Pith. sign in

REVIEW 3 major objections 12 minor 43 references

A novel discontinuous-Galerkin deterministic neutronics model for Fusion applications: development and benchmarking

T0 review · 3 major / 12 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A deterministic transport method combining arbitrary-order discontinuous Galerkin spatial discretization, discrete ordinates, multigroup energy groups, and matrix-free iterative solvers delivers fusion blanket neutron fluxes within 10% of…

desk verdict Solid methods paper with honest benchmarks; the abstract overclaims the 3D fusion capability, which is only demonstrated on a slab with rectangular cutouts and no timing comparison, but the core derivation and verification are worth a referee. read the letter →

arxiv 2411.16369 v2 pith:GKR5MJWR submitted 2024-11-25 physics.comp-ph

Timo Jos Bogaarts , Felix Warmer This is my paper · ORCID
classification physics.comp-ph
keywords discontinuousGalerkindiscreteordinatesmultigroupneutrontransportfusionblankettritiumbreedingratiomatrix-freeiterativesolversunstructuredmesh
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper introduces a deterministic neutron transport solver intended to fill the gap between slow, high-fidelity Monte Carlo simulations and fast but drastically reduced models used in early fusion design. It solves the six-dimensional transport equation on unstructured meshes using arbitrary-order discontinuous Galerkin spatial discretization, discrete-ordinates angular discretization, multigroup energy discretization, arbitrary-order anisotropic scattering, and matrix-free iterative solvers. The claim is that this combination makes three-dimensional neutron-response calculations fast enough for iterative design while staying close to Monte Carlo accuracy, demonstrated in a breeding blanket benchmark with scalar flux within 10% for $E>10$ eV and tritium breeding within 1.5% of a continuous-energy Monte Carlo reference (1% with problem-specific cross sections). Verification against analytic solutions and literature scattering tests shows optimal spatial convergence orders in one, two, and three dimensions. A sympathetic reader would take the central claim to be that this method provides a practical deterministic alternative for neutronic design assessment in fusion.

What carries the argument

The load-bearing machinery is the discontinuous Galerkin spatial discretization with an upwind numerical flux, which makes the spatial sweep a topological ordering of a directed acyclic graph: each element's solution depends only on its inflow neighbours, so elements are solved one by one with element-local matrices of size $p\times p$ instead of one global system. The same structure is then accelerated by a simplex simplification in which the volume, flux, and edge matrices are evaluated once on a reference element, so each element solve reduces to matrix additions and matrix-vector products. Scattering-source convergence is handled by matrix-free iterative solvers acting on spherical-harmonic moments of the angular flux, whose dimension is far smaller than the number of discrete angles, and the no-upscatter assumption makes the multigroup coupling lower-triangular so energy groups are solved sequentially from the highest to the lowest energy.

What would settle it

Run the method on a well-moderated problem where most of the scalar flux lies below $1$ eV at material temperatures near $1000$ K, and compare the sub-eV flux and a neutronically sensitive response (such as an absorption rate in a detector nuclide) against a continuous-energy Monte Carlo reference; if the disagreement exceeds the 10% margin established in the blanket benchmark, the central accuracy claim is limited to energies where upscatter is negligible.

Watch

Extended reading notes

Core claim

The paper's central claim is that a particular combination of existing discretizations produces a neutron transport scheme that is simultaneously arbitrary-order, unstructured, and fast enough for design-loop use: discontinuous Galerkin in space with an upwind numerical flux, discrete ordinates in angle, a multigroup energy structure solved from high to low energy, arbitrary-order anisotropic scattering, and matrix-free Krylov iterative solvers for the within-group scattering source. The authors derive the single-element transport sweep on straight-sided convex elements, simplify it for simplexes so that element matrices are precomputed and no integrals are needed during sweeps, and then verify the implementation against an analytic solution, isotropic and anisotropic scattering tests, and a fusion-relevant breeding blanket benchmark in both one- and three-dimensional geometries. The blanket test shows scalar flux agreement within 10% for energies above $10$ eV, tritium breeding 1.5% lower than the Monte Carlo reference with standard cross sections and 1% lower with problem-specific cross sections, and qualitative agreement across five orders of magnitude of flux and twelve orders of magnitude of energy spectrum.

Load-bearing premise

The load-bearing premise is that neutrons never gain energy in scattering events, so group-to-group upscatter is neglected; the paper acknowledges this breaks down below about $10$ eV, and the low-energy blanket-edge flux is exactly where benchmark agreement degrades.

