This is a general-purpose Gaussian approximation potential (GAP [1]) for carbon and hydrogen based materials (CH). The potential is capabe of simulating various materials and molecules composed of C and H elements. The interatomic potential has been fitted with QUIP/GAP [1,2] using an extensive dataset of different configurations at the PBE level of theory [3] using the VASP code [4,5]. The dataset contains following structures : Dimers of carbon and hydrogen Trimers CH containing "soup" structures generated during iterative training QM9[6] molecules augmented to C and H containing molecules only Interactive molecules generated using active learning a-C dataset from [7] Bulk and surface carbon structures CH structures geneated using high pressure This potential includes van der Waals (vdW) corrections at the Tkatchenko-Scheffler (TS) level of theory [8] via a machine learning based local parametrization of dispersion interactions [9].

2024