Title: Machine learning for phase diagrams of complex fluids

Speaker: Prof. Arun Yethiraj (UW-Madison, USA)

Abstract: Machine learning (ML) has become an important tool in computational chemistry. This work describes the use of supervised and unsupervised ML methods to obtain the phase diagram of complex fluids from computer simulations. A convolutional neural network approach, based on a grid interpolation of particle positions, successfully predicts the phase behavior of off-lattice systems, e.g., the Widom-Rowlinson mixture and symmetric polymer blends. The method is too computationally intensive, however, for more complex polymeric systems. A deep neural network approach, based on structural and thermodynamic input from computer simulations predicts the phase diagram of polymers in ionic liquids. The disadvantage of supervised methods is that they require some knowledge of the phase diagram for training purposes. Unsupervised methods based on principal component analyses can predict the phase diagram of off-lattice systems, without any prior knowledge. The method is, however, sensitive to the choice of the input features. ML methods are therefore an attractive route to obtaining phase diagrams via computation, but physical insight plays an important role in their implementation.