A small group of scientists in the Information Technology and Materials Research Laboratories have been been applying neural networks to examining a number of problems in chemical physics. One problem, the Kovats retention indices used in gas chromatography, has already been successfully attacked using these approaches (Predicting Kov\'{a}ts Retention Indices Using Graph Neural Networks ). We have achieved an almost fourfold increase in predicitive capabilities of our model based on graph neural networks over previous atom additivity approaches. We are eager to extend these ideas more broadly to predicting mass specta, and the positions and intesities of IR spectral lines. The work has immediate application to the identification of unknown compounds of interest to the larger industrial community. 

References

Chen Qu, Barry I. Schneider, Anthony J. Kearsley, Walid Keyrouz and Thomas C. Allison, Predicting Kov\'{a}ts Retention Indices Using Graph Neural Networks Journal of Chromatogaphy A

Keywords:

artificial intelligence; deep learning; mass spectra; collisions; IR spectra; gas chromatography

Citizenship:  Open to U.S. citizens

Level:  Open to Postdoctoral applicants