AI 中文总结
本研究对比了苯系烃的求和连通性（SGO）与乘积连通性（PGO）Gourava指数，发现SGO在π电子能量预测、退化性等验证中表现更优，适用于QSPR建模等理论研究。
AI 中文摘要
本研究评估了求和连通性（$SGO$）与乘积连通性（$PGO$）Gourava指数作为苯系烃的分子描述符。利用包含30个苯系结构的数据集，我们比较了用于预测π电子能量（$E_{\rm{\text{π}}}$）的最小二乘回归模型，发现$SGO$在各类分子边类型上的拟合效果显著优于$PGO$。该指数还通过三类验证设计进一步评估：（i）相关性分析，其中$SGO$与标准描述符$M_1、M_2、SO、DSO$及$ABS$呈强但非完全的负相关，相关系数$r$在[-0.9923,-0.8936]范围内，表明其包含互补的结构信息；（ii）辛烷、壬烷及10阶树数据集的退化性分析，$SGO$的退化率分别为22.22%、40.00%和42.45%；（iii）10阶树的结构敏感性分析，其敏感性较$DSO$高74%，同时保持较高的结构突变率（$SA = 0.474386$）。总体而言，$SGO$在判别能力与数值稳定性间取得了良好平衡，支持其在QSPR建模及相关理论研究中的适用性。
英文摘要
This study evaluates the sum-connectivity ($SGO$) and product-connectivity ($PGO$) Gourava indices as molecular descriptors for benzenoid hydrocarbons. Using a dataset of 30 benzenoid structures, we compare least-squares regression models for predicting $π$-electronic energies ($E_π$) and find that $SGO$ yields a markedly better fit than $PGO$ across molecular edge types. The indices are further assessed using three validation designs: (i) correlation analysis, in which $SGO$ exhibits strong yet non-perfect inverse correlations with standard descriptors ($M_1, M_2, SO, DSO,$ and $ABS$; $r\in[-0.9923,-0.8936]$), suggesting complementary structural information; (ii) degeneracy analysis on Octane, Nonane, and order-$10$ tree datasets, where $SGO$ attains low degeneracy rates (22.22\%, 40.00\%, and 42.45\%); and (iii) structure-sensitivity analysis on trees of order $n=10$, showing 74\% higher sensitivity than $DSO$ while maintaining a high structure-abruptness ratio ($SA = 0.474386$). Overall, $SGO$ offers a favorable balance between discriminative power and numerical stability, supporting its applicability in QSPR modeling and related theoretical studies.
Comments15 pages, 6 figures
