Numerical analysis of the homogeneous Landau equation: approximation, error estimates and simulation

Francis Filbet (IMT); Ling-Bing He (THU); Yanzhi Gui (THU)

REVIEW 1 major objections 4 minor 52 references

Numerical analysis of the homogeneous Landau equation: approximation, error estimates and simulation

T0 review · 1 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read A Fourier spectral method for the Landau–Coulomb equation is proven to converge with explicit error bounds: for any fixed time interval and any tolerance, sufficiently large domain and mode count bring the numerical solution within toleranc

desk verdict The paper aims at the first explicit convergence bound for a spectral discretization of the Landau-Coulomb equation, but the nonnegativity proof in Section 3.3 relies on a false identity, so the main theorem is not established as written. read the letter →

arxiv 2509.09276 v1 pith:QHVXSS7A submitted 2025-09-11 math.AP cs.NAmath.NA

Francis Filbet (IMT) , Yanzhi Gui (THU) , Ling-Bing He (THU) This is my paper · ORCID

classification math.APcs.NAmath.NA MSC 65M1582C4035B6535R0965M7065N35

keywords LandauequationCoulombpotentialFourier-Galerkinspectralmethoderrorestimateconvergenceanalysiskinetictheorycollisionoperator

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

This paper proves convergence with explicit error bounds for a Fourier–Galerkin spectral method solving the spatially homogeneous Landau equation with Coulomb potential. The numerical solution is built in two steps: the infinite velocity space is truncated to a box of half-size $L$, and the periodized collision operator is discretized with $N$ Fourier modes per direction. The authors show that the $L^2$ distance between the numerical solution and the exact solution on a fixed time interval $[0,T]$ is bounded by a term decaying like $L^{-l}$ plus a term decaying like $(L/N)^{n-2}$, up to exponential-in-time factors. Hence for any tolerance, sufficiently large $L$ and $N$ bring the error below it. This gives the Coulomb case a rigorous error analysis that the literature had only provided for harder potentials.

What carries the argument

The argument is carried by three pieces. (1) A coercivity/anisotropic-energy machinery for the Coulomb collision operator: with a weighted Sobolev norm and an angular-derivative norm $H^1_A$, the linearized collision operator gives a negative definite contribution, while commutator estimates control the remainder. (2) The periodic Landau operator $Q^\#$, whose Fourier symbol is computed exactly in Proposition 4.1: $Q^\#(e^{i\pi \ell \cdot v/L}, e^{i\pi m \cdot v/L}) = \beta(\ell, m) e^{i\pi (\ell + m) \cdot v/L}$ with $\beta(\ell, m)$ an explicit closed-form expression in $\ell$ and $m$. This preserves the convolution structure and allows the spectral method to be analyzed mode-by-mode. (3) A two-level error decomposition: the truncation residual produces th

What would settle it

Run the scheme for a smooth admissible initial datum, and at a fixed time compare the $L^2$ error for $(L,N)$ pairs that double $L$ and $N$ according to Theorem 1.3 against a very high-resolution reference solution: the error should drop roughly like $\max(L^{-l}, (L/N)^{n-2})$ with the proven rates; a slower rate would falsify the spectral projection or truncation estimates. A separate check: if one exhibits an admissible initial datum whose Landau-Coulomb solution develops a singularity or a growing $H^{n+2}_{k+l}$ norm in finite time, Proposition 1.1, and with it Theorem 1.3, is false.

Watch

Extended reading notes

Core claim

The central result is Theorem 1.3. Under the paper's assumption that the initial datum lies in a weighted Sobolev space with sufficiently high regularity and moment decay, the exact solution is globally regular with a uniform bound $M$. For any time $T>0$, once the velocity box half-size $L$ and the number of Fourier modes $N$ satisfy the explicit conditions $L \ge 2\tilde R(T)$ and $N/L \ge \tilde N(T,L/2)$, the $L^2(\mathbb{R}^3)$ error between the numerical solution $f^{R,N}$ and the exact solution $f$ obeys $\| f^{R,N}(t) - f(t) \|_{L^2} \leq C \left[ e^{\kappa_0 t}/L^\ell + (L/N)^{n-2} e^{\kappa L^{1/2} t} \right]$ for all $t \in [0,T]$, with constants depending only on $M,n,k,l$. Because both terms can be made arbitrarily small by increasing $L$ and $N$, the sche

