Micro-Macro Tensor Neural Surrogates for Uncertainty Quantification in Collisional Plasma
Authors: Wei Chen, Giacomo Dimarco, Lorenzo Pareschi
Organizations: School of Mathematical Sciences, Xiamen University, China · Department of Mathematics and Computer Science, University of Ferrara, Italy · Maxwell Institute for Mathematical Sciences and Department of Mathematics, School of Mathematical and Computer Sciences, Heriot-Watt University, Edinburgh, UK
Plasma kinetic equations exhibit pronounced sensitivity to microscopic perturbations in model parameters and data, making reliable and efficient uncertainty quantification (UQ) essential for predictive simulations. However, the cost of uncertainty sampling, the high-dimensional phase space, and multiscale stiffness pose severe challenges to both computational efficiency and error control in traditional numerical methods. These aspects are further emphasized in presence of collisions where the high-dimensional nonlocal collision integrations and conservation properties pose severe constraints. To overcome this, we present a variance-reduced Monte Carlo framework for UQ in the Vlasov--Poisson--Landau (VPL) system, in which neural network surrogates replace the multiple costly evaluations of the Landau collision term. The method couples a high-fidelity, asymptotic-preserving VPL solver with inexpensive, strongly correlated surrogates based on the Vlasov--Poisson--Fokker--Planck (VPFP) and Euler--Poisson (EP) equations. For the surrogate models, we introduce a generalization of the separable physics-informed neural network (SPINN), developing a class of tensor neural networks based on an anisotropic micro-macro decomposition, to reduce velocity-moment costs, model complexity, and the curse of dimensionality. To further increase correlation with VPL, we calibrate the VPFP model and design an asymptotic-preserving SPINN whose small- and large-Knudsen limits recover the EP and VP systems, respectively. Numerical experiments show substantial variance reduction over standard Monte Carlo, accurate statistics with far fewer high-fidelity samples, and lower wall-clock time, while maintaining robustness to stochastic dimension.
Figures & tables
Figure 1: The Tensor Neural Network.
Figure 2: UQ-SPINN for Vlasov–Poisson–Fokker–Planck equation: the macro tensor network T1 builds the anisotropic Maxwellian background, while the tensor network T2 learns only the kinetic correction. A third tensor network T3 solves the Poisson problem.
Category
Notation
Description
Subscript
FP
Homogeneous Fokker–Planck
EP
Euler-Poisson
VPFP
Vlasov-Poisson-Fokker–Planck
VPL
Vlasov–Poisson–Landau
Superscript
Calibrated Model
Table 1: Symbols and abbreviations used in the numerical examples.
Figure 3: Left: the L1 and L∞ errors between Landau operator Q(f,f) and the calibrated Fokker–Planck operator μP(f) ; Right: the model error with respect to VPL for standard model, calibrated model without data, and calibrated model augmented with VPL data.
Figure 4: Different decomposition. Left: background term M ; Right: perturbation term g ; Top: anisotropic reference Maxwellian; Middle: isotropic reference Maxwellian; Bottom: moment-matched Maxwellian.
Figure 5: Training loss for Example 6.1 .
Figure 6: L1 error of expectation for Example 6.2 . The sample sizes used to compute the expected value and to construct the control variate are K=5 and L=2500 , respectively. Left: short time behavior; Right: long time behavior.
Figure 7: L1 error of expectation for Example 6.2 under different training region.
Method
Wall-clock time
Relative ratio
EP-based UQ-SPINN
493.132+0.009×200
1
Calibrated-VPFP-based UQ-SPINN
2312.217+0.015×200
4.68
Deterministic EP solver
13.828×200
5.58
Deterministic calibrated VPFP solver
72633.966×200
2.94×104
Deterministic VPL solver
101151.912×200
4.09×104
Table 2: Wall-clock time comparison for 20000 samples.
Figure 8: Left: global training and windowed training at ε=1e+6 ; Right: windowed training at ε=1 .
Figure 9: Results for Example 6.3 . The sample sizes used to compute the expected value and to construct the control variate are K=15 , L=20000 for calibrated VPFP and EP, respectively. Left: L1 error of expectation E[ρ] ; Right: expectation E[ζ] and confidence bands; Top: ε=1e−4 ; Middle: ε=1 ; Bottom: ε=1e+6 .
Example
Method
Equation
Training
Inference
(s)
(s / 100 samples)
6.3
UQ-SPINN
EP
493.132
0.009
UQ-SPINN
calibrated VPFP
2312.217
0.015
Deterministic Solver
EP
N/A
13.828
Deterministic Solver
calibrated VPFP
N/A
72633.966
Deterministic Solver
VPL
N/A
101151.912
Table 3: Runtime comparison for Example 6.3 – 6.5 . “Training” is the wall-clock training time for the UQ-SPINN (not applicable to the deterministic baseline). “Inference” is the wall-clock time to evaluate 100 samples.
Figure 10: Results for Example 6.4 . The sample sizes used to compute the expected value and to construct the control variate are K=15 , L=20000 for calibrated VPFP and EP, respectively. Left: L1 error of expectation E[ρ] ; Right: expectation E[ζ] and confidence bands; Top: ε=1e−4 ; Middle: ε=1 ; Bottom: ε=1e+6 .
