Embedded Files
Welcome
Welcome
In the ML4Simulations' Group, led by Dr. Huziel E. Sauceda, we work on the development and application of machine learning-based methods to accelerate predictive materials and molecular systems' simulations. On the machine learning side, the main research topics are the development of physics-based machine-learned force fields (interatomic potentials) and physical time-series interpolation and forecasting. On the application's part, we are interested in understanding the implications of the nuclear quantum effects (i.e. treating atomic nuclei as quantum particles) on the physical properties of materials and molecular systems.
Our group is based in the vibrant Mexico City, at Instituto de Física of the Universidad Nacional Autonoma de México (UNAM).
Latest News
Intercavity polariton dynamics to Nature Commun!
Intercavity polariton dynamics to Nature Commun!
[04/Apr/24] Our paper on the generalization to periodic systems of the force field learning GDML framework was accepted in Nature Communications!
1500 downloads of Bi-LSTM!
1500 downloads of Bi-LSTM!
[18/Aug/22] Our paper on bidirectional-LSTMs as interpolators for complex molecular dynamics trajectories is being very well received! Take a look at it:
Vibrations of fullerenes goes to the European Physical Journal D!
Vibrations of fullerenes goes to the European Physical Journal D!
[18/Jul/22] The fullerene family's vibrational low-frequency behavior still follows a linear dependence on the size. Our article reporting these results was accepted in European Physical Journal D!
BIGDML to Nature Commun!
BIGDML to Nature Commun!
[29/Jun/22] Our paper on the generalization to periodic systems of the force field learning GDML framework was accepted in Nature Communications!
1000 downloads of Bi-LSTM!
1000 downloads of Bi-LSTM!
[22/Jun/22] Our paper on bidirectional-LSTMs as interpolators for complex molecular dynamics trajectories is being very well received! Take a look at it:
Bi-LSTM-driven trajectory interpolation accepted in MLST!
Bi-LSTM-driven trajectory interpolation accepted in MLST!
[30/May/22] Using local-time reversibility in thermostated molecular dynamics trajectories via bidirectional recurrent neural networks, we have achieved unprecedented interpolation accuracy. This work, part of the PhD thesis of Ludwig Winkler, was just published in Machine Learning: Science and Technology.
Page updated
Google Sites
Report abuse