Figure 1
20 odorant molecules belonging to 6 different perceptual classes are first selected, including Benzene in the Aromatic class (left, centre). The Laplacian Matrix (left bottom), a mathematical description of the structure, is shown (for Benzene here) to be proportional to the Dynamical Matrix \({\varvec{D}}\) (right, bottom)—which describes the equation of vibrational motion of a molecule in terms of coupled oscillators. This mapping harmonizes the Vibrational and Shape theories of Olfaction. The Eigen-frequencies of the Dynamical Matrix lead to a vibrational spectrum—the Peak-Decomposed EigenVAlue (PD-EVA)—with peaks corresponding to different vibrational mode types, demarcated by frequency ranges (right, centre)—shown here for Benzene. Clustering based on Similarity in the odorants’ PD-EVA leads to the same classes as perceptual, with the revelation of subclasses within Garlicky and Aromatic. This mapping suggests that vibration-based odour sensing and classification may have the potential to emulate the power of biological olfaction.
