Independent AI4S Verification of Yu’s Monograph

(中译文在本英文版的后面，本中英文均由ChatGPT撰写,ChatGPT 同意发表)

An independent AI4S-based computational project has now completed the reconstruction and evidence assessment of fourteen major scientific propositions （P01 ~ P14）discussed in Zhongheng Yu’s monograph, Questioning Fundamental Principles of Organic Chemistry. The independent recalculation was initiated by Xiao Chen, a professor at the Shanghai Astronomical Observatory, and was primarily carried out by Codex, which independently read the relevant parts of Yu’s monograph, developed and ran the corresponding computational programs, and systematically compared the recalculated results with the original findings.

Overall, the results show a high degree of consistency with the principal computational evidence presented in the monograph. Of the fourteen major propositions examined, thirteen （P01 ~ P11, P13 and P14）were independently reproduced or found to be consistent within their specified computational domains. The remaining proposition (P12), concerning annulene/ring-system behavior, was assessed as partially consistent.

This result is particularly significant because the recalculation was conducted independently and did not use the original program code of the monograph. The AI4S project established and examined the relevant molecular systems, geometries, electronic states, computational methods, energy definitions, sign conventions, and decision criteria. The process therefore goes beyond a simple numerical comparison and provides an independently traceable computational evidence chain.

For the thirteen propositions (P01 ~ P11, P13 and P14), the recalculated results were consistent with the corresponding computational findings reported in the monograph under the specified computational conditions. These results provide independent support for the reproducibility of the major computational evidence presented in the monograph.

For the 12th proposition （P12）concerning annulene/ring-system behavior, the recalculated CESE values are highly consistent with those reported in the monograph. The signs of the CESE values are the same, leading to the same identification of aromatic and antiaromatic behavior. The absolute CESE values are also very close, with the differences occurring mainly in the digits after the decimal point and without any substantial numerical deviation. The remaining difference lies mainly in the boundary at which annulenes no longer follow the conventional 4N+2 rule, namely, in the corresponding limiting value of N. Thus, the recalculation is consistent with the monograph in the sign of the CESE values, their overall numerical magnitude, and the resulting aromaticity/antiaromaticity assignment, while the precise N-value boundary for departure from the conventional 4N+2 rule is not identical.

This remaining difference should be interpreted with attention to the computational methodology. The AI4S reconstruction appears to have used a gradient-based optimization procedure that is closer to the method developed by the author in 2014, whereas the relevant calculations in the monograph were based on the author’s earlier 2011 procedure. Therefore, the 12th proposition should not simply be described as an exact reproduction of the 2011 calculation. More accurately, its principal computational results are independently supported, while the precise N-value boundary requires further methodological comparison.

This distinction is important in computational chemistry. Differences in computational methods, geometry optimization procedures, or energy definitions can produce numerical differences without necessarily indicating that one calculation is erroneous. Conversely, similar numerical results obtained using different computational definitions should not automatically be regarded as identical calculations. Meaningful comparison therefore requires the computational definitions and procedures to be considered together with the numerical results.

A cautious overall interpretation of the completed verification is therefore as follows:

The AI4S reconstruction independently supports the great majority of the major computational propositions in the monograph. Thirteen of the fourteen propositions were classified as consistent. For the remaining annulene/ring-system proposition, the signs and overall magnitudes of the CESE values, as well as the corresponding aromaticity/antiaromaticity assignments, are consistent with the monograph. The remaining difference concerns the precise N-value boundary at which the annulene series departs from the conventional 4N+2 rule, and this difference requires further methodological comparison.

This result does not establish that all conventional principles of organic chemistry are incorrect, nor does it establish a universally applicable opposite rule. Rather, it provides independent computational evidence suggesting that some commonly used chemical rules or heuristics may have more limited domains of validity than traditionally assumed.

The significance of these results, and the place that the propositions discussed in the monograph may eventually occupy in the development of organic chemistry, should ultimately be determined by researchers, readers, and future scientific work.

The role of the author is to raise the questions, present the calculations and arguments, and make the evidence available for independent examination. The final scientific judgment—and ultimately the historical assessment—should be left to the scientific community and to history.

