Nonlinear Fokker-Planck Acceleration for Forward-Peaked Transport Problems in Slab Geometry

05/13/2020 ∙ by J. J. Kuczek, et al. ∙ Intra-Germanic The Ohio State University 0

This paper introduces a nonlinear acceleration technique that accelerates the convergence of solution of transport problems with highly forward-peaked scattering. The technique is similar to a conventional high-order/low-order (HOLO) acceleration scheme. The Fokker-Planck equation, which is an asymptotic limit of the transport equation in highly forward-peaked settings, is modified and used for acceleration; this modified equation preserves the angular flux and moments of the (high order) transport equation. We present numerical results using the Screened Rutherford, Exponential, and Henyey-Greenstein scattering kernels and compare them to established acceleration methods such as diffusion synthetic acceleration (DSA). We observe three to four orders of magnitude speed-up in wall-clock time compared to DSA.



There are no comments yet.


page 1

page 2

page 3

page 4

This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.

1 Introduction

Transport problems with highly forward-peaked scattering are prevalent in a variety of areas, including astrophysics, medical physics, and plasma physics HGK ; aristova ; multiphysics . For these problems, solutions of the transport equation converge slowly when using conventional methods such as source iteration (SI) adamslarsen and the generalized minimal residual method (GMRES) gmres . Moreover, diffusion-based acceleration techniques like diffusion synthetic acceleration (DSA) alcouffe and nonlinear diffusion acceleration (NDA) smithetall are generally inefficient when tackling these problems, as they only accelerate up to the first moment of the angular flux JapanFPSA . In fact, higher-order moments carry important information in problems with highly forward-peaked scattering and can be used to further accelerate convergence japanDiss .

This paper focuses on solution methods for the monoenergetic, steady-state transport equation in homogeneous slab geometry. Under these conditions, the transport equation is given by equationparentequation

with boundary conditions

Here, represents the angular flux at position and direction , is the macroscopic total cross section, is the differential scattering cross section, and is an internal source.

New innovations have paved the way to better solve this equation in systems with highly forward-peaked scattering. For instance, work has been done on modified equations and modified scattering cross section moments to accelerate convergence of anisotropic neutron transport problems khattab . In order to speed up the convergence of radiative transfer in clouds, a quasi-diffusion method has been developed aristova . In addition, the DSA-multigrid method was developed to solve problems in electron transport more efficiently trucksin .

One of the most recent convergence methods developed is Fokker-Planck Synthetic Acceleration (FPSA) JapanFPSA ; japanDiss . FPSA accelerates up to moments of the angular flux and has shown significant improvement in the convergence rate for the types of problems described above. The method returns a speed-up of several orders of magnitude with respect to wall-clock time when compared to DSA JapanFPSA .

In this paper, we introduce a new acceleration technique, called Nonlinear Fokker-Planck Acceleration (NFPA). This method returns a modified Fokker-Planck (FP) equation that preserves the angular moments of the flux given by the transport equation. This preservation of moments is particularly appealing for applications to multiphysics problems multiphysics , in which the coupling between the transport physics and the other physics can be done through the (lower-order) FP equation. To our knowledge, this is the first implementation of a numerical method that returns a Fokker-Planck-like equation that is discretely consistent with the linear Boltzmann equation.

This paper is organized as follows. Section 2 starts with a brief description of FPSA. Then, we derive the NFPA scheme. In Section 3, we discuss the discretization schemes used in this work and present numerical results. These are compared against standard acceleration techniques. We conclude with a discussion in Section 4.

2 Fokker-Planck Acceleration

In this section we briefly outline the theory behind FPSA, describe NFPA for monoenergetic, steady-state transport problems in slab geometry, and present the numerical methodology behind NFPA. The theory given here can be easily extended to higher-dimensional problems. Moreover, extending the method to energy-dependence shall not lead to significant additional theoretical difficulties.

To solve the transport problem given by Eq. 1.1 we approximate the in-scattering term in Eq. 1.1a with a Legendre moment expansion:




Here, is the Legendre moment of the angular flux, is the Legendre coefficient of the differential scattering cross section, and is the -order Legendre polynomial. For simplicity, we will drop the notation in the remainder of this section.

The solution to Eq. 2.1 converges asymptotically to the solution of the following Fokker-Planck equation in the forward-peaked limit pomraning1 :


where is the momentum transfer cross section and is the macroscopic absorption cross section.

