Neural-Network Chemical Emulator for First-Star Formation: Robust Iterative Predictions over a Wide Density Range
Sojun Ono and Kazuyuki Sugimura
我们提出了用于第三人口恒星形成中热和化学演化的神经网络模拟器。 模拟器在21个数量级(10^-3-10^18 cm^-3)的宽密度范围内精确再现热化学进化,跟踪六个原始物种:H,H_2,e^-,H^+,H^-和H_2^+。 为了处理广泛的动态范围,我们将密度范围划分为五个次区域,并训练每个区域的单独深度运营商网络(DeepOnets)。 当应用于随机采样的热化学状态时,模拟器实现相对误差低于10
We present a neural-network emulator for the thermal and chemical evolution in Population III star formation. The emulator accurately reproduces the thermochemical evolution over a wide density range spanning 21 orders of magnitude (10^-3-10^18 cm^-3), tracking six primordial species: H, H_2, e^-, H^+, H^-, and H_2^+. To handle the broad dynamic range, we partition the density range into five subregions and train separate deep operator networks (DeepONets) in each region. When applied to randoml...