JapanTech

Autonomous Agentic Workflow for High-Performance Tensor Network Software

Talk on LLM-driven agentic AI that autonomously writes and optimizes high-performance tensor network code, plus the TCAPI interface. Hybrid.

When
Where
Open Space, RIKEN AIP Nihonbashi Office·Hybrid
Region
Kanto (Tokyo)
Organizer
RIKEN Center for Advanced Intelligence Project
Language
EN
Source
Doorkeeper
Autonomous Agentic Workflow for High-Performance Tensor Network Software

Summary

Dr. Rong-Yang Sun of The Chinese University of Hong Kong presents a fully autonomous agentic workflow, driven by large language models, for generating and optimizing high-performance tensor network applications. The talk shows that the approach can match state-of-the-art implementations written by human experts on realistic computational problems. The talk also introduces the Tensor Computing API (TCAPI), a unified cross-platform interface for tensor computations, and explains how its clear abstraction boundaries and concise interface contracts enable autonomous development. The schedule is a 45-minute talk followed by 15 minutes of Q&A, in English. Online attendance via Zoom is open to everyone who registers. In-person attendance at the RIKEN AIP Nihonbashi Office is limited to RIKEN members. It suits researchers and engineers interested in quantum physics, machine learning, high-performance computing and AI-assisted software development.
#agentic-ai#tensor-networks#quantum-computing#hpc#llm#research-talk#code-generation