Assembly theory as a way of defining the biotic/abiotic boundary has been established for molecules, but not yet for crystal structures. This is an assembly algorithm that calculates the complexity of biotic and abiotic minerals in order to constrain the quantitative fundamentals of "life". The calculation utilizes the Hermann-Mauguin space group symmetry and Wyckoff sites of mineral unit cells to calculate the path-building complexity of a crystal structure. 5,644 minerals from the American Mineralogist COD database were run through the algorithm. The five structures with the highest information complexity were a mix of biotic and abiotic minerals, indicating that further calculations on larger datasets would be pertinent. Furthermore, an expansion of the definition of mineral to include biotically synthesized solids would further research efforts aimed at using minerals as possible biomarkers.
This thesis proposes a graph based neural network architecture, SwarmNet, for learning the swarming behaviors of multi-agent systems. Given observation of only the trajectories of an expert multi-agent system, the SwarmNet is able to learn sensible representations of the internal low-level interactions on top of being able to approximate the high-level behaviors and make long-term prediction of the motion of the system. Challenges in scaling the SwarmNet and graph neural networks in general are discussed in detail, along with measures to alleviate the scaling issue in generalization is proposed. Using the trained network as a control policy, it is shown that the combination of imitation learning and reinforcement learning improves the policy more efficiently. To some extent, it is shown that the low-level interactions are successfully identified and separated and that the separated functionality enables fine controlled custom training.