论文标题
部分可观测时空混沌系统的无模型预测
How magnetic field and stellar radiative feedback influences the collapse and the stellar mass spectrum of a massive star forming clump
论文作者
论文摘要
储层计算是预测湍流的有力工具,其简单的架构具有处理大型系统的计算效率。然而,其实现通常需要完整的状态向量测量和系统非线性知识。我们使用非线性投影函数将系统测量扩展到高维空间,然后将其输入到储层中以获得预测。我们展示了这种储层计算网络在时空混沌系统上的应用,该系统模拟了湍流的若干特征。我们表明,使用径向基函数作为非线性投影器,即使只有部分观测并且不知道控制方程,也能稳健地捕捉复杂的系统非线性。最后,我们表明,当测量稀疏、不完整且带有噪声,甚至控制方程变得不准确时,我们的网络仍然可以产生相当准确的预测,从而为实际湍流系统的无模型预测铺平了道路。
In spite of decades of theoretical efforts, the physical origin of the stellar initial mass function (IMF) is still debated. We aim at understanding the influence of various physical processes such as radiative stellar feedback, magnetic field and non-ideal magneto-hydrodynamics on the IMF. We present a series of numerical simulations of collapsing 1000 M$_\odot$ clumps taking into account radiative feedback and magnetic field with spatial resolution down to 1 AU. Both ideal and non-ideal MHD runs are performed and various radiative feedback efficiencies are considered. We also develop analytical models that we confront to the numerical results. The sum of the luminosities produced by the stars in the calculations is computed and it compares well with the bolometric luminosities reported in observations of massive star forming clumps. The temperatures, velocities and densities are also found to be in good agreement with recent observations. The stellar mass spectrum inferred for the simulations is, generally speaking, not strictly universal and in particular varies with magnetic intensity. It is also influenced by the choice of the radiative feedback efficiency. In all simulations, a sharp drop in the stellar distribution is found at about $M_{min} \simeq$ 0.1 M$_\odot$, which is likely a consequence of the adiabatic behaviour induced by dust opacities at high densities. As a consequence, when the combination of magnetic and thermal support is not too large, the mass distribution presents a peak located at 0.3-0.5 M$_\odot$. When magnetic and thermal support are large, the mass distribution is better described by a plateau, i.e. $d N / d \log M \propto M^{-Γ}$, $Γ\simeq 0$. Abridged