Load-bearing premise

The load-bearing premise is the uniform-in-time weighted Sobolev bound $\| f \|_{L^\infty([0,\infty); H^{n+2}_{k+l})} < M$ for the exact solution (Proposition 1.1), imported from prior analytic results: if the Landau-Coulomb solution can lose this regularity or the bound is not finite, the constants in Theorems 1.1–1.3 do not exist and the convergence guarantee collapses.

Editorial extensions

If this is right

For any fixed time interval and any error tolerance, the spectral method with sufficiently large L and N is guaranteed to approximate the Landau-Coulomb solution within that tolerance in L2.
The truncated equation preserves nonnegativity of the distribution function for R large enough, so the physical meaning of the numerical solution is retained at the truncation level.
The error rate in the number of modes is spectral in the ratio L/N: for solutions with H^n regularity, the discretization error scales like (L/N)^{n-2}, so smoother solutions give faster convergence.
The explicit Fourier symbol β(l,m) supplies a rigorous justification of the fast spectral algorithms used in practice, which evaluate the collision operator through convolutions and FFT.
The numerical experiments for the Coulomb case show that the scheme reproduces the expected decay of entropy, relative entropy, and Fisher information, and that moderate N (e.g., 48 per direction) already matches a reference solution on long time intervals.

Reading between the lines

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

The exponential factor exp(κ L^{1/2} t) in the spectral error suggests that for very long time horizons the number of modes must grow rapidly with the domain size; in practice one might need adaptive or time-dependent truncation to keep the provable bound useful.
The error estimate is in L2; because the scheme also propagates high Sobolev regularity, an L∞ or pointwise error bound should be derivable by Sobolev embedding, though the paper does not state one.
The explicit β(l,m) formula, independent of the rest of the analysis, could be reused to build conservative or entropy-stable spectral discretizations of related kinetic equations, or to analyze time-discretization errors in the Fourier basis.
The proof relies on the global-in-time regularity bound M imported from analytic theory; if future work weakens that hypothesis, the same error structure would carry over, making the numerical analysis conditional only on the regularity of the exact solution.

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.

Desk Editor's Note

read the letter

Quick take: the paper targets a real gap—there is no previous error estimate with explicit rates for Fourier spectral methods applied to the Landau-Coulomb case. The two-level truncation (ψ_R inside and outside the collision operator) is a sensible device, and the final error structure (R^{-l} + (R/N)^{n-2}) matches what one expects. The exact formula for the periodic collision multiplier β(l,m) in Proposition 4.1 is a genuinely nice piece of algebra.

But there is a load-bearing flaw. Section 3.3 asserts the identity

∫ Q(h,h) h 1_{h<0} dv = -∫ (a*h):∇h⊗∇h 1_{h<0} dv + 4π∫ h³ 1_{h<0} dv,

with h = f^R ψ_R. That is not the weak form. The correct expression contains the double integral ∫∫ a(v-v′) ∇h(v′) · ∇h(v) h(v) 1_{h(v)<0} plus surface terms living on the zero set of h; it does not collapse to a local cubic term. Try h = -c on a large ball and zero outside: Q(h,h) is identically zero on the ball, so the left side is zero, while the cubic term is a large negative number. The identity fails. The entire nonnegativity proof of f^R depends on this identity, and Theorem 1.2 later uses f^R ≥ 0 to make the leading term I111 nonpositive. I see no other argument for nonnegativity in the paper, so the convergence guarantee (1.15) is unsupported as written. This is not a typo; it is the central estimate.

Elsewhere, the quality is uneven but mostly real. The analytic machinery is sophisticated, and the imported regularity bound M from the authors' arXiv preprint [32] is a genuine validation risk—if that preprint does not survive review, the constants are not grounded—but the use of the Fisher information monotonicity from [27] is legitimate. The numerics are illustrative: no code or data are provided, so they do not independently verify the theorem. The introduction's promise that the scheme 'preserves non-negativity' is an overstatement; the discrete solution does not, and the authors concede this later in the text.

