?
Postdoctoral Fellow
Research group |?Condensed Matter Physics
Main supervisor |?Anders Malthe-S?renssen
Co-supervisor |?-
Affiliation |?Department of Physics, UiO
Contact |?eevargas@uio.no
Short bio
I started at the Department of Chemical Engineering, Universidad de Concepción (Chile), where I pursued my doctoral thesis in Engineering Sciences with co-supervision from the Department of Mechanical Engineering at the Technical University of Denmark. I investigated the water capillary flow in silica nanochannels and proposed a Brownian water-pump using carbon nanotubes based on MD simulations.?
Then, I held a postdoctoral position at Swinburne University of Technology, Department of Computing Technologies, Australia, conducting analyses of CO2 and water with MD simulations, evaluating the influence of quantum corrections on thermodynamic properties. Later, I moved to Denmark, to work as a postdoc under the supervision of Frank Jensen at Aarhus University. This project focused on the development of a charge-flow polarization model, transferring information from ab initio computations to classical MD simulations.?
Currently, I am a postdoctoral researcher at UiO under the DSTrain MSCA fellowship, working at the Njord Centre, Department of Physics.
Research interests and hobbies
As a researcher, my motivation is to contribute to the understanding of intermolecular interactions, particularly at fluid-solid interfaces. This endeavor is rooted in elucidating the underlying chemical description of matter to predict macroscale behaviors with applications in engineering. I am interested in the confluence of statistical mechanics, chemistry, and fluid dynamics. I like to write my own MD routines using Fortran or C++ with MPI parallelization.
In my spare time, I like to spend time with my family and enjoy music.
DSTrain project
Towards Accurate Modelling of Water-Mineral Interfaces: Development and Validation of Machine Learning Potentials for Atomistic Simulations of Geological Systems.
Fluid-rock interactions play a fundamental role in the physicochemical properties of Earth's lithosphere, linking chemical reactions with the transport of mass and energy. Limited access to deep lithosphere samples makes computational simulations a valuable alternative for investigating these processes. Molecular Dynamics (MD) simulations offer atomic-scale insights at conditions comparable to those found in the Earth's lithosphere. However, to describe their interactions, researchers usually employ simplified force fields due to the high computational costs associated. In the last decade, Machine Learning Potentials (MLP) have been introduced, capable of describing the potential energy surface of a multiatomic system upon training from ab initio computations, such as density functional theory (DFT) methods. Consequently, MLPs have emerged as an alternative to force field methods, offering improved predictive capabilities—comparable to DFT—at computational costs similar to classical methods. This research proposal aims to demonstrate how advancements in computational modelling, such as MD simulations using MLPs, are essential for understanding geophysical and geochemical processes at atomic and nanoscale levels. The trained MLPs will be evaluated in large-scale systems, and key properties relevant to geoscience will be considered. This interdisciplinary research integrates chemistry, geoscience, nanofluidics, and computational science to provide enhanced accuracy in the modelling of fluid-solid interactions relevant to understanding the processes in Earth's lithosphere.
Publications
DSTrain publications
?
Previous publications
?
?