As deep learning, learning algorithm based on multilayer neural networks, which can figure out regularities in huge data (extracted from experiment), and express them in equations; is developing. How would this development affect computational physics? Will we still be using methods like first-principles calculation etc. as often?
Dear Abbas, I think this paper could be useful for your analysis of relations between Neural Networks and Computational Chemistry and Physics. Gianluca
G. Montovan, Neural Networks for Computational Chemistry: Pitfalls and Recommendations, 2013
So far, first principles provide very accurate result in the atomic level. Otherwise the neural network method can produce better accuracy than first principles calculation, the people will still use first principle calculation.
- Badges
- Science topic