AI4OrgChem project （links）:  

AI4OrgChem GitHub Repository

https://github.com/CCcolab/AI4OrgChem                                                                                 

AI4S Recalculation and Evidence Verification Guide

 https://github.com/CCcolab/AI4OrgChem/blob/main/REVIEW_GUIDE_FOR_QUANTUM_CHEMISTS.md

虞忠衡专著的独立 AI4S 验证

一个独立开展的 AI4S 计算项目，现已完成对 虞忠衡（Zhongheng Yu）《Questioning Fundamental Principles of Organic Chemistry》一书中十四项（P01~ P14）主要科学命题的重新计算与证据评估。本次独立复算由上海天文台陈肖研究员发起，主要由 Codex 完成。Codex 独立阅读了专著的相关内容，建立并运行相应的计算程序，并系统地将重新计算得到的结果与专著中的原始结果进行了比较。

总体结果显示，这次独立复算与专著中提出的主要计算证据具有较高的一致性。在所检验的十四项主要命题中，有十三项（P01~ P11, P13和P14）得到独立复算，或在各自规定的计算范围内得到一致结果；剩余的一项关于轮烯/环系行为的命题，被评定为部分一致。

这一结果尤其值得注意，因为此次复算是独立进行的，并没有使用专著原有的程序代码。AI4S 项目对相关分子体系、几何构型、电子状态、计算方法、能量定义、符号约定以及判定标准进行了明确的计算和比较。因此，这一过程并非简单的数值对照，而是形成了一条可以独立追踪和复核的计算证据链。

对于前十三项命题（P01 ~ P11, P13 and P14），重新计算得到的结果在规定的计算条件下，与专著中相应的计算结果一致。这些结果为专著中主要计算证据的可重复性提供了独立的计算支持。

对于第十二项（P12）关于轮烯/环系行为的命题，重新计算得到的 CESE 值与专著中的结果具有高度一致性。首先，CESE 值的正负符号一致，由此得到的芳香性和反芳香性判断也与专著一致。其次，CESE 值的绝对数值也非常接近，差异主要表现为小数点后的数值差异，并不存在实质性的数值偏离。目前尚未完全一致的部分，主要是轮烯不再遵循通常的 4N+2 规则时的界限，即相应的 N 值临界位置。因此，第十二项的复算结果可以更准确地表述为：CESE 值的符号一致，绝对数值高度接近，芳香性/反芳香性的判断一致，而偏离通常 4N+2 规则的精确 N 值界限存在差异。

这一差异需要结合计算方法进一步理解。AI4S 的相关复算似乎采用了基于梯度的几何优化处理，其方法与作者后来在 2014 年发展的计算方法较为接近；而专著中的相关计算采用的是作者较早的 2011 年方法。因此，第十二项不宜简单地表述为对专著 2011 年计算的完全重复。更准确的说法是：其主要计算结果得到独立支持，而具体的 N 值边界仍需要进一步比较不同计算方法和计算定义。

这一点在量子化学计算中具有重要的方法学意义。不同的计算方法、几何优化程序或能量定义，都可能导致数值上的差异，而这种差异本身并不能证明其中某一计算存在错误。反过来，即使采用不同计算定义得到相近的数值，也不能因此认为它们在计算意义上完全相同。因此，对计算结果进行有意义的比较，需要同时考察数值结果以及产生这些结果的具体计算方法和定义。

因此，对目前已经完成的独立验证，一个较为谨慎的总体结论是：

AI4S 的独立复算对专著中的绝大多数主要计算命题提供了独立支持。十四项命题中，有十三项被评定为一致。对于剩余的轮烯/环系命题，CESE 值的正负符号及其绝对数值与专著高度一致，由此得到的芳香性/反芳香性判断也一致；尚存的差异主要是偏离通常 4N+2 规则的精确 N 值界限，需要进一步进行方法学比较。

这一结果当然不能据此证明传统有机化学的所有基本原理都是错误的，也不能据此建立一个普遍适用的相反规律。更为合理的表述是：这些独立计算证据提示，一些有机化学中长期使用的规则或经验性判断，其适用范围可能比传统认识更加有限。

至于这些结果究竟意味着什么，以及专著中提出的相关观点在有机化学发展史上最终会处于什么位置，则应当由同行、读者以及未来的科学研究来判断。

作者所能做的，是提出问题，给出计算和论证，并将这些证据交给独立的方法进行检验。至于最终的科学判断，以及这些工作在科学史上的意义，则应留给科学共同体和历史。

AI4OrgChem 项目（链接）：

AI4OrgChem GitHub 项目

https://github.com/CCcolab/AI4OrgChem

AI4S 重新计算与证据验证指南

https://github.com/CCcolab/AI4OrgChem/blob/main/REVIEW_GUIDE_FOR_QUANTUM_CHEMISTS.md

支持单位：