|
|
各位老师好，我是动力学模拟小白，最近涉及到模拟化学反应过程，该过程设计到化学键的断裂与生成，这种情况根据社长的博文，应该使用AIMD。我调研了一些文献，发现Gaussian中的BOMD和orca中的AIMD似乎都可以模拟这个过程，但是我在具体操作过程中遇到很多的问题，查阅了论坛也找不到合适的解答，现在发帖求助，可能问题很初级，请各位老师解答下，谢谢。
（1）首先，我先把调研的文献描述贴出来，第一篇(ACS Catal. 2021, 11, 6816−6824)的计算方法描述：Quasiclassical trajectories (QCTs) were simulated with direct MD at the level of SMD-M06-L/BS-I by interfacing Singleton’s ProgDyn code with G09. All direct MD simulations were performed at 298.15 K to reproduce reaction conditions. A pruned (99, 590) grid (G09 keyword: int = ultrafine) was applied to improve the accuracy of energy, energy gradient, and vibrational frequency evaluations. For each ambimodal TS (reoptimized with the more accurate ultrafine grid), a total of 100 initial configurations were
generated by normal-mode sampling. Along forward and backward directions, molecular configurations and velocities were propagated using the velocity Verlet algorithm with a timestep of 1 fs until the starting material or product is formed. Thresholds for bond formations and trajectory terminations are listed in Table S4.
我关心的是，他使用的产生100个初始结构的normal-mode sampling，是如何使用以及随后的velocity verlet算法如何实现计算，因为按照我的理解，他从过渡态开始进行动力学模拟？这100个initial configuration是100个不同构型的过渡态？还是说给予相同的过渡态100个不同的初始速度？我个人倾向于后者的解释，这样对么？就是给予相同过渡态100个不同的初始速度，就可以跑出100条不同的轨迹。
第二篇文献（Chem 4, 1952–1966）的计算方法描述：Quasi-classical Born-Oppenheimer molecular dynamics simulations were performed with the ‘‘BOMD’’ keyword implemented in Gaussian09 package. The initial coordinates and momenta of all the trajectories were generated from the normal mode sampling (at 300 K) from the transition state TS2-B or TS2-A optimized at the PBE1PBE/6-31G** level of theory. From each set of initial coordinates and momenta, a trajectory was propagated with the classical equations of motion with energies and forces computed with PBE1PBE/6-31G** in THF (SMD). Trajectory propagation was performed in both forward and reverse directions until either C2–C8 or C3–C8 bond was fully constructed or a total time of 2 ps was reached.
相同的，他使用Gaussian 09产生了100个trajectory，也是说给予过渡态100个不同的初始速度么？这个使用Gaussian 09应该如何实现？我翻阅gaussian 09的手册，似乎可以手动添加初始速度，但是他说的normal mode sampling是怎么实现的？
（2）Gaussian 09的BOMD计算中似乎每一步之后都去算Hessian矩阵和Force constant，这对于大体系来说，也不能接受吧？毕竟如果模拟时间到了2ps，按照每一步1fs来计算，每一步计算Hessian这样的计算量也太大了吧？似乎gaussian 09可以设置每隔几部之后再去更新hessian，但是我还是没用明白。随后我阅读了社长写的orca做AIMD的文章，我使用了ORCA 4.2跑了一下，但是我发现orca也没有产生initial coordinates and momenta的关键词啊？这实在对我来说是太困惑了，而且我发现，我设置orca跑2000步，它跑了800多步就自动结束，是平衡了么？
|
|
发表于 Post on 2021-8-23 10:56:04
收藏 Add to favorites
楼主 Author
发表于 Post on 2021-8-23 15:37:18