From data to interpretable models: machine learning for soil moisture forecasting.

Soil moisture is critical to agricultural business, ecosystem health, and certain hydrologically driven natural disasters. Monitoring data, though, is prone to instrumental noise, wide ranging extrema, and nonstationary response to rainfall where ground conditions change. Furthermore, existing soil moisture models generally forecast poorly for time periods greater than a few hours.

Interaction of Hydrogen with MB (M = Ba, Ca, La, and Sr) Surfaces from First Principles.

We show results of basic energetics and interacting behavior of hydrogen with metal hexaboride surfaces using a combination of self-consistent density functional calculations and dynamics based on the Car-Parrinello method. Our results show that hydrogen is strongly attracted to localized exposed boron atoms and interactions with the terminal cations are strictly repulsive. From these, preliminary local adsorption energy calculations suggest that a single hydrogen molecule per surface unit-cell is possible (one ML).

A generalized method for the inversion of cohesive energy curves from isotropic and anisotropic lattice expansions.

Cohesive energy curves contain important information about energetics of atomic interactions in crystalline materials, and these are more often obtained using ab initio methods such as density functional theory. Decomposing these curves into the different interatomic contributions is of great value to evaluate and characterize the energetics of specific types of atom-atom interactions. In this work, we present and discuss a generalized method for the inversion of cohesive energy curves of crystalline materials for pairwise interatomic potentials extraction using detailed geometrical descriptions of the atomic interactions to construct a list of atomic displacements and degeneracies, which is modified using a Gaussian elimination process to isolate the pairwise interactions.