Source Iteration adamslarsen is generally used to solve Eq. 2.1, which can be rewritten in operator notation:




and is the iteration index. This equation is solved iteratively until a tolerance criterion is met. The FP approximation shown in Eq. 2.3 can be used to accelerate the convergence of Eq. 2.1.

2.1 FPSA: Fokker-Planck Synthetic Acceleration

In the FPSA scheme JapanFPSA ; japanDiss , the FP approximation is used as a preconditioner to synthetically accelerate convergence when solving Eq. 2.1 (cf. adamslarsen for a detailed description of synthetic acceleration). When solving Eq. 2.4, the angular flux at each iteration has an error associated with it. FPSA systematically follows a predict, correct, iterate scheme. A transport sweep, one iteration in Eq. 2.4, is made for a prediction. The FP approximation is used to correct the error in the prediction, and this iteration is performed until a convergence criterion is met. The equations used are: equationparentequation


where we define as


In this synthetic acceleration method, the FP approximation is used to correct the error in each iteration of the high-order (HO) equation (2.6a). Therefore, there is no consistency between the angular moments of the flux in the HO and low-order (LO) equations.

2.2 NFPA: Nonlinear Fokker-Planck Acceleration

Similar to FPSA, NFPA uses the FP approximation to accelerate the convergence of the solution. We introduce the additive term to Eq. 2.3, obtaining the modified FP equation


The role of is to force the transport and modified FP equations to be consistent. Subtracting Eq. 2.8 from Eq. 2.1 and rearranging, we obtain the consistency term


The NFPA method is given by the following equations: equationparentequation

HO (2.10a)
LO (2.10b)
Consistency term (2.10c)

where is the angular flux obtained from the HO equation and is the angular flux obtained from the LO equation. The nonlinear HOLO-plus-consistency system given by Eq. 2.10 can be solved using any nonlinear solution technique kelley . Note that the NFPA scheme returns a FP equation that is consistent with HO transport. Moreover, this modified FP equation accounts for large-angle scattering which the standard FP equation does not. The LO equation (2.3) can then be integrated into multiphysics models in a similar fashion to standard HOLO schemes patelFBR . To solve the HOLO-plus-consistency system above, we use Picard iteration kelley : equationparentequation

Transport Sweep for HO (2.11a)
Evaluate Consistency Term (2.11b)
Solve LO Equation (2.11c)

where and are given in Eq. 2.5, and are given in Eq. 2.7, is the identity operator, and is the iteration index. Iteration is done until a convergence criterion is met.

The main advantage of setting up the LO equation in this fashion is that the stiffness matrix for LO needs to be setup and inverted only once, just as with FPSA JapanFPSA ; japanDiss . This has a large impact on the method’s performance. A flowchart of this algorithm is shown in Fig. 1.

Initial guess of flux moments



One sweep in transport equation

Flux moments converged?

Solve for consistency term

Solve for FP angular flux

Convert angular flux to moments



Figure 1: NFPA algorithm

3 Numerical Experiments

In Section 3.1 we describe the discretization methods used to implement the algorithms. In Section 3.2 we provide numerical results for 2 different choices of source and boundary conditions. For each choice we solve the problem using 3 different scattering kernels, applying 3 different choices of parameters for each kernel. We provide NFPA numerical results for these 18 cases and compare them against those obtained from FPSA and other standard methods.

All numerical experiments were performed using MATLAB. Runtime was tracked using the tic-toc functionality matlab17 , with only the solver runtime being taken into consideration in the comparisons. A 2017 MacBook Pro with a 2.8 GHz Quad-Core Intel Core i7 and 16 GB of RAM was used for all simulations.

3.1 Discretization

The Transport and FP equations were discretized using linear discontinuous finite element discretization in space mpd1 , and discrete ordinates (S) in angle landm . The Fokker-Planck operator was discretized using moment preserving discretization (MPD) mpd1 . Details of the derivation of the linear discontinuous finite element discretization can be seen in japanDiss ; martin . The finite element discretization for the Fokker-Planck equation follows the same derivation.

A brief review for the angular discretization used for the FP equation is given below. First, we use Gauss-Legendre quadrature to discretize the FP equation in angle:


for . Here, term is the discrete form of the angular Laplacian operator evaluated at angle .

The MPD scheme is then shown as


where is the MPD discretized operator defined by equationparentequation


for . Here, are the Legendre polynomials evaluated at each angle and are the respective weights.

is defined as a (N x N) operator for a vector of

angular fluxes , at spatial location .

