BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:From Paper to Program: Overcoming the AI Coding Bottleneck in Scie
 nce
DTSTART;VALUE=DATE-TIME:20260610T020000Z
DTEND;VALUE=DATE-TIME:20260610T030000Z
DTSTAMP;VALUE=DATE-TIME:20260630T094400Z
UID:indico-event-5100@indico-tdli.sjtu.edu.cn
DESCRIPTION:Speakers: Yi Zhou(周毅) (Institute of Physics\, Chinese Acad
 emy of Sciences)\n\nHost: Prof. Xin Liu  Venue: TDLI Meeting Room N600
   Abstract:Large Language Models (LLMs) are highly capable at general pr
 ogramming\, yet they consistently fail to translate complex scientific pap
 ers into working code. In this talk\, I will demonstrate that the true bot
 tleneck is not the AI's coding ability\, but knowledge externalization. Re
 search papers are written for human experts and routinely omit the tacit\,
  computational "how-to" knowledge required to actually build the software.
 To solve this\, we designed a multi-stage "Virtual Research Group" workflo
 w. Instead of asking the AI to code directly from a paper\, an intermediat
 e AI agent first drafts a rigorous technical blueprint. Tested on a notori
 ously difficult quantum physics algorithm (DMRG)\, this approach boosted t
 he AI's success rate from 46% to 100% across 16 different model combinatio
 ns—compressing weeks of graduate-level coding into under 24 hours.  Bi
 ography:Yi Zhou is a professor at Institute of Physics\, Chinese Academy o
 f Sciences. He received his B.S. degree and Ph. D. in physics from Tsinghu
 a University in 1998 and 2004 respectively. After postdoctoral journey\, h
 e became a faculty member in Zhejiang University in 2009 and moved to curr
 ent position in 2019. His main research interests include quantum many-par
 ticle physics and theoretical condensed matter physics.\n\nhttps://indico-
 tdli.sjtu.edu.cn/event/5100/
LOCATION:Tsung-Dao Lee Institute/N6F-N600 - Lecture Room (Tsung-Dao Lee In
 stitute)
URL:https://indico-tdli.sjtu.edu.cn/event/5100/
END:VEVENT
END:VCALENDAR