Data assimilation and application in radiation physics and shielding

1.在高保真问题上开发和测试新的、高效的Scon敏感技术，证明它们比天真的方法更快。将这一敏感性分析应用于劳斯莱斯感兴趣的一个代表性问题。这可能包括获得对材料边界的灵敏度，这方面的一些工作已经在本征值环境中进行了[3]。将获得的灵敏度与数据同化框架中的实验测量相结合，以最大限度地减少不确定性或校准不确定的模型输入。对感兴趣的问题产生更高阶的敏感度，从而能够在例如模式输入的协方差已知的情况下使用更复杂的数据同化方法。这允许放松输入参数中响应分布为高斯分布的假设，并更好地表征真实分布。探索通过混合多组和连续能量计算加速权重窗口生成。

1. Develop and test new, efficient sensitivity techniques in SCONE on high-fidelity problems, demonstrating their speed-up over naïve approaches.2. Apply this sensitivity analysis to a representative problem of interest for Rolls Royce. This may include obtaining sensitivities to material boundaries, for which some work has been performed in the eigenvalue context [3].3. Combine the obtained sensitivities with experimental measurements in a data assimilation framework in order to minimise uncertainties or calibrate uncertain model inputs.4. Generate higher-order sensitivities for the problems of interest, allowing use of more sophisticated data assimilation approaches where, e.g., the covariance of the model inputs isknown. This allows for relaxing the assumption that the response distribution is Gaussian in the input parameters and better characterising the true distribution.5. Explore accelerated weight-window generating through hybrid multi-group and continuous energy calculations.