Editorial extensions

If this is right

  • Blanket scalar fluxes above $10$ eV and tritium breeding ratios can be obtained from a deterministic solve in one, two, and three dimensions without running millions of Monte Carlo histories.
  • Because the mesh is unstructured and the sweep is element-local, the solver can be applied to curved, component-level, or simplified full-reactor geometries and coupled to other engineering design codes.
  • Problem-specific multigroup cross sections generated once from a Monte Carlo run improve agreement significantly and can be reused for nearby design variants, which is useful in design optimization.
  • The full angular flux distribution is available from the deterministic solve rather than only tally-based results, so high-fluence or neutron-penetration regions can be identified directly during design.
  • The matrix-free formulation keeps memory use low because the full transport matrix is never formed; only the action of the operator on a moment vector is needed for GMRES- or BiCGSTAB-style iterations.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct but unstated consequence is that removing the no-upscatter assumption with outer iterations over the lowest energy groups would likely shrink the observed $E<10$ eV edge discrepancies, since the paper's own comparison attributes part of that gap to neglected group-to-group upscatter.
  • Not stated in the paper but a natural next test: the method's speed advantage should be largest in strongly three-dimensional, curved geometries such as stellarator blankets, where Monte Carlo sampling and CAD faceting are hardest, so a direct head-to-head wall-clock comparison in such a geometry would sharpen the design-cycle claim.
  • The paper lists differentiable solutions only as a future possibility, but because the element matrices and sweep order are fixed and cheap, a shape-optimization loop that differentiates tritium breeding with respect to blanket geometry appears to be a plausible extension of the same machinery.
  • An untested but direct generalization is to include upscatter by iterating the energy-group loop rather than solving it in a single downward pass; the paper describes this cost but does not quantify it, so a benchmark with a thermal-energy design quantity would show whether the trade-off is acceptable.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 12 minor

Summary. The manuscript presents a deterministic neutron transport solver aimed at early-stage fusion design. The method combines multigroup energy discretization (solved from high to low energy under the no-group-to-group-upscatter assumption), discrete-ordinates angular discretization with TN quadratures, arbitrary-order discontinuous Galerkin spatial discretization on straight-edged convex elements, Legendre-expanded anisotropic scattering, and matrix-free GMRES/BiCGSTAB inner iterations on spherical-harmonic flux moments; the upwind DG flux yields a per-element transport sweep whose matrix assembly and inversion are element-local (Eq. 22), with a quadrature-free simplification for simplexes (Appendix C). Verification proceeds in three stages: an analytical slab problem showing optimal error scaling N^-(p+1) in space for p = 0-4 and Q^-1 in angle; the Reed isotropic-scattering test and a two-group anisotropic-scattering test, which agree with the multigroup mode of OpenMC to within about 2% in well-resolved regions; and a helium-cooled-pebble-bed blanket benchmark against continuous-energy OpenMC using FENDL/TRANSX and OpenMC-generated multigroup cross sections, agreeing within 10% for E > 10 eV and within 1.5% on tritium breeding, with larger deviations below 10 eV traced to the upscatter assumption. A single three-dimensional benchmark with rectangular void cutouts in the blanket slab is also reported, using the OpenMC-generated cross sections.

Significance. If the results hold, this is a genuinely useful tool for the fusion design toolbox: a deterministic solver with demonstrated high-order spatial convergence, a memory-efficient moment-only formulation (Eq. 33), honest and quantified treatment of the upscatter approximation (Fig. 13), and validation against independent references (analytical solution; OpenMC in both multigroup and continuous-energy modes) with cross sections taken from FENDL and from OpenMC rather than fitted to benchmark outputs; there are no free parameters in the derivations. The paper's strengths include explicit assumption statements (Section 2.1), the element-independent calligraphic-matrix construction that makes high-order simplex sweeps quadrature-free (Appendix C), and the clean separation of verification (Section 3) from application (Section 4).

