本文印刷
全画面プリント
東京確率論セミナー
過去の記録 ～05/02｜次回の予定｜今後の予定 05/03～
| 開催情報 | 月曜日 16:00～17:30 数理科学研究科棟(駒場) 126号室 |
|---|---|
| 担当者 | 佐々田槙子、中島秀太（明治大学）、星野壮登（東京科学大学） |
| セミナーURL | https://sites.google.com/view/tokyo-probability-seminar23/ |
2024年11月25日(月)
16:00-17:30 数理科学研究科棟(駒場) 126号室
15:15〜 ２階のコモンルームでTea timeを行います。ぜひこちらにもご参加ください。
Leon Frober 氏 (Grand Valley State University)
Free energy and ground state of the spiked SSK spin-glass model
15:15〜 ２階のコモンルームでTea timeを行います。ぜひこちらにもご参加ください。
Leon Frober 氏 (Grand Valley State University)
Free energy and ground state of the spiked SSK spin-glass model
[ 講演概要 ]
Spin-glasses are essentially mathematical models of particle interactions, and were originally describing magnetic states characterized by randomness in condensed matter physics. Due to the versatility of these types of models, however, they are now studied much more broadly for various complex systems such as statistical inference problems, weather/climate models or even neural networks. In this talk we will lay out the basic concepts of spin-glass models, while then focusing on the spiked SSK variant and its free energy as well as ground state energy. Furthermore we will discuss how one can determine these quantities including their lower order fluctuations with a so called "TAP approach" that was in this comprehensive form introduced in 2016 by N. Kistler and D. Belius, and what its benefits are compared to the earlier established "Parisi approach".
Spin-glasses are essentially mathematical models of particle interactions, and were originally describing magnetic states characterized by randomness in condensed matter physics. Due to the versatility of these types of models, however, they are now studied much more broadly for various complex systems such as statistical inference problems, weather/climate models or even neural networks. In this talk we will lay out the basic concepts of spin-glass models, while then focusing on the spiked SSK variant and its free energy as well as ground state energy. Furthermore we will discuss how one can determine these quantities including their lower order fluctuations with a so called "TAP approach" that was in this comprehensive form introduced in 2016 by N. Kistler and D. Belius, and what its benefits are compared to the earlier established "Parisi approach".
