F. Jirasek

The hybrid Enhanced Stokes-Einstein model is extended to concentration-dependent self-diffusion coefficients and multicomponent solvents with HADES, which significantly outperforms benchmark prediction methods.

A new hybrid model predicts electrolyte behaviors for 9,296 systems, overcoming the limitations of traditional models that require extensive experimental data.

Forget complex simulations: this new model predicts liquid diffusion coefficients with state-of-the-art accuracy using only SMILES strings as input.

By combining tensor completion with active learning, this method achieves markedly improved predictions of temperature-dependent diffusion coefficients in mixtures, even with limited experimental data.