Update 2024-10-18.md

talks/2024-10-18.md

@@ -1,8 +1,12 @@
 ---
-title: TBD
+title: Data-Driven Prediction of Molecular Crystal Structures Using Mathematical and Topological Descriptors
 author: Nikos Galanakis
 date: 2024-10-18
 category: talk
 ---
 
-TBD
+Predicting the structure of organic molecular crystals is a critical challenge with applications in various industries, from pharmaceuticals to materials science. Traditional prediction methods often rely on complex interaction models, which can be highly sensitive to fine-tuned parameters. In this presentation, we introduce a novel mathematical approach to predict molecular crystal structures without the need for these interaction models.
+
+By leveraging topology and geometric principles, we propose that stable molecular arrangements can be identified by aligning molecular axes and ring plane vectors with crystallographic directions, and positioning heavy atoms at locations that minimize geometric order parameters. This approach translates the problem of crystal structure prediction into an optimization problem, where stable structures are identified by minimizing an objective function. The final structures are then filtered using statistical distributions of free volume and intermolecular contacts from known crystal data.
+
+This approach highlights how molecular crystal structures can be predicted through a purely mathematical framework, providing an alternative to traditional, chemistry-based methods. In this presentation, we will explore the mathematical principles behind the model and discuss its efficiency in predicting stable structures and polymorphs across different compounds.