major comments (3)
  1. [Abstract; §4.4; §5] The abstract claims a method 'capable of quickly and accurately assessing the neutron response of a fusion reactor, even in three-dimensional geometry,' but the reported evidence supports only part of this claim. No wall-clock, CPU-hour, or iteration-count figures appear anywhere in the manuscript; the only quantitative efficiency statement is the qualitative remark in §4.4 that the deterministic code 'use[s] significantly less time CPU-hours in all considered cases,' and the conclusions (§5) explicitly defer benchmarking 'in a more complex, three-dimensional reactor geometry, including computational speed comparisons' to a future paper. Moreover, the only fully three-dimensional inhomogeneous benchmark (§4.4) is a slab with two axis-aligned rectangular void beams, using cross sections generated from the one-dimensional problem, so it does not exercise curved or toroidal geometry, and the mesh is effectively Cartesian rather than genuinely unstructured. Section 2.8 further notes that only angle-parallelization has been used and that scaling is future work. The fix is within scope: either temper the abstract and §1 to claim demonstrated accuracy on slab-like geometries with speed presented as a qualitative observation, or add a minimal quantitative performance comparison (e.g., a CPU-time or iteration-count table for the Section 4 blanket case). As written, the headline claim is ahead of the evidence.
  2. [§2.1; §4.3; Fig. 13] The abstract's 'accurately assessing' claim is unqualified, but the method deliberately neglects group-to-group upscatter (§2.1), and the authors' own measurements show the consequences: Fig. 13 demonstrates that the upscatter assumption 'partly explains the observed differences' for E < 10 eV, where Figs. 11 and 16 show deterministic/OpenMC disagreements at the blanket edge that exceed the 10% figure quoted for E > 10 eV. The paper is admirably transparent about this limitation in the body text, so the issue is claim scope rather than a hidden flaw: the abstract should state the demonstrated domain of validity, namely energies above roughly 10 eV and integral quantities such as tritium breeding (within 1.5% of OpenMC) in the tested slab-like geometries. Without this qualification, the abstract overstates the method's accuracy in the thermal range.
  3. [§4.4] The three-dimensional benchmark uses only the OpenMC-generated multigroup cross sections from the one-dimensional configuration ('Only OpenMC-generated cross sections (from the 1D problem) are used'), and no three-dimensional run with the FENDL/TRANSX cross sections is reported. This is not circular in the strict sense, because the reference solution is continuous-energy OpenMC and the cross sections are inputs rather than fits, but it does reduce the independence of the 3D validation: the multigroup data embedded in the deterministic solve were produced by the same code against which the solution is compared, on the parent configuration of the modified 3D case, while the FENDL path that a design tool would actually use is validated only in 1D. Either a single FENDL-based 3D run (even with the expected lower agreement in the breeding zone) or an explicit statement that §4.4 is a cross-section-transferability demonstration rather than a validation of the FENDL chain would make the scope of the 3D claim precise.
minor comments (12)
  1. [Abstract] The abstract as circulated contains the duplicated phrase 'capable of quickly and quickly assessing,' whereas the abstract inside the manuscript reads 'capable of quickly and accurately assessing'; the duplication should be removed and the two versions made identical.
  2. [Throughout] Several typos should be corrected: 'benificial' (§1), 'this is comes' (§2.1), 'quadarature' (§3.2), and 'the mimum feature size' (Fig. 7 caption).
  3. [§3.1; §3.3; Fig. 10] The convergence-order notation is inconsistent: §3.1 states 'optimal N^{p+1} convergence rate' and §3.3/Fig. 10 state 'N^{-p+1},' while the plotted fits in Figs. 4 and 10 are N^-1, N^-2, N^-3, N^-4, N^-5 for p = 0-4, i.e., error proportional to N^-(p+1); the text and captions should be unified on this form.
  4. [§2.3, Eq. (6); Appendix A, Eq. (A.3)] The Legendre expansion of the scattering cross section is written in Eq. (6) without the (2l+1) factor but in Eq. (A.3) with it; the footnote in Appendix A explains the convention, but a forward pointer in §2.3 would prevent readers from misassembling Eq. (7).
  5. [§2.4] The statement that topologically sorting the inflow-dependency graph 'yields a solution order' should be qualified: cyclic sweep dependencies are a known phenomenon for arbitrary unstructured S_N meshes (see Ref. [10]), so the authors should state the conditions under which the graph is guaranteed to be acyclic or describe the fallback (e.g., iteration) used when it is not.
  6. [§4.4] The sentence 'use significantly less time CPU-hours in all considered cases' should specify the conditions of the comparison (serial versus parallel settings, machine, numbers of cores, and whether the OpenMC tallies were converged) so that the reader can at least reproduce the setup of the qualitative speed statement.
  7. [§4.4] No convergence check is reported in the three-dimensional geometry (the TN9 quadrature set is chosen 'to limit ray-effects' and the mesh is refined once); a single refinement or quadrature-order test in 3D would show that the residual differences in Fig. 16 are not discretization-driven.
  8. [§2.8] The paper states that one-, two-, and three-dimensional models are implemented, but no two-dimensional benchmark is presented; a sentence clarifying whether the 2D models are exercised only as special cases of the 3D runs would close this gap.
  9. [Figs. 6 and 9] The vertical axis of Fig. 6 is labeled phi [m^-3], whereas comparable flux plots elsewhere (e.g., Fig. 9) use phi [m^-2 s^-1]; the units should be harmonized and checked.
  10. [§2.6] The operator M that converts angle values to moment values shares its symbol with the element M of §2.4 and with the moment space M defined in the same paragraph; renaming the operator (for example, Phi) would remove a genuine source of confusion.
  11. [References] References [19] (Lyytinen et al., Nuclear Fusion, 2024) and [21] (Boyd et al., Nuclear Technology, 2019) lack complete volume/page or article identifiers.
  12. [Code availability] The manuscript does not state whether the code or the benchmark input files will be made available; a data/code availability statement would materially aid reproducibility of the verification results.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the numerical derivation is self-contained and benchmarks are external; the few self-citations are motivational only.

