Johannes Krotz,Cory D. Hauck,Ryan G. McClarren
Abstract:We present a hybrid method for time-dependent particle transport problems that combines Monte Carlo (MC) estimation with deterministic solutions based on discrete ordinates. For spatial discretizations, the MC algorithm computes a piecewise constant solution and the discrete ordinates uses bilinear discontinuous finite elements. From the hybridization of the problem, the resulting problem solved by Monte Carlo is scattering free, resulting in a simple, efficient solution procedure. Between time steps, we use a projection approach to ``relabel'' collided particles as uncollided particles. From a series of standard 2-D Cartesian test problems we observe that our hybrid method has improved accuracy and reduction in computational complexity of approximately an order of magnitude relative to standard discrete ordinates solutions.
What problem does this paper attempt to address?
This paper attempts to solve the trade - off between computational complexity and accuracy in particle transport problems. Specifically, it proposes a hybrid method that combines Monte Carlo (MC) estimation and the deterministic solution method based on discrete ordinates, in order to obtain a more efficient and accurate solution in time - dependent particle transport problems.
### Problem Background
The kinetic transport equations are important tools for describing the propagation behavior of particles (such as neutrons, photons, etc.) in media. Numerical solutions to such equations are generally divided into two types of methods: deterministic methods and Monte Carlo methods.
- **Deterministic methods**: Directly discretize the phase space (physical space, flight direction, particle energy) and time. For large - scale seven - dimensional problems (three - dimensional physical space, two - dimensional flight direction, one - dimensional energy and one - dimensional time), it is very difficult to construct high - resolution solutions, and the computational cost and memory requirements are huge.
- **Monte Carlo methods**: Simulate the particle transport process through sampling techniques. The advantage is that it can handle complex three - dimensional geometric structures and can continuously handle the interaction between particles and the background medium. However, the uncertainty and computational complexity of the Monte Carlo method are relatively high, especially when uniform spatial estimation is required.
### Solutions Proposed in the Paper
The paper proposes a Hybrid Monte Carlo - Discontinuous Galerkin method, aiming to combine the advantages of the two methods while reducing their respective disadvantages. Specifically:
1. **Monte Carlo treatment of uncollided particles**: For particles that have not collided, use the Monte Carlo method to track them, avoiding considering the collision process, thereby simplifying the calculation.
2. **Deterministic treatment of collided particles**: For particles after collision, use the discrete ordinates method to solve them, and use bilinear discontinuous finite elements for spatial discretization.
3. **Remapping step**: Between each time step, use the projection method to relabel the collided particles as uncollided particles to prevent the exponential decay of the number of uncollided particles.
### Method Advantages
Through the above - mentioned hybrid method, the paper has achieved the following improvements:
- **Reduced computational complexity**: Compared with the pure deterministic method, the computational complexity is reduced by approximately one order of magnitude.
- **Improved accuracy**: In the standard two - dimensional Cartesian test problem, the hybrid method shows higher accuracy.
### Mathematical Model
The particle transport equation can be expressed as:
\[
\frac{1}{c}\frac{\partial\Psi}{\partial t}+\Omega\cdot\nabla_x\Psi + \sigma_t\Psi=\frac{\sigma_s}{4\pi}\langle\Psi\rangle+Q,\quad x\in X,\Omega\in S^2,t > 0
\]
where:
- \(\Psi(x,\Omega,t)\) is the angular flux,
- \(\sigma_t\) is the total cross - section,
- \(\sigma_s\) is the scattering cross - section,
- \(Q\) is the known particle source,
- \(\langle\Psi\rangle=\int_{S^2}\Psi d\Omega\) represents the integral on the unit sphere.
By treating the uncollided and collided parts separately:
\[
\begin{aligned}
&\frac{\partial\Psi_u}{\partial t}+\Omega\cdot\nabla_x\Psi_u+\sigma_t\Psi_u = Q,\\
&\frac{\partial\Psi_c}{\partial t}+\Omega\cdot\nabla_x\Psi_c+\sigma_t\Psi_c=\frac{\sigma_s}{4\pi}(\langle\Psi_u\rangle+\langle\Psi_c\rangle)
\end{aligned}
\]