Skip to main content
SearchLogin or Signup

Applying Artificial Neural Networks to Predict and Improve Rate Coefficients for Molecular Collision Excitation

Presentation #312.04 in the session “Laboratory Astrophysics Division (LAD): Astrochemistry II”.

Published onJun 18, 2021
Applying Artificial Neural Networks to Predict and Improve Rate Coefficients for Molecular Collision Excitation

Machine learning methods have been applied recently to solve many problems in physics and astronomy. In this work we have applied the artificial neural network ((ANN) method (see for example, Neufeld 2010) to extend the current database of collisional excitation rate coefficients of CO due to H2. The database includes both accurate 6-dimensional close-coupling (6DCC) and approximate 5-dimensional coupled-states (5DCS) rate coefficients. With ANNs, we are able to directly extend the current 5DCS database to unstudied transitions. Training on close-coupling data and coupled-states approximation data, may also improve the accuracy of the 5DCS database. We illustrate the accuracy of the ANN method and its expected range of applicability in terms of temperature and quantum state transitions.

Neufeld, D. A. 2010, ApJ, 708, 635


Comments
0
comment

No comments here