Figure 11: Results for Example 6.4 . L1 error of expectation E[ρ] . Left: based on EP model; Right: based on calibrated VPFP model.
Figure 12: Results for Example 6.5 . Expectation of projections PVPFP∗ at t=0,1.4, and 8 ; Top: ε=1e−4 ; Bottom: ε=1 .
Figure 13: Results for Example 6.5 . Expectation of projections PVPFP∗ at t=0,15, and 20 for ε=1e+6 .
Figure 14: L1 error of expectation E[ρ] for Example 6.5 . The sample sizes used to compute the expected value and to construct the control variate are K=15 , L=20000 for calibrated VPFP and EP, respectively. Top Left: ε=1e−4 ; Top Right: ε=1 ; Bottom: ε=1e+6 .
The high dimensionality of kinetic equations with stochastic parameters poses major computational challenges for uncertainty quantification (UQ). Traditional Monte Carlo (MC) sampling methods, while widely used, suffer from slow convergence and high variance, which become increasingly severe as the dimensionality of the parameter space grows. To accelerate MC sampling, we adopt a multiscale control variates strategy that leverages low-fidelity solutions from simplified kinetic models to reduce variance. To further improve sampling efficiency and preserve the underlying physics, we introduce surrogate models based on structure and asymptotic preserving neural networks (SAPNNs). These deep neural networks are specifically designed to satisfy key physical properties, including positivity, conservation laws, entropy dissipation, and asymptotic limits. By training the SAPNNs on low-fidelity models and enriching them with selected high-fidelity samples from the full Boltzmann equation, our method achieves significant variance reduction while maintaining physical consistency and asymptotic accuracy. The proposed methodology enables efficient large-scale prediction in kinetic UQ and is validated across both homogeneous and nonhomogeneous multiscale regimes. Numerical results demonstrate improved accuracy and computational efficiency compared to standard MC techniques.
Wei Chen, Giacomo Dimarco, Lorenzo Pareschi
School of Mathematical Sciences, Xiamen University, China · Department of Mathematics and Computer Science, University of Ferrara, Italy · Maxwell Institute for Mathematical Sciences and Department of Mathematics, School of Mathematical and Computer Sciences, Heriot-Watt University, Edinburgh, UK
Plasma modeling is central to the design of nuclear fusion reactors, yet simulating collisional plasma kinetics from first principles remains a formidable computational challenge: the Vlasov-Maxwell-Landau (VML) system describes six-dimensional phase-space transport under self-consistent electromagnetic fields together with the nonlinear, nonlocal Landau collision operator. A recent deterministic particle method for the full VML system estimates the velocity score function via the blob method, a kernel-based approximation with O(n2) cost. In this work, we replace the blob score estimator with score-based transport modeling (SBTM), in which a neural network is trained on-the-fly via implicit score matching at O(n) cost. We prove that the approximated collision operator preserves momentum and kinetic energy, and dissipates an estimated entropy. We also characterize the unique global steady state of the VML system and its electrostatic reduction, providing the ground truth for numerical validation. On three canonical benchmarks -- Landau damping, two-stream instability, and Weibel instability -- SBTM is more accurate than the blob method, achieves correct long-time relaxation to Maxwellian equilibrium where the blob method fails, and delivers 50% faster runtime with 4× lower peak memory.
Vasily Ilin, Jingwei Hu
Department of Mathematics, University of Washington, Seattle, WA, USA. · Department of Applied Mathematics, University of Washington, Seattle, WA, USA.
Kinetic equations in the high-field regime describe particle transport dominated by electric-field acceleration and collisional relaxation. These equations are widely used in simulations of plasmas and semiconductor devices. Typical models include the Vlasov--Poisson--Fokker--Planck (VPFP) and Boltzmann--Poisson systems. Their high dimensionality and multiscale nature pose two of the major challenges to accurate and efficient numerical solution. To address these challenges, we propose an asymptotic-preserving tensor neural network (APTNN) method. Specifically, to reduce high-dimensional computational costs and memory requirements, we adopt the canonical polyadic (CP) decomposition structure of TNNs to approximate their solutions. We show that the equation residuals retain this decomposition form, allowing high-dimensional loss integrals to be evaluated through combinations of one-dimensional integrals. The resulting computational and storage costs are reduced from exponential to polynomial growth with dimension. To improve the prediction accuracy in the high-field limit, we enhance the baseline AP loss by incorporating the first moment equation. The added loss term directly constrains the limiting flux closure to reduce its approximation error in the high-field regime. The enhanced AP loss can still be evaluated dimension-wise. Numerical results show that APTNN improves prediction accuracy in the high-field regime with significantly reduced computational costs and memory usage. For a 3D3V VPFP problem with multiple random inputs, training takes less than 30 minutes on a single GPU.
Shi Jin, Tian-ai Zhang
School of Mathematical Sciences, Shanghai Jiao Tong University, Shanghai, China · Institute of Natural Sciences, MOE-LSC, Shanghai Jiao Tong University, Shanghai, China · Shanghai Artificial Intelligence Laboratory, Shanghai, China