Sensor Placement for Tsunami Early Warning via Large-Scale Bayesian Optimal Experimental Design

Collection of arXiv papers (2026-04-13) covering scalable scientific and ML systems research: large-scale Bayesian optimal experimental design for tsunami sensor placement (multi-GPU, Schur-complement greedy OED); Nemo-Nemo, a DAG-based crash-fault-tolerant (CFT) consensus protocol for WANs; Watt Counts, an energy-aware LLM inference benchmark across heterogeneous GPUs; TensorHub and Reference-Oriented Storage for efficient RL weight-transfer; MATCHA for heterogeneous multi-accelerator SoC DNN deployment; lessons from HPL/HPL-MxP on Aurora at exascale; EdgeFlow for fast mobile LLM cold starts;