论文标题
部分可观测时空混沌系统的无模型预测
Primordial non-Guassianity in inflation with gravitationally enhanced friction
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。
The gravitationally enhanced friction can reduce the speed of the inflaton to realize an ultra-slow-roll inflation, which will amplify the curvature perturbations. The amplified perturbations can generate a sizable amount of primordial black holes (PBHs) and induce simultaneously a significant background gravitational waves (SIGWs). In this paper, we investigate the primordial non-Gaussianity of the curvature perturbations in the inflation with gravitationally enhanced friction. We find that when the gravitationally enhanced friction plays a role in the inflationary dynamics, the non-Gaussianity is noticeably larger than that from the standard slow-roll inflation. During the regime in which the power spectrum of the curvature perturbations is around its peak, the non-Gaussianity parameter changes from negative to positive. When the power spectrum is at its maximum, the non-Gaussianity parameter is near zero ($\sim \mathcal{O}(0.01)$). Furthermore, the primordial non-Gaussianity promotes the formation of PBHs, while its effect on SIGWs is negligible.
