ИСТИНА

During the last decades, computer simulations have become an indispensable part of studies of mesoscopic liquid crystal (LC) systems, such as droplets, channels, and layers. Computer simulations offer knowledge about the three-dimensional structure of a studied object. In many cases, finding it corresponds to an optimization problem. There is no ‘best’ method for any optimization problem. In general, the roughness of the energy profile and dimensionality of the system are the two crucial points for the choice. Liquid crystal systems are characterized by a very rough energy landscape with plenty of local minima, particularly for cholesteric systems. Moreover, experiments often deal with metastable states with a short lifetime. Thus it is practically essential to be able to scan the energy landscape and find different structures corresponding to the most probable metastable states. It makes Monte Carlo approach a good candidate. Here we present the application of the Stochastic Approximation Monte Carlo (SAMC) method to the liquid crystal simulations problem. SAMC was proven to be an extremely efficient method to obtain energy minima, discover energy landscape profiles and metastable states in molecular and polymer systems. We demonstrate the effective implementation of SAMC on massive-parallel GPU architectures. Furthermore, we show its usability with the extended Frank Elastic Continuum method to study systems with multiple metastable states on examples of nematic and cholesteric LC droplets. At the same time, the approach and its implementation are not limited by Elastic Continuum theory and could be used with other methods too, for example, with Landau – de Gennes one.