Machine learning‐assisted study of correlation between post‐transition‐state bifurcation and initial phase information at the ambimodal transition state

• Published in: Journal of physical organic chemistry 2023-11, Vol.36 (11)
• Main Authors: Murakami, Tatsuhiro, Kikuma, Yuya, Ibuki, Shunichi, Matsumoto, Naoki, Ogino, Kanon, Hashimoto, Yu, Takayanagi, Toshiyuki
• Format: Article
• Language: English
• Subjects: Bifurcations; Butadiene; Classification; Cycloaddition; Cyclopentadiene; Diels-Alder reactions; Machine learning; Molecular dynamics
• ISSN: 0894-3230; 1099-1395
• DOI: 10.1002/poc.4561

The Diels–Alder cycloaddition of cyclopentadiene and nitroethene, the intramolecular cycloaddition between a diene and triene, and the Diels–Alder cycloaddition of 2‐hydroxyacrolein with 1,3‐butadiene involving post‐transition‐state bifurcation (PTSB) were studied. These cycloaddition reactions were investigated using quasi‐classical trajectory (QCT), classical molecular dynamics (MD), ring‐polymer molecular dynamics (RPMD) simulations, and supervised machine‐learning binary classification techniques. Room‐temperature dynamics simulations started from the ambimodal transition state (TS) using the QCT, classical MD, and RPMD methods presented similar dynamics. Binary classification revealed that the initial geometry displacement from the ambimodal TS for the Diels–Alder cycloaddition of cyclopentadiene and nitroethene contributed to the branching dynamics and that the initial momenta for the intramolecular cycloaddition between a diene and triene and the Diels–Alder cycloaddition of 2‐hydroxyacrolein with 1,3‐butadiene played a significant role in the bifurcation dynamics.