On behalf of our authors, below is the full version of reply we sent to editorial board on 25/04/2024.
-------------------------------------------------------------------------------------------------------
Dear Prof. Gagliardi, First and foremost, we would like to express our sincere gratitude to you for selecting our paper to be featured as an ACS Editors' Choice in addition to being published in Journal of Chemical Theory and Computation. It is truly a great honor, and we are thrilled to have our work recognized in such a prestigious manner. The main purpose of this email is to address the questions raised by Yumiao Ma. When filling out Table 1, we fully trusted the software/program manuals provided by the strong teams such as Gaussian, ORCA, etc. However, for the sake of thoroughness, we conducted tests on some personally developed programs including KST48 mentioned in Table 1. As limited by length of article, we did not include every reason for the assessments included in Table 1. Here, we would like to provide the following point-to-point reply in order to address the three major concerns raised by Ma. 1. About ONIOM We noticed that KST48 had claimed to add ONIOM function several months ago. Then, we downloaded KST48 from GitHub and tested the example provided by KST48 (S0-T1 MECP of CH3CH2NO2 including high/medium/low layers). We found that using the same input file provided by KST48, the calculation job failed in 3 steps due to SCF convergence failure. The relevant calculation files can be found in the attached "oniom_test1.zip". Furthermore, we added “SCF(maxcyc=512,xqc)” keyword into the input file, yet the job still showed no convergence trend within 100 steps. These files can be found in the attached "oniom_test2.zip". As such, we believed that KST48 cannot perform ONIOM calculations although the author of KST48 might have tested the example. This is why we marked “no” for KST48 in ONIOM column in Table 1. 2. About spin-flip TD-DFT Regarding this part, there is no specific explanation in KST48's manual. The author only claims that his/her program can call SF-TD-DFT through "tail1" and "tail2" keywords. That means no one knows how to perform SF-TD-DFT by using KST48. Due to this issue with the program, we were unable to obtain specific results. Even so, we still tried to performed SF-TD-DFT using KST48 based on our knowledge. But the result was not satisfied. Spin contamination cannot be avoided in spin-flip TD-DFT calculation. Therefore, the main practical issue should be carefully treated is to find the targeted states within involved excited states to calculate the gradient. The strategy used by XMECP is actually same as Q-Chem. Taking S0-S1 MECP as an example, the lowest two excited states with the spin contamination lower than threshold (default: 1.20) will be treated as the targeted states, being accepted for the following gradient calculations. However, we found that KST48 cannot do so. Without such strategy, user need to monitor the optimization process time to time in order to ensure the correct states are selected by “root”. This is why ORCA commonly fails in using SF-TD-DFT to optimize MECP. We suggest future development of KST48 should consider this issue. Moreover, we have to point out that KST48 has the command problem when working with ORCA to calculate SF-TD-DFT using "tail1" and "tail2" keywords. The code errors include but may not limit to: 1. Error occurs if the absolute path of ORCA executable file contains “program”. Minimum change is to modify line 285 into: elif “program” in l and “orca_comm” not in l: 2. With above modification, error still occurs since KST48 requires .out suffix for output files, while the running command of ORCA in KST48 generates .log suffix. Based on above tests, we did not further check the source code and we think the claimed capability of SF-TD-DFT with ORCA in KST48 was not tested before its release, and might be just based on Ma’s personal idea. Therefore, we believed that KST48 cannot support SF-TD-DFT and gave a judgment of "no" for KST48 in SF-TD-DFT column in Table 1. All the related files can be found in attached SF-TD-DFT_test.zip. 3. About Gaussian According to the license required by Gaussian Inc., users should not compare Gaussian with other programs in terms of performance. Here “performance” refers to computation speed, accuracy, etc. While the capabilities of Gaussian are not within such aspect. Whether Gaussian is able to perform MECP optimization at any theoretical levels has already been stated in Gaussian user manual. We just summarized the information which is entirely open to the whole theoretical chemistry society. Therefore, we do not agree with Ma that our review contents in the paper have violated any term required by Gaussian license. Our paper is not only an introduction article of XMECP program, but also reports new conclusions reached in several biochemical systems. The marks in Table 1 are not intended to criticize anyone or any programs. We suggest Ma spare more time reading “Results and Discussion” section before making his/her conclusion. We think a good paper, or a good program, should raise the general interest of the whole chemical society, providing the convenient tools and the investigation schemes to readers/users. That’s why we used the complicated multiscale examples in our paper to demonstrate that XMECP is trying to reach “state-of-the-art”. We believe our paper has the potential for broad public interest and will be valuable in related fields. We hope our reply address Ma’s concerns.
Sincerely,
Chunsen Li, on behalf of all authors. 2024.4.25
| 
发表于 Post on 2024-7-18 14:29:48
|
只看该作者 Only view this author
|只看大图 View big picture
|倒序浏览 Reverse view
|阅读模式 Reading model
收藏 Add to favorites
楼主 Author