Research Overview

If you pinch a stress ball for a few seconds and after you let go it almost immediately returns to its original state. On the other hand, pinch your spouse and don't let go for a few seconds and you will likely be hearing about for the next few minutes, hours, maybe even weeks...*

Memory, in the context of dynamical systems, is the property of having a delayed temporal response to an external stimulus. In the example above, we might say that a stress ball exhibits less memory than your spouse, upon the stimulus of being pinched. Memory effects are extremely important and ubiquitous in nature. My PhD studied how memory manifests in the dynamics of several chemical systems of interest, spanning both classical and quantum regimes.  The most significant part of that work was our study of how small molecules couple to surface vibrations of solids. 

*This is purely a humorous example, not a suggestion to start  pinching people and testing their responses. Please respect others' physical boundaries, especially your spouses. 

New developments in equivariant neural network forcefields (EqNNs) have made it possible to do molecular dynamics simulations using quantum chemical methods (typically DFT) with fraction of the normal computational cost and negligible loss in accuracy. 

However, these EqNNs are still slaved to the fidelity of the original quantum chemical method they were parametrized from. I am working on developing quantum-embedding theories which can be used to parametrize EqNNs, but which are more robust than standard DFT. I am particularly interested in the potential application of such methods to challenging open problems in materials science, such as solvent decomposition at the solid-electrolyte interface.