In summary, if we write the FP equation as


then is Diag for , is a vector of source terms , and is represented by .

3.2 Numerical Results

It is shown that for slowly converging problems, typical convergence methods like suffer from false convergence adamslarsen . To work around this issue, the criterion is modified to use information about the current and previous iteration:


Two problems were tested using 200 spatial cells, = 400, , = 15, and = 16. Problem 1 has vacuum boundaries and a homogeneous isotropic source for . Problem 2 has no internal source and an incoming beam at the left boundary. The source and boundary conditions used are shown in Table 1.

Problem 1 Problem 2
Q(x) 0.5 0
0 0

Table 1: Problem Parameters

We consider three scattering kernels in this paper: Screened Rutherford pomraning1 , Exponential pomraning2 , and Henyey-Greenstein HGK . Three cases for each kernel were tested. The results obtained with NFPA are compared with those obtained using GMRES, DSA, and FPSA with the MPD scheme.

3.2.1 SRK: Screened Rutherford Kernel

The Screened Rutherford Kernel pomraning1 ; JapanFPSA is a widely used scattering kernel for modeling scattering behavior of electrons SRK . The kernel depends on the parameter , such that


The SRK has a valid FP limit as approaches 0 patelFBR . Three different values of were used to generate the scattering kernels shown in Fig. 2. GMRES, DSA, FPSA, and NFPA all converged to the same solution for problems 1 and 2. Figure 3 shows the solutions for SRK with .

Figure 2: Screened Rutherford Kernels
(a) Problem 1
(b) Problem 2
Figure 3: Results for SRK Problems with

Parameter Solver Runtime (s) Iterations
GMRES 98.8 12
DSA 2380 53585
FPSA 1.21 26
NFPA 1.39 26
GMRES 208 84
DSA 3040 69156
FPSA 0.747 16
NFPA 0.857 16
GMRES 174 124
DSA 3270 73940
FPSA 0.475 10
NFPA 0.542 10

Table 2: Runtime and Iteration Counts for Problem 1 with SRK

Parameter Solver Runtime (s) Iterations
GMRES 52.4 187
DSA 1107 25072
FPSA 0.953 20
NFPA 1.14 20
GMRES 108 71
DSA 1434 32562
FPSA 0.730 14
NFPA 0.857 14
GMRES 94.1 185
DSA 1470 33246
FPSA 0.438 8
NFPA 0.484 8

Table 3: Runtime and Iteration Counts for Problem 2 with SRK

The results of all solvers are shown in Tables 3 and 2. We see that NFPA and FPSA tremendously outperform GMRES and DSA in runtime for all cases. FPSA is a simpler method than NFPA, requiring less calculations per iteration; therefore, it is expected that it outperforms NFPA in runtime. We see a reduction in runtime and iterations for FPSA and NFPA as the FP limit is approached, with DSA and GMRES requiring many more iterations by comparison as approaches 0.

An advantage that NFPA offers is that the angular moments of the flux in the LO equation will remain consistent with those of the transport equation even as a problem becomes less forward-peaked. On the other hand, the moments found using only the FP equation and source iteration lose accuracy. To illustrate this, Problem 1 was tested using different Screened Rutherford Kernels with increasing parameters. The percent errors (relative to the transport solution) for the scalar flux obtained with the LO equation and with the standard FP equation at the center of the slab are shown in Fig. 4. It can be seen that the percent relative errors in the scalar flux of the FP solution is orders of magnitude larger than the error produced using the LO equation. The same trend can be seen when using the exponential and Henyey-Greenstein kernels.

Figure 4: Log Scale of Relative Error vs for Problem 1 at the Center of the Slab with SRK

3.2.2 EK: Exponential Kernel

The exponential kernel pomraning2 ; JapanFPSA is a fictitious kernel made for problems that have a valid Fokker-Planck limit pomraning1 . The zero moment, , is chosen arbitrarily; we define as the same zero moment from the SRK. The parameter determines the kernel: the first and second moments are given by equationparentequation

and the relationship for is

As is reduced, the scattering kernel becomes more forward-peaked.

The EK has a valid FP limit as approaches 0 patelFBR . Three different values of were used to generate the scattering kernels shown in Fig. 5. The generated scattering kernels are shown in Fig. 5. GMRES, DSA, FPSA, and NFPA all converged to the same solution for problems 1 and 2. Figure 6 shows the solutions for EK with .

