Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 34: Emerging Topics in Chemical and Polymer Physics, New Instruments and Methods I

CPP 34.11: Vortrag

Donnerstag, 21. März 2024, 12:30–12:45, H 0106

Characterisation of the different polymorphs of syndiotactic polystyrene using data-driven methods — •Atreyee Banerjee¹, Tristan Bereau², and Joseph F. Rudzinski^{1, 3} — ¹Max Planck Institute for Polymer Research, Mainz, Germnay — ²Heidelberg University, Germany — ³Physics Department and IRIS Adlershof, Humboldt-Universität zu Berlin, Berlin

Syndiotatic polysterene (sPS) exhibits complex polymorphic behaviour, resulting in rugged free energy landscapes (FELs). Enhanced sampling methods have the potential to remedy to explore the rugged FEL with prior knowledge of collective variables (CVs) that can resolve the relevant transition pathways, typically identified through extensive physical or chemical expertise. Recently, data-driven methods have attracted considerable attention for learning relevant CVs without significant a priori insight. In this work, we adapt an atomic representation used in machine learning to efficiently describe the local environment of sPS monomers. These descriptors do not require the incorporation of excessive system-specific intuition and demonstrate good transferability properties. We then investigate low-dimensional projections from methods of varying complexity: Principal Component Analysis (linear projection), Uniform Manifold Approximation, and Projection (manifold learning). These dimensionality techniques, in conjunction with clustering, provide physically meaningful polymorphic states.

Keywords: Polymer Polymorphism; Data-driven Methods; Principal Component Analysis; Unsupervised Learning Methods; Free Energy Landscape