A framework for integration of empirically measured molecular structure, as inferred by moment of inertia, and <i>ab initio</i> structure prediction is presented to enable recovery of molecular structure representative of empirical conditions. Structural elucidation is a ubiquitous endeavor in chemistry, from crystal structure determination to molecular structure and conformational studies. As systems of interest grow larger and more complex, predicting structure ab initio and finding agreement between theoretical results and empirical structures becomes more challenging. Multi-agent consensus equilibrium (MACE) is a method which enables the fusion of multiple models of the same system. MACE is a flexible formalism which computes the equilibrium point between several models, known as agents, which are optimized simultaneously; the returned state description (e.g., atomic coordinates for molecular structure) represents the intersection of each manifold and is not equivalent to the average optimum state for each agent. Using MACE, any two descriptions of the same system can be fused. The moment of inertia, determined directly from microwave spectroscopy measurements, provides a convenient model with which to test the application of MACE to quantum chemical modeling. MACE results reported combine gradient descent optimization of each ab initio agent with an agent that predicts chemical structure based on root-mean-square deviation of the predicted inertia tensor with experimentally measured moments of inertia. The resulting structure more accurately resembles empirical results than pure computational chemistry approaches yet retains structural effects from accounting for quantum mechanics.

Learning from computations and experiments simultaneously: Multi-agent consensus equlibrium (MACE) for improving chemical structure determination

ACS Fall 2020 Virtual Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.