问题一

Question 1

如何构建“可用于决策”的可信模型，并形成可追溯、可量化的真实性评估体系？

How can we build decision-grade credible models and establish a traceable, quantitative framework for assessing model truthfulness?

问题二

Question 2

如何将不确定性量化与模型校准深度内嵌进端到端仿真流程？

How can UQ and model calibration be deeply embedded into an end-to-end simulation workflow?

问题三

Question 3

面向多物理场、跨尺度耦合系统，如何建立统一的建模—耦合—验证范式？

For multiphysics, multiscale coupled systems, how can we establish a unified paradigm for modelling, coupling, and validation?

问题四

Question 4

如何构建可个体化、可验证的多尺度人体器官生理模型，并实现临床可用的保真—准实时仿真？

How can we construct individualized, verifiable multiscale physiological models of human organs and achieve clinically usable high-fidelity, near-real-time simulation?

问题五

Question 5

无物理样机条件下，如何在复杂工况与随机载荷下可信预测超大吨位装备结构疲劳寿命？

Without physical prototypes, how can we credibly predict fatigue life of ultra-large-tonnage equipment under complex operating conditions and random loads?

问题六

Question 6

如何在异构硬件上实现大型复杂装备的 保真“准实时/实时”整机仿真？

How can we realize high-fidelity “quasi-real-time/real-time” full-machine simulation of large complex equipment on heterogeneous hardware?

问题七

Question 7

大模型能否成为可泛化的“世界模型”，在物理系统中实现可解释、可外推的建模与推演？

Can foundation models become generalizable “world models” that enable interpretable and out-of-distribution physical modelling and inference?

问题八

Question 8

如何对 DFT 中的交换-相关近似进行不确定性量化，并传递到多尺度模型与材料设计？

How can we quantify uncertainty from exchange–correlation (XC) approximations in DFT and propagate it into multiscale models and materials design?

问题九

Question 9

面向亿级规模芯片时序签核，如何构建兼具物理约束与不确定性估计的时序替代模型体系？

For billion-scale timing sign-off, how can we build timing surrogate models that respect physics and provide uncertainty estimates?

问题十

Question 10

如何以 LLM 为核心引擎构建 逼真社会仿真，同时确保可验证性、可解释性与伦理稳健性？

How can we build high-fidelity LLM-centered social simulations while ensuring verifiability, interpretability, and ethical robustness?