Water, acetonitrile, and methanol MD simulations driven by many-body ML potentials

<p>Input, output, and trajectories of molecular dynamics (MD) simulations of water, acetonitrile, and methanol. Simulations were driven by many-body machine learning (mbML) potentials including explicit 1-, 2-, and 3-body contributions. <a href="https://keithgroup.github.io/mbGDML/">GDML</a>, <a href="https://libatoms.github.io/GAP/">GAP</a>, and <a href="https://schnetpack.readthedocs.io/en/stable/">SchNet</a>&nbsp;models are provided in a <a href="https://doi.org/10.5281/zenodo.7112163">separate repository</a>. All simulations were performed in the <a href="https://wiki.fysik.dtu.dk/ase/">atomic simulation environment (ASE)</a>.&nbsp;Analyses including radial distribution function (rdf) curves are provided <a href="https://github.com/keithgroup/mbgdml-h2o-meoh-mecn">here</a>.</p> <p><strong>Manifest</strong></p> <p>The following simulations are included in this repository for each solvent.</p> <ul> <li>1 ps hexamer NVE MD simulation driven by MP2/def2-TZVP (in ORCA v4.2.0), mbGDML, mbGAP, mbSchNet, and GFN2-xTB started with the same positions and velocities. Velocities were initialized at 298.15 K with a Maxwell-Boltzmann distribution.</li> <li>Periodic NVT MD simulation at 298.15 K for 10 or 30 ps with a 1 fs time step driven by mbGDML. These simulations contained <ul> <li>58 or <strong>137</strong> water molecules,</li> <li><strong>67</strong> or 122 acetonitrile molecules,</li> <li><strong>61</strong> methanol molecules.</li> </ul> </li> </ul> <p>Systems that are not bolded were used for testing purposes.</p>