Figure 5: Exponential Kernels
(a) Problem 1
(b) Problem 2
Figure 6: Results for EK Problems with

The runtimes and iterations for GMRES, DSA, FPSA, and NFPA are shown in Tables 5 and 4. We see a similar trend with the EK as seen with SRK. Smaller values lead to a reduction in runtime and iterations for NFPA and FPSA, which greatly outperform DSA and GMRES in both categories.

Parameter Solver Runtime (s) Iterations
GMRES 196 142
DSA 3110 70140
FPSA 0.514 11
NFPA 0.630 11
GMRES 156 132
DSA 3120 70758
FPSA 0.388 7
NFPA 0.393 7
GMRES 81 127
DSA 3120 70851
FPSA 0.292 6
NFPA 0.318 6

Table 4: Runtime and Iteration Counts for Problem 1 with EK

Parameter Solver Runtime (s) Iterations
GMRES 110 73
DSA 1455 33033
FPSA 0.492 10
NFPA 0.613 10
GMRES 82.7 79
DSA 1470 33309
FPSA 0.358 7
NFPA 0.431 7
GMRES 56.8 90
DSA 1470 33339
FPSA 0.273 5
NFPA 0.319 5

Table 5: Runtime and Iteration Counts for Problem 2 with EK

3.2.3 HGK: Henyey-Greenstein Kernel

The Henyey-Greenstein Kernel HGK ; JapanFPSA is most commonly used in light transport in clouds. It relies on the anisotropy factor , such that


As goes from zero to unity, the scattering shifts from isotropic to highly anisotropic.

Figure 7: Henyey-Greenstein Kernels
(a) Problem 1
(b) Problem 2
Figure 8: Results for HGK Problems with

The HGK does not have a valid FP limit patelFBR . The three kernels tested are shown in Fig. 7. GMRES, DSA, FPSA, and NFPA all converged to the same solution for problems 1 and 2. Figure 8 shows the solutions for HGK with . The results of each solver are shown in Tables 7 and 6.

Parameter Solver Runtime (s) Iterations
GMRES 9.88 76
DSA 24.5 554
FPSA 1.50 32
NFPA 1.39 27
GMRES 12.2 131
DSA 47.7 1083
FPSA 1.75 38
NFPA 1.83 35
GMRES 40.0 27
DSA 243 5530
FPSA 3.38 74
NFPA 3.93 73

Table 6: Runtime and Iteration Counts for Problem 1 with HGK

Parameter Solver Runtime (s) Iterations
GMRES 24.3 135
DSA 14.8 336
FPSA 1.15 23
NFPA 1.35 24
GMRES 31.3 107
DSA 29.7 675
FPSA 1.56 32
NFPA 1.90 33
GMRES 41.4 126
DSA 146 3345
FPSA 3.31 67
NFPA 3.99 67

Table 7: Runtime and Iteration Counts for Problem 2 with HGK

Here we see that NFPA and FPSA do not perform as well compared to their results for the SRK and EK. Contrary to what happened in those cases, both solvers require more time and iterations as the problem becomes more anisotropic. This is somewhat expected, due to HGK not having a valid Fokker-Planck limit. However, both NFPA and FPSA continue to greatly outperform GMRES and DSA. Moreover, NFPA outperforms FPSA in iteration count for problem 1.

4 Discussion

This paper introduced the Nonlinear Fokker-Planck Acceleration technique for steady-state, monoenergetic transport in homogeneous slab geometry. To our knowledge, this is the first nonlinear HOLO method that accelerates all moments of the angular flux. Upon convergence, the LO and HO models are consistent; in other words, the (lower-order) modified Fokker-Planck equation preserves the same angular moments of the flux obtained with the (higher-order) transport equation.

NFPA was tested on a homogeneous medium with an isotropic internal source with vacuum boundaries, and in a homogeneous medium with no internal source and an incoming beam boundary. For both problems, three different scattering kernels were used. The runtime and iterations of NFPA and FPSA were shown to be similar. They both vastly outperformed DSA and GMRES for all cases by orders of magnitude. However, NFPA has the feature of preserving the angular moments of the flux in both the HO and LO equations, which offers the advantage of integrating the LO model into multiphysics models.

In the future, we intend to test NFPA capabilities for a variety of multiphysics problems and analyze its performance. To apply NFPA to more realistic problems, it needs to be extended to include time and energy dependence. Additionally, the method needs to be adapted to address geometries with higher-order spatial dimensions. Finally, for the NFPA method to become mathematically “complete”, a full convergence examination using Fourier analysis must be performed. However, this is beyond the scope of this paper and must be left for future work.


