Gaussian Approximation Potentials Sample Clauses

Gaussian Approximation Potentials. ‌ One of the ML potentials is the Gaussian Approximation Potential (GAP) developed by Xxxxxx and his collaborators [9] and they are specifically implemented using the Gaussian process inference of the PES surface. In the scenario of a GAP, the total energy Etotal is constructed to be sum of the atomic energy {ϵi} centred on each of the atoms and the function relation between atomic energy and atomic environments are the underlying target for the ML techniques to address. For localised bonding, such as covalent bonds, this local energy concept is justified. For long-range interactions, such as the Coulomb or dispersion, these terms are usually separated from the local energies {ϵi} to assure a much better localisation for the ML to perform [9]. Σ Natoms total E = i=1