Core Insights - A systematic breakthrough in the "Kissing Number Problem" has been achieved by a research team from Shanghai Institute of Science and Intelligence, Peking University, and Fudan University, utilizing a multi-agent reinforcement learning system called PackingStar [1][2] - The research has resulted in the discovery of new optimal structures in dimensions 25 to 31, breaking long-standing records in dimensions 14, 17, 12, 20, and 21, and identifying over 6000 new configurations [2] Group 1 - The Kissing Number Problem, proposed by Isaac Newton and David Gregory in 1694, seeks to determine the maximum number of identical spheres that can touch another identical sphere in three-dimensional space [1] - The PackingStar system transforms the high-dimensional packing problem into a cosine matrix filling problem, significantly enhancing computational efficiency and establishing a stable fault-tolerant mechanism for large-scale calculations [1][2] Group 2 - The research team has developed a stable human-AI collaboration model, allowing for the breakdown of large scientific goals into specific projects, thus enabling AI and researchers to work together effectively [3] - This advancement signifies a new exploration path in mathematical research, driven by the integration of artificial intelligence into fundamental scientific fields [3]