用神经马尔可夫核实现零样本辐射-物质相互作用模拟，速度快且可微。
BRICKS: Compositional Neural Markov Kernels for Zero-Shot Radiation-Matter Simulation
- 基于黎曼流匹配的混合离散-连续变换器构建下一粒子预测核
- 单核执行在GPU上比传统模拟快数倍，支持零样本大尺度材料仿真
- 模型可微且提供可计算似然，适合需要快速推断的物理仿真任务
我们提出一种用于辐射-物质相互作用的组合式神经代理新策略，该任务涵盖粒子物理、核工程、空间工程到医学物理等多个领域。利用粒子相互作用的局部性和马尔可夫特性，我们基于流形上的黎曼流匹配，构建了一个混合离散-连续变压器的「下一粒子预测」核，能够生成由入射粒子与材料体积相互作用产生的变长类型化粒子及辐射效应集合。该核可组合以实现对未见大规模材料分布的零样本仿真。与机制模拟器不同，本模型设计为可微分，可提供未来下游应用所需的可计算似然。单核执行时，相较于依赖CPU的机制模拟，观察到显著的计算加速。我们在核级别评估模型，展示了多轮自回归推演中的预测稳定性。此外，我们还发布了全新的2000万事件辐射-物质相互作用数据集，供进一步研究使用。
原文摘要 · Abstract (English)
We introduce a new strategy for compositional neural surrogates for radiation-matter interactions, a key task spanning domains from particle physics through nuclear and space engineering to medical physics. Exploiting the locality and the Markov nature of particle interactions, we create a \emph{next-particle prediction} kernel using hybrid discrete-continuous transformer models based on Riemannian Flow Matching on product manifolds. The model generates variable-sized typed sets of particles and radiation side effects that are the result of the interaction of an incident particle with a material volume. The resulting kernel can be composed to simulate unseen large-scale material distributions in a zero-shot manner. Unlike mechanistic simulators, our model is designed to be differentiable, provides tractable likelihoods for future downstream applications. A significant computational speed-up on GPU compared to CPU-bound mechanistic simulation is observed for single-kernel execution. We evaluate the model at the kernel level and demonstrate predictive stability over multi-round autoregressive rollouts. We additionally release a novel 20M-event radiation-matter interaction dataset for further research.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。