论文标题
部分可观测时空混沌系统的无模型预测
Will the last be the first? School closures and educational outcomes
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。
Governments have implemented school closures and online learning as one of the main tools to reduce the spread of Covid-19. Despite the potential benefits in terms of reduction of cases, the educational costs of these policies may be dramatic. This work identifies the educational costs, expressed as decrease in test scores, for the whole universe of Italian students attending the 5th, 8th and 13th grade of the school cycle during the 2021/22 school year. The analysis relies on a difference-in-difference model in relative time, where the control group is the closest generation before the Covid-19 pandemic. The results suggest a national average loss between 1.6-4.1% and 0.5-2.4% in Mathematics and Italian test scores, respectively. After collecting the precise number of days of school closures for the universe of students in Sicily, we estimate that 30 additional days of closure decrease the test score by 1%. However, the impact is much larger for students from high schools (1.8%) compared to students from low and middle schools (0.5%). This is likely explained by the lower relevance of parental inputs and higher reliance on peers inputs, within the educational production function, for higher grades. Findings are also heterogeneous across class size and parental job conditions, pointing towards potential growing inequalities driven by the lack of in front teaching.