My bottom line: the paper's goal is important and several components are solid, but its headline result rests on a false computation. Send it to a serious referee by all means; the authors should be asked to replace the nonnegativity argument or to state the main theorem conditionally. I would not cite it as a proof until that is fixed.

Referee Report

1 major / 4 minor

Summary. The paper analyzes a Fourier-Galerkin spectral method for the spatially homogeneous Landau equation with Coulomb potential. The strategy is a two-step approximation: first, the equation is truncated in velocity with a smooth cutoff ψ_R, and an O(R^{-l}) error estimate between the truncated and exact solutions is derived (Theorem 1.1). The truncated solution is then periodized on a torus and discretized by N Fourier modes per direction, yielding an O((R/N)^{n-2} exp(κ R^{1/2} T)) spectral error (Theorem 1.2). Combining the two gives the main convergence statement (Theorem 1.3): for any fixed T and tolerance, sufficiently large L and N make the L2 error arbitrarily small. The proofs are based on weighted Sobolev energy estimates, commutator inequalities, a nonnegativity argument for the truncated solution, and exact Fourier symbol computations for the periodic Landau operator. Numerical experiments for Maxwellian and Coulombian interactions are included.

Significance. If the analysis is correct, this is a significant contribution: it gives the first explicit, parameter-free a priori convergence rates for a spectral method applied to the full Landau-Coulomb equation, with the expected error structure R^{-l} + (L/N)^{n-2}. The proof is largely self-contained at the numerical-analysis level and carefully tracks all constants. The main external input is a global uniform Sobolev bound M imported from the analytic theory, in particular from the recent preprint [32] by He, Ji and Luo. This dependency is heavy and should be clearly stated, but it is not by itself a defect.

major comments (1)

[§2.3 (proof of Proposition 1.1, final paragraph)] The proof states: 'Finally by applying Theorem A.2-(1) with r = 1/2, m = ... we obtain that (2.19) holds.' However Theorem A.2-(1) is stated only for r ∈ [-1/2, 0]. The desired membership (2.19) is exactly the r = -1/2 case of that theorem, giving C([0,∞); H^{-1/2}_m) ∩ L^2_loc([0,∞); H^{1/2}_{m-3/2}). Since (2.19) is used as the bootstrap hypothesis for the interval-induction yielding the uniform H^{n+2}_{k+l} bound, this is a load-bearing step. Please correct the value of r and verify the hypotheses (in particular f0 ∈ L^1_{2m+1}) under Assumption 1.1.

minor comments (4)

[§3.3 (nonnegativity identity)] The identity ∫ Q(h,h) h 1_{h<0} dv = -∫ (a*h):∇h⊗∇h 1_{h<0} dv + 4π∫ h^3 1_{h<0} dv is correct for smooth h: the second term follows from ∇·(a*∇h) = -8πh and integration by parts with g = h^2 1_{h<0}/2. The stress-test counterexample with h a genuine step function is outside the admissible class, since products of surface distributions are not defined. Still, the paper's 'By further computation' is too terse; a short derivation should be added, especially because the negative-part function f^R 1_{f^R<0} is nonstandard (it is nonpositive rather than the usual positive negative part).
[Lemma 3.1] The condition R > 3 is not enough to guarantee ―f ψ_R―_{L1} > 1/2 from the energy bound ∫|v|^2 f = 3; Chebyshev gives mass outside |v|>R/2 ≤ 12/R^2, so one needs R > √24. Since the theorem takes R large anyway, this is a minor quantitative slip, but the threshold should be corrected or replaced by 'R sufficiently large'.
[Throughout] Typographical issues: 'employee' should be 'employ' (§1.2), 'functionnal' → 'functional', 'moldecules' → 'molecules' (§1.3 and §5), 'W e' → 'We' (§4.2). Also, the remark after Theorem 1.2 that the estimate requires N > L is dimensionally odd (L has length units); the condition should be phrased as N sufficiently large relative to R, as in the theorem statement.
[Appendix A / Theorem A.2] The statement of Theorem A.2-(1) should be checked against [32] and its hypotheses listed with the range of r used later. In particular, the proof of Proposition 1.1 additionally invokes Theorem A.2-(2) and A.2-(5); making the exact dependencies explicit would help the reader, since all later constants depend on M.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the convergence proof derives truncation and spectral errors from independent analytic regularity inputs; the flagged nonnegativity identity is a correctness gap, not a circular reduction.