full rationale

The derivation chain in Sections 2.2-2.7 and Appendices A-C is self-contained: the discontinuous-Galerkin upwind discretization (Eqs. 13-22), the topological sweep ordering (Eqs. 27-32), and the matrix-free operator form (Eq. 33, Appendix B) are derived from the transport equation without assuming the target results. Verification uses an analytical slab solution (Eq. 42) and two external reference codes: OpenMC's multigroup mode for the isotropic and anisotropic scattering tests, and continuous-energy OpenMC for the breeding-blanket comparison. Cross sections come from the external FENDL library or from OpenMC tallies; the OpenMC-generated multigroup data are presented transparently as problem-specific data preparation (Section 4.2), while the FENDL case provides an independent check (Section 4.3). The no-upscatter assumption is a physical simplification, not a fitted input, and its thermal-range breakdown is explicitly tested in Figure 13. The only self-citations ([19], [38]) support motivational statements about stellarator design and blanket geometry, not the correctness of the method, and no uniqueness theorem or ansatz is imported from the authors' prior work. The paper's overclaims about speed and 3D reactor capability (no wall-clock comparison; only a slab-with-cutout 3D test; explicit deferral in Section 5) are correctness and scope risks, not circularity.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The method introduces no fitted physical constants and no new physical entities. The central claim rests on standard transport discretization plus domain assumptions about upscatter, mesh shape, material homogeneity, Legendre truncation, quadrature, and multigroup cross sections.

assumptions (6)
  • domain assumption Neutrons in fusion applications do not gain energy from scattering events; group-to-group upscatter is neglected.
    Section 2.1 first bullet. The authors acknowledge breakdown in the thermal range, and Section 4.3 shows it partly explains E<10 eV discrepancies.
  • domain assumption Elements in the spatial discretization have straight edges and are convex.
    Section 2.1 second bullet. This makes the dot product of direction with outward normal constant on each edge, enabling the sweep; mesh refinement can mitigate the restriction.
  • domain assumption Each element consists of a single material so cross sections can be moved outside finite-element integrals.
    Section 2.1 third bullet. Standard for material-defined meshes.
  • domain assumption The scattering cross section can be represented by a finite Legendre expansion of order L.
    Equation (6) in Section 2.3. Truncation is an approximation; the blanket case uses P5 anisotropy.
  • standard math Angular moments of the flux can be approximated by numerical quadrature over a finite set of discrete ordinates.
    Equation (9) in Section 2.3. Quadrature convergence is verified in Section 3.
  • domain assumption Multigroup cross sections from FENDL and from OpenMC adequately represent the continuous-energy data needed for the method.
    Section 4.2. The accuracy of the deterministic result depends on these input data, and the authors attribute part of the observed difference to the multigroup approximation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A novel discontinuous-Galerkin deterministic neutronics model for Fusion applications: development and benchmarking." pith.science (2026). https://pith.science/paper/GKR5MJWR

@misc{pith2026241116369,
  author       = {Pith},
  title        = {Pith review of: A novel discontinuous-Galerkin deterministic neutronics model for Fusion applications: development and benchmarking},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GKR5MJWR}},
  note         = {Machine review of arXiv:2411.16369}
}
read the original abstract

