GRAPHICAL ANALYSIS OF THE COMPUTATIONAL METHODS PERFORMANCE IN THE STUDY OF THERMOCHEMICAL PROPERTIES

In this manuscript we have evaluated the computational methods performance in the calculation of thermochemical properties using graphical analysis. The analysis was carried out in a set of 74 organic molecules considering the following thermochemical properties: standard enthalpy of formation, ionization energy and electron and proton affinities. The evaluated methodologies were: G3, G3MP2, G3B3, G3MP2B3, G4 and G4MP2, and the exchange-correlation func- tionals: SOGGA11, RevTPSS, B3LYP, PBE0, B98, M06-2X, BMK and HSE06. Larger basis function sets (G3LargeXP) were employed with these functionals. It was observed that Gaussian-4 theory using reduced order perturbation theory (G4MP2) offered the best relationships between accuracy and computational cost.

本文通过图形分析手段，评估了各类计算方法在热化学性质计算中的性能表现。本次分析以74种有机分子组成的数据集为研究对象，考量的热化学性质包括：标准生成焓、电离能、电子亲和能与质子亲和能。本次评估的计算方法涵盖：G3、G3MP2、G3B3、G3MP2B3、G4以及G4MP2；同时评估的交换相关泛函（exchange-correlation functionals）包括：SOGGA11、RevTPSS、B3LYP、PBE0、B98、M06-2X、BMK与HSE06。上述泛函均搭配大基组G3LargeXP进行计算。研究结果表明，采用低阶微扰理论的高斯4理论（G4MP2）在计算精度与计算成本间实现了最优权衡。