that local ripple wells and misaligned field maxima stay negligible at the aspect ratios of interest; if those finite-aspect-ratio effects become significant, the near-axis $\\epsilon_{\\mathrm{eff}}$ estimate and the measures built on it lose their predictive power.","fun_headline_variants_meta":{"raw":{"variants":["Second-order near-axis terms dictate stellarator trade-offs","Measure stellarator transport and stability from near-axis data","Near-axis toolkit predicts stellarator behaviour before global solves","Second-order expansion captures stellarator confinement and stability"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000215,"raw_usage":{"total_tokens":1441,"prompt_tokens":970,"completion_tokens":471,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":586,"completion_tokens_details":{"reasoning_tokens":409}},"tokens_in":586,"tokens_out":471,"duration_ms":5831,"temperature":1.0,"reasoning_tokens":409,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:50:16.473187+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a family of near-axis QI configurations with deliberately flattened $B_0$ minima, construct global equilibria at $A = 6, 8, 10$, and compare the predicted $\\epsilon_{\\mathrm{eff}}^{3/2} = \\epsilon_{\\mathrm{eff}}^{3/2,(0)} + r^2 \\epsilon_{\\mathrm{eff}}^{3/2,(2)}$ and the predicted first-ripple aspect ratio $A_w$ with direct neoclassical and field-line calculations; if the effective ripple deviates by more than a factor of two while $A_w$ still predicts no well, or if a new well appears at $A > A_w$, the central claim fails.","supporting_citations":[{"cited_title":"2021 Figures of merit for stellarators near the magnetic axis","cited_arxiv_id":null,"evidence_quote":"Supplies the finite-aspect-ratio matching construction from which $\\delta B_{\\mathrm{ar}}$ and the critical radius $A_c$ are derived."},{"cited_title":"& Sengupta, W","cited_arxiv_id":null,"evidence_quote":"Provides the inverse-coordinate near-axis equilibrium equations and the second-order solve used throughout."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the direct construction of omnigenous fields near the magnetic axis and the buffer-region notion behind $\\epsilon_{\\mathrm{eff}}^{3/2,(0)}$."},{"cited_title":"Journal of Plasma Physics 89 (6), 905890609","cited_arxiv_id":null,"evidence_quote":"Derives the second-order omnigeneity condition $\\Delta B_{2c}^{\\mathrm{QI}}$ that drives the ripple expression."},{"cited_title":", Plunk, G","cited_arxiv_id":null,"evidence_quote":"Previously applied the field-error goodness criterion to first-order QI fields and motivates its second-order generalisation."}],"review_version":1}