This three-year, $1.1 million project grant from the National Science Foundation's Computer and Information Science and Engineering program aims to develop highly scalable and sample-efficient spectral methods for learning graph topologies from high-dimensional data samples. The grantee, the Trustees of the Stevens Institute of Technology, will investigate spectral graph densification frameworks to efficiently estimate attractive Gaussian Markov random fields. A unique feature of the learned graphs will be that effective resistance distances encode sample similarities in the original data. This research is expected to advance scalable, data-driven physics-informed algorithms for modeling, simulation, optimization and verification of integrated circuits. Successful outcomes may also contribute to dimensionality reduction, scientific computation, data visualization and machine learning. The grant reflects NSF's mission to support merit-reviewed projects advancing computing and information sciences through September 2025.