研究目的
To propose an automatic approach for analyzing on-the-fly trajectory surface hopping simulation results in multi-channel nonadiabatic photoisomerization dynamics by considering trajectory and configuration similarities, using the phytochromobilin chromophore model as an example.
研究成果
The proposed analysis protocol successfully identifies multiple reaction channels in nonadiabatic photoisomerization dynamics, extracts major molecular motions, and determines branching ratios without prior knowledge. It is a powerful tool for automatic analysis of trajectory-based simulations, with potential applications to other types of nonadiabatic dynamics.
研究不足
The approach may not work well in systems with very complex molecular motions or when trajectories do not form clear clusters. The Fréchet distance is an approximation and may not capture all aspects of trajectory similarity. Computational cost and memory requirements can be high for large numbers of trajectories.
1:Experimental Design and Method Selection:
The study uses trajectory surface hopping (TSH) dynamics combined with Fréchet distance for trajectory similarity and multidimensional scaling (MDS) for dimensionality reduction. DBSCAN clustering is employed for trajectory grouping.
2:Sample Selection and Data Sources:
The phytochromobilin (PΦB) chromophore model is used, with trajectories generated from TSH simulations at the semiempirical OM2/MRCI level. Initial conditions are sampled using Wigner distribution.
3:List of Experimental Equipment and Materials:
Computational simulations are performed using the JADE code interfaced with MNDO package for quantum chemistry calculations. A homemade FORTRAN code is used for RMSD calculations, and Python with Scikit-learn toolkit is used for data analysis.
4:Experimental Procedures and Operational Workflow:
Trajectories are propagated up to 1 ps, with snapshots taken at specific time intervals. Fréchet distance is computed for trajectory pairs, followed by MDS and DBSCAN clustering to identify reaction channels. Dimensionality reduction is applied to geometries within clusters to extract major motions.
5:Data Analysis Methods:
Statistical analysis includes Fréchet distance calculation, MDS for dimensionality reduction, DBSCAN for clustering, and visualization of geometric evolution using key coordinates.
独家科研数据包,助您复现前沿成果,加速创新突破
获取完整内容