编码理论
Search documents
AI与数学“双向奔赴” 中国团队突破亲吻数问题
Zhong Guo Qing Nian Bao· 2026-02-27 01:27
Core Insights - The research team from Shanghai Institute of Science and Intelligence, Peking University, and Fudan University has made significant breakthroughs in the classical mathematical problem of kissing numbers, utilizing the PackingStar reinforcement learning system to explore complex high-dimensional spaces beyond human intuition [1][2] Group 1: Breakthroughs in Kissing Numbers - The team has achieved new best-known kissing number structures in dimensions 25-31, and has broken long-standing records in dimensions 14, 17, 12, 20, and 21 [1] - The PackingStar system transforms high-dimensional packing problems into multi-agent game learning on cosine matrices, allowing for a unified approach to complex geometric issues [3] Group 2: Applications and Implications - Kissing number problems are not just abstract geometric challenges; they are central to discrete geometry and coding theory, with applications in satellite communication, quantum coding, and data compression [2] - The advancements in this area could lead to improved signal distribution methods in engineering, highlighting the practical significance of the research [2] Group 3: Research Methodology and Efficiency - The research has resulted in a 2-3 times increase in search efficiency and saved over 100,000 GPU hours, showcasing the computational potential of the AI model [3] - The methods developed through the PackingStar project have become a reusable interdisciplinary computational paradigm, enabling systematic exploration of previously deemed "intractable" scientific problems [3] Group 4: Team and Institutional Support - The project is led by young researchers, including AIMath researcher Ma Chengdong, who aims to tackle more niche and high-risk problems after achieving existing academic results [3][5] - The Shanghai Institute of Science and Intelligence promotes an inclusive environment for young researchers, encouraging independent exploration without hierarchical constraints [5]
盟云全息股价近期波动,探索区块链技术应用
Jing Ji Guan Cha Wang· 2026-02-11 13:32
Group 1 - The stock price of Alliance Hologram (HOLO.OQ) has shown volatility over the past week, with a cumulative decline of 4.27% and a fluctuation range of 11.97% [1] - On February 4, the stock price fell by 4.70% to close at $2.23; on February 5, it continued to drop by 6.73% to $2.08; on February 6, it rebounded by 5.77% to $2.20; on February 9, it slightly decreased by 0.91% to $2.18; and on February 10, it increased by 2.75% to $2.24, with a trading volume of approximately $1.72 million and a volume ratio of 1.26 indicating a recovery in trading activity [1] - During the same period, the electronic components sector declined by 1.80%, and the Nasdaq index fell by 0.59%, indicating that the company's stock price volatility was greater than that of the sector and the broader market [1] Group 2 - Alliance Hologram announced the exploration of coding theory applications in the field of blockchain scalability, utilizing sharding technology and coding symbol processing to reduce node storage burdens, enhance system fault tolerance, and improve communication efficiency [2] - This technological direction aims to address challenges posed by the growth of blockchain data and may provide technical support for the company's business areas such as digital twins [2]