Neutron interactions in a fusion power plant play a pivotal role in determining critical design parameters such as coil-plasma distance and breeding blanket composition. Fast predictive neutronic capabilities are therefore crucial for an efficient design process. For this purpose, we have developed a new deterministic neutronics method, capable of quickly and quickly assessing the neutron response of a fusion reactor, even in three-dimensional geometry. It uses a novel combination of arbitrary-order discontinuous Galerkin spatial discretization, discrete-ordinates angular and multigroup energy discretizations, arbitrary-order anisotropic scattering, and matrix-free iterative solvers, allowing for fast and accurate solutions. One, two, and three-dimensional models are implemented. Cross sections can be obtained from standard databases or from Monte-Carlo simulations. Benchmarks and literature tests were performed, concluding with a successful blanket simulation.

Figures

Figures reproduced from arXiv: 2411.16369 by the authors.

Figure 1
Figure 1. A numbered triangular mesh, together with a solution direction Ω. Boundaries corresponding to outflow boundaries (defined as Ωˆ · nˆ > 0) are indicated by a red plus, while inflow boundaries (Ωˆ · nˆ < 0) are indicated by a green minus. right-hand-side does not depend on other elements (no ψ O l term) and can be solved immediately. For element 2, the only inflow boundary is connected to element 1, such that it can b… view at source ↗
Figure 3
Figure 3. The total algorithm for solving the neutron transport equation (1) using the multigroup, discrete-ordinates, Discontinuous Galerkin, discretization. Red dashed boxes indicate sequential loops, while blue dotted boxes denote parallel loops. For each energy group Eg, the downscatter source is calculated, after which for all angles Ωˆ k (in parallel), the spatial PDE of equation (13) is solved. This is done by (potenti… view at source ↗
Figure 4
Figure 4. figure 4. For the spatial convergence, all order show the [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: The L2 error for the numerical solution of equation (41) with the tetrahedral finite elements as a function of the number of finite element blocks at Ωx = 0.24 (a) The order p of the finite element is indicated in the plot. All orders show the optimal Np+1 convergence …
Figure 5
Figure 5. Figure 5: The domain setup for the isotropic scattering test. σt is the total cross section in m−1 , σs the isotropic scattering cross section in m−1 , and q the isotropic source in m−3 . Data from Ref. [35]. The solutions of both the deterministic code (20 element blocks per cm…
Figure 7
Figure 7. Figure 7: L2 difference of the flux ϕ between OpenMC and the deterministic code for the isotropic scattering test as a function of the number of angles for the different regions (a) and as a function of the number of elements per cm (b). For the quadrature convergence, 20 third-…
Figure 8
Figure 8. Figure 8: Geometry for the anisotropic multigroup scattering test of references [11, 36, 37]. The precise scattering data and matrices are given in Appendix D. The shaded area denotes an isotropic source with strength [9.7702 · 104 , 4.5451 · 105 ] m−3 s−1 respectively for the t…
Figure 9
Figure 9. Figure 9: (a) Results for deterministic simulation (solid lines) and an OpenMC simulation (dashed lines) for both energy groups (E0 red dots, E1 blue triangles) for the anisotropic scattering test on the domain of figure 8. Black dashed lines indicate the region boundaries of fi…
Figure 10
Figure 10. Figure 10: L2 difference of the flux ϕ between OpenMC and the deterministic code as a function of the number of angles Q (a) and as a function of the number of elements per 10 cm N. Fits are shown with dashed lines. For the quadrature convergence, 52 second order elements per 10…
Figure 11
Figure 11. Figure 11: Scalar flux as a function of position in the breeding blanket as defined in tables 2 and 1 for the OpenMC simulation (solid lines), and the deterministic code with FENDL cross sections (dotted lines) and the OpenMC generated cross sections (dashed lines). The scalar f…
Figure 12
Figure 12. Figure 12: Scalar flux spectrum integrated over the regions in the breeding blanket as defined in tables 1, 2 and for the OpenMC simulation (solid lines), and the deterministic code with FENDL cross sections (dotted lines) and the OpenMC generated cross sections (dashed lines). …
Figure 13
Figure 13. Figure 13: Scalar flux in the energy range E < 10 eV where upscatter is significant for the OpenMC simulation (cyan solid line, shown also in figure 11), an OpenMC simulation where group-to-group upscatter was manually removed (blue dotted line) and the deterministic simulation …
Figure 15
Figure 15. Figure 15: Scalar flux for E > 1 MeV averaged over the cutout in the z-dimension for the geometry of figure 14. Figure (a) shows the OpenMC results, and figure (b) the deterministic results with the cross sections from the 1D OpenMC simulation. The cutout in the x, y dimensions …
Figure 16
Figure 16. Figure 16: Scalar flux in the geometry of figure 14 as a function of x, averaged over the cutout in the z-dimension (−0.1 < z < 0.10) and over the region below the cutout ((a), y < −0.2), over the cutout itself ((b), −0.2 < y < 0) and over the region above the cutout ((c), y > 0…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

