AI & ML Research 24 Hours

Last updated: July 26, 2026, 8:30 AM ET

AI & ML Research

Researchers explored methods for optimizing vector search infrastructure, specifically addressing the rising costs of RAM. Discussions covered the trade-offs between on-disk and in-memory Approximate Nearest Neighbor (ANN) indexes, including HNSW, SPANN, and Disk ANN, to achieve cost-effective solutions for large-scale vector databases navigated trade-offs. Concurrently, a novel approach to fluid simulation was presented, generating visual phenomena like the Kármán vortex street without directly solving traditional fluid equations, instead leveraging the Lattice Boltzmann Method and implemented in C++ simulated fluid.