Login / Signup

Reproducing the Reaction Route Map on the Shape Space from Its Quotient by the Complete Nuclear Permutation-Inversion Group.

Hiroshi TeramotoTakuya SaitoMasamitsu AokiBurai MurayamaMasato KobayashiTakenobu NakamuraTetsuya Taketsugu
Published in: Journal of chemical theory and computation (2023)
This study develops an algorithm to reproduce reaction route maps (RRMs) in the shape space from the outputs of potential search algorithms. To demonstrate the algorithm, global reaction route mapping is utilized as a potential search algorithm, but the proposed algorithm should work with other potential search algorithms in principle. The proposed algorithm does not require any encoding of the molecular configurations and is thus applicable to complicated realistic molecules for which efficient encoding is not readily available. We show that subgraphs of an RRM mapped to each other by the action of the symmetry group are isomorphic and also provide an algorithm to compute the set of feasible transformations in the sense of Longuet-Higgins. We demonstrate the proposed algorithm in toy models and in more realistic molecules. Finally, we remark on absolute rate theory from our perspective.
Keyphrases
  • machine learning
  • deep learning
  • neural network
  • high resolution
  • magnetic resonance
  • computed tomography
  • risk assessment
  • contrast enhanced