full rationale

The claimed convergence chain is Theorem 1.3 = Theorem 1.1 + Theorem 1.2. Theorem 1.1 controls the truncation error via an energy estimate on g = f^R - f, with the residual bounded by R^{-l} ||f||^2_{H^2_{k+l}}; Theorem 1.2 controls the spectral error via the projection estimate (4.2), the explicit Fourier symbol bound (4.5), and a Gronwall argument. No step fits a parameter to the target error or defines the target quantity in terms of itself. The constants M, C, and kappa are inputs inherited from the analytic regularity Proposition 1.1, which is cited from [32] (He-Ji-Luo) and [8] (Carrapatoso-Desvillettes-He). This is heavy same-author reliance, but it is not circular: Theorem A.2 states global well-posedness and smoothing for the Landau equation with no reference to the numerical scheme, so the convergence conclusion is not presupposed. Assumption (1.5) is a sufficient condition on the initial data, not a reformulation of the error bound. One non-circular proof gap is flagged: Section 3.3 uses the identity int Q(h,h) h 1_{h<0} dv = -int (a*h): grad h ⊗ grad h 1_{h<0} dv + 4π int h^3 1_{h<0} dv, which is algebraically false in general; for h constant negative on a ball, the left side is zero while the 4π term is not. Since the nonnegativity of f^R enters the proof of I111 in Theorem 1.2, the proof as written is incomplete. This is a correctness problem, not a circularity reduction, so the circularity score is 0.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper's central claim rests on the analytic well-posedness and regularity theory of the Landau-Coulomb equation, imported from cited literature (partly the authors' own), plus the regularity assumptions on the initial data. No parameters are fitted to data; all constants are generic. The only 'invented' objects are the truncated/periodic collision operators Q# and Q_N^R, which are mathematical constructs, not new physical entities.

assumptions (5)

standard math Theorem A.2, items (1)-(5): global well-posedness, smoothing estimates, propagation of regularity, Fisher information monotonicity, and moment decay for the homogeneous Landau-Coulomb equation, from [32], [27], [8].
Quoted in Section 2.3 to prove Proposition 1.1 (the global regularity bound M). The bound M feeds every error estimate in Theorems 1.1-1.3.
standard math Theorem A.1: sharp dissipativity lower bound for (a*g): grad f tensor grad f, from [32, Prop 2.1].
Used in Lemma 2.3 and Section 3.3 for coercivity and nonnegativity, essential for Theorem 1.1.
standard math Monotonicity of Fisher information for Landau-Coulomb solutions (part of Theorem A.2 item 4, from [27]).
Used in Section 2.3 to get uniform H^{-1/2} bounds for f, a key step in proving the global bound M.
domain assumption Assumption 1.1: f0 in L1_ell intersected H^{n+2}_{k+l} with k>9/2, l>3, n>=5 and inequality (1.5).
Defines admissible initial data; all constants in the error estimates depend on the implied bound M.
standard math Theorem B.1, convolution inequality from [7, Lemma 3.3].
Used repeatedly in Lemmas 2.1 and 2.2 to bound weighted convolution integrals of the collision operator.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Numerical analysis of the homogeneous Landau equation: approximation, error estimates and simulation." pith.science (2026). https://pith.science/paper/QHVXSS7A

@misc{pith2026250909276,
  author       = {Pith},
  title        = {Pith review of: Numerical analysis of the homogeneous Landau equation: approximation, error estimates and simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QHVXSS7A}},
  note         = {Machine review of arXiv:2509.09276}
}

abstract

We construct a numerical solution to the spatially homogeneous Landau equation with Coulomb potential on a domain $D_L$ with N retained Fourier modes. By deriving an explicit error estimate in terms of $L$ and $N$, we demonstrate that for any prescribed error tolerance and fixed time interval $[0, T ]$, there exist choices of $D_L$ and $N$ satisfying explicit conditions such that the error between the numerical and exact solutions is below the tolerance. Specifically, the estimate shows that sufficiently large $L$ and $N$ (depending on initial data parameters and $T$) can reduce the error to any desired level. Numerical simulations based on this construction are also presented. The results in particular demonstrate the mathematical validity of the spectral method proposed in the referenced literature.

