Case Studies of Dimensionality in Chemical Data

The ambition to relate intrinsic features of chemical data to the underlying chemical reaction networks (CRNs) is not new, but has experienced only modest success. This may partly be attributed to a lack of theoretical groundwork connecting idealized theory to actual experimental data with added complexity. In particular: i) many CRNs have species that cannot be directly observed experimentally; ii) the apparent number of underlying reactions is a function of the resolution of the data; iii) chemical phenomena can change the number of discernable independent processes of the data. In this work, we illustrate the application of the recently introduced concept of data dimension, which quantifies the linearly independent dimensions the system composition in a CRN can change. We perform case studies inspecting the dimensionality of chemical data characterizing CRNs, and outline how it can be used for mechanistic interpretation. In some instances, these extended considerations allow us to directly recover the CRN proposed in the literature without any fitting. This demonstrates that, with incomplete information, important clues about CRN structure can still be recovered. Additionally, our approach detects critical subtleties, preventing important candidate reactions from being discarded in mechanistic studies.