The authors acknowledge support under award number NRC-HQ-84-15-G-0024 from the Nuclear Regulatory Commission. The statements, findings, conclusions, and recommendations are those of the authors and do not necessarily reflect the view of the U.S. Nuclear Regulatory Commission.

J. K. Patel would like to thank Dr. James Warsa for his wonderful transport class at UNM, as well as his synthetic acceleration codes. The authors would also like to thank Dr. Anil Prinja for discussions involving Fokker-Planck acceleration.


  • (1) L. C. Henyey, J. L. Greenstein, Diffuse radiation in the galaxy, The Astrophysics Journal 93 (1941) 70–83.
  • (2) E. N. Aristova, V. Y. Gol’din, Computation of anisotropy scattering of solar radiation in atmosphere (monoenergetic case), Journal of Quantitative Spectropy and Radiative Transfer 67 (2) (2000) 139–157.
  • (3) J. K. Patel, R. Vasques, B. D. Ganapol, Towards a multiphyiscs model for tumor response to combined-hyperthermia-radiotherapy treatment, in: Proceedings of International Conference on Mathematics & Computational Methods Applied to Nuclear Science & Engineering, Portland, OR, Aug. 25-29, 2019.
  • (4) M. L. Adams, E. W. Larsen, Fast iterative methods for discrete ordinates particle transport problems, Progress in Nuclear Energy 40 (1) (2002) 3–159.
  • (5) Y. Saad, M. H. Schultz, GMRES: A generalized minimal residual algorithm for solving nonsymmetric linear systems, SIAM Journal on Scientific and Statistical Computing 7 (3) (1986) 856–869.
  • (6) R. E. Alcouffe, Diffusion synthetic acceleration for diamond-differenced discrete-ordinates equations, Nuclear Science and Engineering 64 (2) (1977) 344–355.
  • (7) D. A. Knoll, H. Park, K. Smith, Application of the Jacobian-free Newton-Krylov method to nonlinear acceleration of transport source iteration in slab geometry, Nuclear Science and Engineering 167 (2) (2011) 122–132.
  • (8) J. K. Patel, J. S. Warsa, A. K. Prinja, Acceleration of the S Equations with Highly Anisotropic Scattering using the Fokker-Planck Equation (2019) arXiv:1907.13013.
  • (9) J. K. Patel, Fokker-Planck-based acceleration for S equations with highly forward-peaked scattering in slab geometry, Ph.D. thesis, The University of New Mexico, Albuquerque, NM (2016).
  • (10) K. M. Khattab, E. W. Larsen, Synthetic acceleration methods for linear transport problems with highly anisotropic scattering, Nuclear Science and Engineering 107 (3) (1991) 217–227.
  • (11) B. Trucksin, Acceleration techniques for discrete-ordinates transport methods with highly forward-peaked scatterings, Ph.D. thesis, The University of New Mexico, Albuquerque, NM (2015).
  • (12) G. C. Pomraning, The Fokker-Planck operator as an asymptotic limit, Mathematical Models and Methods in Applied Sciences 2 (1) (1992) 21–36.
  • (13) C. T. Kelley, Solving Nonlinear Equations with Newton’s Method, SIAM, 2003.
  • (14) J. K. Patel, H. Park, C. R. E. de Oliveira, D. A. Knoll, Efficient multiphysics coupling for fast burst reactors in slab geometry, Journal of Computational and Theoretical Transport 43 (2014) 289–313.
  • (15) MATLAB, version (R2017b), The MathWorks Inc., Natick, Massachusetts, 2017.
  • (16) J. S. Warsa, A. K. Prinja, A moment preserving S discretization for the one-dimensional Fokker-Planck equation, Transactions of the American Nuclear Society 106 (2012) 362–365.
  • (17) E. E. Lewis, W. F. Miller, Jr., Computational methods of neutron transport, American Nuclear Society.
  • (18) W. R. Martin, The application of the finite element method to the neutron transport equation, Ph.D. thesis, The University of Michigan, Ann Arbor, MI (1976).
  • (19) G. C. Pomraning, A. K. Prinja, J. W. VanDenburg, An asymptotic model for the spreading of a collimated beam, Nuclear Science and Engineering 112 (4) (1992) 347–360.
  • (20) D. A. Dixon, A computationally efficient moment-preserving Monte Carlo electron transport method with implementation in GEANT4, Ph.D. thesis, The University of New Mexico, Albuquerque, NM (2015).