Figures

Figures reproduced from arXiv: 2509.09276 by the authors.

**Figure 1.** Maxwellian interactions: time evolution of the L 2 error norm in log scale for the scheme (1.12) with respect to L for N = 48 (left) and N = 56 (right). In the second part, we consider a sufficiently large domain DL and focus on the error with respect to the number of Fourier modes per direction N. We then plot the evolution of the error (still on a logarithmic scale) for L = 12 and L = 14 in [PITH_FULL_IMAGE:figur… view at source ↗

**Figure 2.** Maxwellian interactions: time evolution of the L 2 error norm in log scale for the scheme (1.12) with respect to N for L = 12 (left) and L = 14 (right) [PITH_FULL_IMAGE:figures/full_fig_p034_2.png] view at source ↗

**Figure 3.** Maxwellian interactions: error with respect to L/N, which illustrates the spectral accuracy when L ≪ N. This test is used to compute the time evolution of the numerical solution (1.12) and to compare the results with those obtained in [45] and with a reference solution obtained with N = 128 and a small time step [PITH_FULL_IMAGE:figures/full_fig_p034_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Coulombian interactions : time evolution of the (a) entropy H(f), (b) fourth order moment M4, (c) relative entropy H(f | µf ) and (d) L 2 norm of f − µf . (2) Smoothing estimates. Let f ∈ C([0, ∞); Hr m) ∩ L 2 loc([0, ∞); H r+1 m− 3 2 ) be a global solution of the equa…

**Figure 5.** Figure 5: Coulombian interactions : (a) time evolution the distribution function f(t, 0, 0, vz) for N = 64 at time t = 0, 0.5, 1, 1.5 and 3 (b) time evolution of the Fisher information I(f). where C(t) is a locally uniformly bounded function depending only on ∥f0∥Hrm , r, m, n,…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 5 linked inside Pith

[32]

Existence, Uniqueness and Smoothing Effect for Spatially Homogeneous Landau- Coulomb Equation in H − 1 2 Space with Polynomial Tail

He, L.-B., Ji, J., and Luo, Y. Existence, Uniqueness and Smoothing Effect for Spatially Homogeneous Landau- Coulomb Equation in H − 1 2 Space with Polynomial Tail. arXiv pre-print server, arXiv:2412.07287

arXiv
[27]

Guillen, N., and Silvestre, L. (2025). The Landau Equation Does Not Blow Up. Acta Mathematica, 234, 315-375

2025
[1]

J., Gamba, I

Alonso, R. J., Gamba, I. M., and Tharkabhushanam, S. H. (2018). Convergence and Error Estimates for the Lagrangian-Based Conservative Spectral Method for Boltzmann Equations. SIAM Journal on Numerical Analysis, 56(6), 3534-3579

2018
[2]

Bird, G. A. (1994). Molecular Gas Dynamics and the Direct Simulation of Gas Flows. Oxford University Press

1994
[3]

Bobylev, A. V. (1975). Exact Solutions of the Boltzmann Equation. Akademiia Nauk SSSR Doklady, 225, 1296- 1299

1975
[4]

V., Gamba, I

Bobylev, A. V., Gamba, I. M., and Zhang, C. (2017). On the Rate of Relaxation for the Landau Kinetic Equation and Related Models. Journal of Statistical Physics, 168(3), 535-548

2017
[5]

V., and Potapenko, I

Bobylev, A. V., and Potapenko, I. F. (2013). Monte Carlo Methods and Their Analysis for Coulomb Collisions in Multicomponent Plasmas. Journal of Computational Physics, 246, 123-144

2013
[6]

Buet, C., Cordier, S., Degond, P., and Lemou, M. (1997). Fast Algorithms for Numerical, Conservative, and Entropy Approximations of the Fokker–Planck–Landau Equation. Journal of Computational Physics, 133(2), 310-322

