We present recent advances in path-integral formulations designed for unbiased Monte Carlo sensitivity estimation (in the form of partial derivatives) within a coupled physics model. We establish the theoretical foundation and illustrate the approach by estimating instantaneous atmospheric radiative forcings. In climate studies, these quantities amount for the change in top-of-atmosphere (TOA) net radiative flux induced by an isolated change in surface or atmospheric constitution. Based on a path-integral framework, our approach results in estimations consistent with well-established radiative forcings in the climate community. We highlight how physics coupling through path-integral formulations yields unbiased sensitivity estimation of a radiative quantity (integrated TOA flux) to a spectroscopic parameter (fraction change in gas concentration). Furthermore, we emphasize the method’s scalability, demonstrating its compatibility with computer science acceleration techniques. These latter play a key role in rendering the computational time weakly sensitive to the system’s multidimensional and multiphysics complexity.

中文翻译：

---

使用光谱辐射灵敏度的路径积分公式方法对大气辐射强迫进行蒙特卡罗模拟


我们介绍了路径积分公式的最新进展，该公式专为耦合物理模型中的无偏蒙特卡罗灵敏度估计（以偏导数的形式）而设计。我们通过估计瞬时大气辐射强迫建立了理论基础并说明了该方法。在气候研究中，这些量相当于由地表或大气构成的孤立变化引起的大气层顶部（TOA）净辐射通量的变化。基于路径积分框架，我们的方法得出的估计结果与气候界既定的辐射强迫相一致。我们重点介绍如何通过路径积分公式进行物理耦合，从而产生辐射量（积分 TOA 通量）与光谱参数（气体浓度分数变化）的无偏灵敏度估计。此外，我们强调该方法的可扩展性，证明其与计算机科学加速技术的兼容性。后者在使计算时间对系统的多维和多物理复杂性弱敏感方面发挥着关键作用。