  1. [1]

    Lion J, Warmer F, Wang H, Beidler C, Muldrew S and Wolf R 2021 Nuclear Fusion 61 126021

    work page 2021

  2. [2]

    2022 Fusion Engineering and Design 178 113080

    Kembleton R, Morris J, Siccinio M, Maviglia F et al. 2022 Fusion Engineering and Design 178 113080

    work page 2022

  3. [3]

    Franza F, Boccaccini L, Fisher U, Gade P and Heller R 2015 Fusion Engineering and Design 98 1767–1770

    work page 2015

  4. [4]

    Moreno C, Bader A and Wilson P 2024 Frontiers in Nuclear Engineering 3 1384788

    work page 2024

  5. [5]

    2023 Fusion Science and Technology 79 1008–1022

    Valentine A, Berry T, Bradnam S, Chohan H, Eade T, Grove C, Hagues J, Hearn K, Hodson J, Lennon K et al. 2023 Fusion Science and Technology 79 1008–1022

    work page 2023

  6. [6]

    2024 Energy 289 129970

    Palermo I, Alguacil J, Catal´ an J P, Fern´ andez-Berceruelo I, Lion J, Valiente J ´A N, Sosa D, Rapisarda D, Urgorri F R, Warmer F et al. 2024 Energy 289 129970

    work page 2024

  7. [7]

    Kovari M, Fox F, Harrington C, Kembleton R, Knight P, Lux H and Morris J 2016 Fusion Engineering and Design 104 9–20

    work page 2016

  8. [8]

    Lewis E E and Miller W F 1984 Computational methods of neutron transport (United States: John Wiley and Sons, Inc)

    work page 1984

Show all 43 references
  1. [9]

    Evans T M, Stafford A S, Slaybaugh R N and Clarno K T 2010 Nuclear technology 171 171–200

    2010

  2. [10]

    Wareing T A, McGhee J M, Morel J E and Pautz S D 2001 Nuclear science and engineering 138 256–268

    2001

  3. [11]

    Goldberg E, i Duxans M C, Gelabert O O, Mantsinen M and Soba A 2022 Plasma Physics and Controlled Fusion 64 104006

    2022

  4. [12]

    Moller J Y and Lautard J J 2011 Minaret, a deterministic neutron transport solver for nuclear core calculations Proc. Int. Conf. Mathematics and Computational Methods Applied to Nuclear Science and Engineering (Rio de Janeiro, Brazil)

    2011

  5. [13]

    Fournier D, Herbin R and Tellier R L 2013 SIAM Journal on Scientific Computing 35 A936–A956

    2013

  6. [14]

    Hall E, Houston P and Murphy S 2017 SIAM Journal on Scientific Computing 39 B916–B942

    2017

  7. [15]

    Royston K E, Johnson S R, Evans T M, Mosher S W, Naish J and Kos B 2018 Fusion science and technology 74 303– 314

    2018

  8. [16]

    Youssef M, Batistoni P, Patrizzi L, Wareing T and Davis I 2007 Fusion science and technology 52 801–806

    2007

  9. [17]

    H¨ außler A, Fischer U and Warmer F 2017 Fusion Engineering and Design 124 1207–1210

    2017

  10. [18]

    H¨ außler A, Warmer F and Fischer U 2018 Fusion Engineering and Design 136 345–349

    2018

  11. [19]

    Lyytinen T T, Snicker A, Virtanen J, Palermo I, Alguacil J, Bogaarts T J and Warmer F 2024 Nuclear Fusion

    2024

  12. [20]

    Davison B, Sykes J B and Cohen E R 1958 Neutron transport theory (American Institute of Physics)