1997

Show all 52 references

[7]

Carrapatoso, K. (2015). On the Rate of Convergence to Equilibrium for the Homogeneous Landau Equation with Soft Potentials. Journal De Math´ ematiques Pures Et Appliqu´ ees, 104(2), 276-310

2015
[8]

Carrapatoso, K., Desvillettes, L., and He, L. B. (2017). Estimates for the Large Time Behavior of the Landau Equation in the Coulomb Case. Archive for Rational Mechanics and Analysis, 224(2), 381-420

2017
[9]

Carrapatoso, K., and Mischler, S. (2017). Landau Equation for Very Soft and Coulomb Potentials Near Maxwellians. Annals of PDE, 3(1), 1

2017
[10]

Carrapatoso, K., Tristani, I., and Wu, K.-C. (2016). Cauchy Problem and Exponential Stability for the Inhomo- geneous Landau Equation. Archive for Rational Mechanics and Analysis, 221

2016
[11]

A., Delgadino, M

Carrillo, J. A., Delgadino, M. G., and Wu, J. (2023). Convergence of a Particle Method for a Regularized Spatially Homogeneous Landau Equation. Mathematical Models and Methods in Applied Sciences, 33(05), 971-1008

2023
[12]

A., Hu, J., Wang, L., and Wu, J

Carrillo, J. A., Hu, J., Wang, L., and Wu, J. (2020). A Particle Method for the Homogeneous Landau Equation. Journal of Computational Physics: X, 7, 100066

2020
[13]

A., Jin, S., and Tang, Y

Carrillo, J. A., Jin, S., and Tang, Y. (2022). Random Batch Particle Methods for the Homogeneous Landau Equation. Communications in Computational Physics, 31, 997-1019

2022
[14]

and Lucquin-Desreux, B

Degond, P. and Lucquin-Desreux, B. (1992). The Fokker-Planck Asymptotics of the Boltzmann Collision Operator in the Coulomb Case. Mathematical Models and Methods in Applied Sciences, 02(02), 167-182

1992
[15]

Degond, P., and Lucquin-Desreux, B. (1994). An Entropy Scheme for the Fokker-Planck Collision Operator of Plasma Kinetic Theory. Numerische Mathematik, 68(2), 239-262

1994
[16]

Desvillettes, L. (2015). Entropy Dissipation Estimates for the Landau Equation in the Coulomb Case and Appli- cations. Journal of Functional Analysis, 269(5), 1359-1403

2015
[17]

Desvillettes, L., He, L.-B., and Jiang, J.-C. (2023). A New Monotonicity Formula for the Spatially Homogeneous Landau Equation with Coulomb Potential and Its Applications. Journal of the European Mathematical Society, 26, 1747–1793

2023
[18]

Desvillettes, L., and Villani, C. (2000). On the Spatially Homogeneous Landau Equation for Hard Potentials Part I : Existence, Uniqueness and Smoothness. Communications in Partial Differential Equations, 25(1-2), 179-259

2000
[19]

Desvillettes, L., and Villani, C. (2000). On the Spatially Homogeneous Landau Equation for Hard Potentials Part II : H-Theorem and Applications. Communications in Partial Differential Equations, 25(1-2), 261-298

2000
[20]

Du, K., Li, L., Xie, Y., and Yu, Y. (2025). A Structure-Preserving Collisional Particle Method for the Landau Kinetic Equation. arXiv pre-print server, arXiv:2501.00263

2025 arXiv
[21]

Filbet, F. (2020). A Spectral Collocation Method for the Landau Equation in Plasma Physics. arXiv pre-print server, arXiv:2006.15885

2020 arXiv
[22]

Filbet, F., and Mouhot, C. (2011). Analysis of Spectral Methods for the Homogeneous Boltzmann Equation, Transactions of the American Mathematical Society, 363(4), 1947-1980

2011
[23]

Filbet, F., and Pareschi, L. (2002). A Numerical Method for the Accurate Solution of the Fokker–Planck–Landau Equation in the Nonhomogeneous Case. Journal of Computational Physics, 179(1), 1-26

