On the example of 2-, 4- and 2,4-thiouracil we demonstrate the performance of the RASPT2/RASSCF protocol in reproducing the spectral positions and line shapes of linear absorption spectra that have been recorded in water and documented in this work. Through a QM/MM scheme coupled to a room-temperature Wigner sampling we simulate condensed phase spectra, permitting to compare our results against experiments. We discuss the sensitivity of the simulations to: a) the active space size by pushing the limits beyond the full-valence active spaces; b) the consideration of “dark” nπ^∗-states in the state averaging; c) the flavor of RASPT2 technique; d) the basis set. The benchmarking demonstrates that full-π valence active spaces tend to red-shift the absorption band. Increasing the active space rectifies the problem and we obtain near-quantitative agreement between our experiments and calculations. We, furthermore, demonstrate that the choice of RASPT2 flavor has to be made through rigorous benchmarking.

在2-，4-和2,4-硫尿嘧啶的例子中，我们证明了RASPT2 / RASSCF协议在再现已记录在水中并记录在这项工作中的线性吸收光谱的光谱位置和线形方面的性能。通过将QM / MM方案与室温Wigner采样耦合，我们可以模拟凝聚相谱，从而可以将我们的结果与实验进行比较。我们讨论了仿真对以下方面的敏感性：a）通过将极限推超出全价有源空间来限制有源空间的大小；b）考虑“暗”nπ ^∗-状态平均中的状态；c）RASPT2技术的风格；d）基础集。基准测试表明，全π价活性空间倾向于使吸收带发生红移。增加活动空间可以解决该问题，并且我们在实验和计算之间获得了接近量化的协议。此外，我们证明必须通过严格的基准测试来选择RASPT2风味。