    1958

  13. [21]

    Boyd W, Nelson A, Romano P K, Shaner S, Forget B and Smith K 2019 Nuclear Technology

    2019

  14. [22]

    Hesthaven J S and Warburton T 2007 Nodal discontinu- ous Galerkin methods: algorithms, analysis, and appli- cations (Springer Science & Business Media)

    2007

  15. [23]

    Aho A V and Hopcroft J E 1974 The design and analysis of computer algorithms (Pearson Education India)

    1974

  16. [24]

    Joseph D, El Hafi M, Fournier R and Cuenot B 2005 International journal of thermal sciences 44 851–864

    2005

  17. [25]

    Patton B W and Holloway J P 2002 Annals of Nuclear Energy 29 109–136

    2002

  18. [26]

    Saad Y 2003 Iterative methods for sparse linear systems (SIAM)

    2003

  19. [27]

    Saad Y and Schultz M H 1986 SIAM Journal on scientific and statistical computing 7 856–869

    1986

  20. [28]

    Van der Vorst H A 1992 SIAM Journal on scientific and Statistical Computing 13 631–644

    1992

  21. [29]

    Ern A and Guermond J L 2004 Theory and practice of finite elements vol 159 (Springer)

    2004

  22. [30]

    2010 Eigen v3 http://eigen.tuxfamily.org

    Guennebaud G, Jacob B et al. 2010 Eigen v3 http://eigen.tuxfamily.org

    2010

  23. [31]

    Jakob W 2022 nanobind: tiny and efficient c++/python bindings https://github.com/wjakob/nanobind

    2022

  24. [32]

    2024 Nuclear Data Sheets 193 1–78

    Schnabel G, Aldama D, Bohm T, Fischer U, Kunieda S, Trkov A, Konno C, Capote R, Koning A, Breidokaite S et al. 2024 Nuclear Data Sheets 193 1–78

    2024

  25. [33]

    Romano P K, Horelik N E, Herman B R, Nelson A G, Forget B and Smith K 2015 Annals of Nuclear Energy 82 90–97

    2015

  26. [34]

    Thurgood C P, Pollard A and Becker H A 1995 Journal of Heat Transfer 117 1068–1070 ISSN 0022-1481

    1995

  27. [35]

    Reed W H 1971 Nuclear Science and Engineering 46 309– 314

    1971

  28. [36]

    1978 Journal of Nuclear Science and Technology 15 56–71

    Asaoka T, Asano N, Nakamura H, Mizuta H, Chichiwa H, Ohnishi T, Miyasaka S I, Zukeran A, Tsutsui T, Fujimura T et al. 1978 Journal of Nuclear Science and Technology 15 56–71

    1978

  29. [37]

    Quah C 1981 Annals of Nuclear Energy 8 709–715

    1981

  30. [38]

    2017 Fusion Engineering and Design 123 47–53

    Warmer F, Bykov V, Drevlak M, H¨ außler A, Fischer Deterministic Neutronics for Fusion 18 U, Stange T, Beidler C, Wolf R et al. 2017 Fusion Engineering and Design 123 47–53

    2017

  31. [39]

    Lyytinen T 2023 Parametric studies of breeding blanket thickness of a HELIAS stellarator using Monte Carlo neutron transport code Serpent2 Master’s thesis Univer- sity of Jyv¨ askyl¨ a

    2023

  32. [40]

    Martin N, Hebert A and Miss J 2011 The discrete angle technique combined with the subgroup monte carlo method Proc. Int. Conf. Mathematics and Computational Methods Applied to Nuclear Science and Engineering (Rio de Janeiro, Brazil)

    2011

  33. [41]

    Sartori E 1985 JEF/DOC-100

    1985

  34. [42]

    1992 TRANSX 2: a code for interfacing MATXS cross-section libraries to nuclear transport codes (Citeseer)

    MacFarlane R E et al. 1992 TRANSX 2: a code for interfacing MATXS cross-section libraries to nuclear transport codes (Citeseer)

    1992

  35. [43]

    Derivation of anisotropic scattering source Although this is a standard practice, the derivation using the same notation and normalization is given here for completeness

    Fleming M, Morgan L and Shwageraus E 2016 Nuclear Science and Engineering 183 173–184 Appendix A. Derivation of anisotropic scattering source Although this is a standard practice, the derivation using the same notation and normalization is given here for completeness. See [8] ...

    2016

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.