2002
[24]

Fournier, N. (2009). Particle Approximation of Some Landau Equations. Kinetic and Related Models, 2(3), 451-464

2009
[25]

Fournier, N., and Gu´ erin, H. (2009). Well-Posedness of the Spatially Homogeneous Landau Equation for Soft Potentials. Journal of Functional Analysis, 256(8), 2542-2560

2009
[26]

Golding, W., Gualdani, M., and Loher, A. (2025). Global Smooth Solutions to the Landau–Coulomb Equation in L3/2. Archive for Rational Mechanics and Analysis, 249(3), 34

2025
[28]

Guo, Y. (2002). The Landau Equation in a Periodic Box. Communications in Mathematical Physics, 231(3), 391-434. NUMERICAL ANALYSIS OF LANDAU EQUATION 43

2002
[29]

He, L.-B. (2018). Sharp Bounds for Boltzmann and Landau Collision Operators. Annales Scientifiques De L Ecole Normale Superieure, 51(5), 1253-1341

2018
[30]

He, L.-B., and Ji, J. (2022). On the Weak Solution to the Spatially Homogeneous Boltzmann Equation with Moderate Soft Potentials. International Journal of Mathematics, 33(09), 68

2022
[31]

He, L.-B., and Ji, J. (2023). Regularity Estimates for the Non-Cutoff Soft Potential Boltzmann Equation with Typical Rough and Slowly Decaying Data. arXiv pre-print server, arXiv:2305.05856

2023 arXiv
[33]

Henderson, C., Snelson, S., and Tarfulea, A. (2020). Local Solutions of the Landau Equation with Rough, Slowly Decaying Initial Data. Annales de l’Institut Henri Poincar´ e C, Analyse non lin´ eaire, 37(6), 1345-1377

2020
[34]

Hu, J., Qi, K., and Yang, T. (2021). A New Stability and Convergence Proof of the Fourier-Galerkin Spectral Method for the Spatially Homogeneous Boltzmann Equation. SIAM Journal on Numerical Analysis, 59(2), 613- 633

2021
[35]

Huang, Y., and Wang, L. (2025). A Score-based Particle Method for Homogeneous Landau Equation. Journal of Computational Physics, 536, 114053

2025
[36]

Ilin, V., Hu, J., and Wang, Z. (2025). Transport Based Particle Methods for the Fokker-Planck-Landau Equation. Communications in Mathematical Sciences, 23, 1763-1788

2025
[37]

Krook, M., and Wu, T. T. (1977). Exact Solutions of the Boltzmann Equation. The Physics of Fluids, 20(10), 1589-1595

1977
[38]

Y., Jang, J., and Hwang, H

Lee, J. Y., Jang, J., and Hwang, H. J. (2023). opPINN: Physics-Informed Neural Network with Operator Learning to Approximate Solutions to the Fokker-Planck-Landau Equation. Journal of Computational Physics, 480, 112031

2023
[39]

Lemou, M. (1998). Multipole Expansions for the Fokker-Planck-Landau Operator. Numerische Mathematik, 78(4), 597-618

1998
[40]

Li, R., Ren, Y., and Wang, Y. (2021). Hermite Spectral Method for Fokker-Planck-Landau Equation Modeling Collisional Plasma. Journal of Computational Physics, 434, 110235

2021
[41]

Li, R., Wang, Y., and Wang, Y. (2020). Approximation to Singular Quadratic Collision Model in Fokker-Planck- Landau Equation. SIAM Journal on Scientific Computing, 42(3), B792-B815

2020
[42]

Lifshitz, E. M. (1992). Perspectives in Theoretical Physics. Pergamon Press

1992
[43]

A., Churchill, R

Miller, M. A., Churchill, R. M., Dener, A., Chang, C. S., Munson, T., and Hager, R. (2021). Encoder–decoder Neural Network for Solving the Nonlinear Fokker–Planck–Landau Collision Operator in XGC. Journal of Plasma Physics, 87(2), 905870211, Article 905870211

2021
[44]

Noh, H., Lee, J., and Yoon, E. (2025). FPL-net: A Deep Learning Framework for Solving the Nonlinear Fokker–Planck–Landau Collision Operator for Anisotropic Temperature Relaxation. Journal of Computational Physics, 523, 113665

2025
[45]

Pareschi, L., Russo, G., and Toscani, G. (2000). Fast Spectral Methods for the Fokker–Planck–Landau Collision Operator. Journal of Computational Physics, 165(1), 216-236

2000
[46]

A., and Gamba, I

Pennie, C. A., and Gamba, I. M. (2019). Entropy Decay Rates for Conservative Spectral Schemes Modeling Fokker-Planck-Landau Type Flows in the Mean Field Limit. arXiv pre-print server, arXiv:1910.03110

2019 arXiv
[47]

Toscani, G., and Villani, C. (2000). On the Trend to Equilibrium for Some Dissipative Systems with Slowly Increasing a Priori Bounds. Journal of Statistical Physics, 98(5), 1279-1309

2000
[48]

Villani, C. (1998). On a New Class of Weak Solutions to the Spatially Homogeneous Boltzmann and Landau Equations. Archive for Rational Mechanics and Analysis, 143(3), 273-307

1998
[49]

Villani, C. (1998). On the Spatially Homogeneous Landau Equation for Maxwellian Molecules. Mathematical Models and Methods in Applied Sciences. 08(06), 957-983

1998
[50]

Wollman, S. (2017). Numerical Approximation of the Spatially Homogeneous Fokker–Planck–Landau Equation. Journal of Computational and Applied Mathematics, 324, 173-203

2017
[51]

Wollman, S. (2024). Finite Difference Approximations of the Spatially Homogeneous Fokker–Planck–Landau Equation. Journal of Computational and Applied Mathematics, 449, 115928

2024
[52]

Zhang, C., and Gamba, I. M. (2017). A Conservative Scheme for Vlasov Poisson Landau Modeling Collisional Plasmas. Journal of Computational Physics, 340, 470-497. 44 FRANCIS FILBET, YANZHI GUI, AND LING-BING HE (Francis Filbet) Institut de Math´ematiques de Toulouse, Universit ...

2017

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [32]

Existence, Uniqueness and Smoothing Effect for Spatially Homogeneous Landau- Coulomb Equation in H − 1 2 Space with Polynomial Tail

He, L.-B., Ji, J., and Luo, Y. Existence, Uniqueness and Smoothing Effect for Spatially Homogeneous Landau- Coulomb Equation in H − 1 2 Space with Polynomial Tail. arXiv pre-print server, arXiv:2412.07287

arXiv

[2] [27]

Guillen, N., and Silvestre, L. (2025). The Landau Equation Does Not Blow Up. Acta Mathematica, 234, 315-375

2025

[3] [1]

J., Gamba, I

Alonso, R. J., Gamba, I. M., and Tharkabhushanam, S. H. (2018). Convergence and Error Estimates for the Lagrangian-Based Conservative Spectral Method for Boltzmann Equations. SIAM Journal on Numerical Analysis, 56(6), 3534-3579

2018

[4] [2]

Bird, G. A. (1994). Molecular Gas Dynamics and the Direct Simulation of Gas Flows. Oxford University Press

1994

[5] [3]

Bobylev, A. V. (1975). Exact Solutions of the Boltzmann Equation. Akademiia Nauk SSSR Doklady, 225, 1296- 1299

1975

[6] [4]

V., Gamba, I

Bobylev, A. V., Gamba, I. M., and Zhang, C. (2017). On the Rate of Relaxation for the Landau Kinetic Equation and Related Models. Journal of Statistical Physics, 168(3), 535-548

2017

[7] [5]

V., and Potapenko, I

Bobylev, A. V., and Potapenko, I. F. (2013). Monte Carlo Methods and Their Analysis for Coulomb Collisions in Multicomponent Plasmas. Journal of Computational Physics, 246, 123-144

2013

[8] [6]

Buet, C., Cordier, S., Degond, P., and Lemou, M. (1997). Fast Algorithms for Numerical, Conservative, and Entropy Approximations of the Fokker–Planck–Landau Equation. Journal of Computational Physics, 133(2), 310-322

1997