At ICCM 2025, Jiaxin Sun from Julin Technology States: "Mathematics is the Only Precise Language for 'What You Simulate Is What You Get'"

On January 3, 2026, the 10th International Congress of Chinese Mathematicians (ICCM) opened in Shanghai under the theme "New Frontiers in Mathematics: A Driving Force Transforming Science and Humanity". JinongChen, Secretary of the Shanghai Municipal Committee of the Communist Party of China, attended and addressed the opening ceremony. ZhengGong , Deputy Secretary of the Shanghai Municipal Committee and Mayor of Shanghai, was also present.

At the opening ceremony, addresses were delivered successively by Shing-Tung Yau, Chairman of the 10th International Congress of Chinese Mathematicians (ICCM) and Chairman of the Shanghai Academy of Mathematics and Interdisciplinary Sciences; JiachangChen, Vice Minister of the Ministry of Science and Technology; XinQiu, Secretary of the Party Committee of Fudan University; LumingLi, President of Tsinghua University; and Caucher Birkar, Foreign Member of the Chinese Academy of Sciences and Fields Medal laureate. Also in attendance were YuanHua and JieChen, leaders of the Shanghai Municipal Government, and KuilingDing, President of Shanghai Jiao Tong University.

At the subsequent "Mathematics + Industry" Forum, JiaxinSun, Founder of Julin Technology, delivered a special report entitled A Brief Discussion on Mathematical Challenges in EDA Software. From the perspective of the underlying mathematical engine, he systematically elaborated on how Julin Technology, through in-depth cooperation with the Yau Mathematical Sciences Center at Southeast University, has leveraged innovative algorithms to solve the simulation challenges of chips in high-frequency and high-density scenarios. He also analyzed the pivotal role of industrial mathematics in breaking the performance bottlenecks of EDA software.

I. EDA: The "Lithography Machine" of the Information Age

In the post-Moore era, where chip processes are advancing toward the angstrom level and system complexity is growing exponentially, EDA is no longer merely an auxiliary tool. In his speech, JiaxinSun put forward an accurate analogy: "If lithography machines determine the precision of physical manufacturing, then EDA determines the precision of design success."

Faced with the GHz/THz high-frequency era, chip design is confronted with the severe challenge of "high-order coupling and nonlinearity". The traditional paradigm of circuit design has become obsolete, replaced by the problem of electromagnetic wave propagation and coupling in three-dimensional space. JiaxinSun pointed out that the core value of EDA lies in building a mathematical insurance mechanism. Through high-fidelity simulation, it achieves "what you simulate is what you get", ensuring that the risks of soaring power density and physical effect failure are resolved before tape-out, thus realizing First-Time-Right.

II. Maxwell's Equations: The Physical Cornerstone of High-Fidelity Simulation

"Mathematics is the only precise language that connects the virtual world with physical reality." JiaxinSun emphasized. The technical path of Julin Technology is precisely built on the ultimate solution of physical laws — especially Maxwell's equations.

In response to the paradigm shift from "circuits" to "waveguides", Julin Technology has constructed three major technical pillars to address the core contradiction between continuous physical equations and discrete numerical solutions:

· Precision Revolution in Geometric Discretization: To tackle the inherent contradiction between mesh density and computing resources, Julin has introduced adaptive mesh refinement technology. Based on error estimation methods (such as superconvergent recovery), the system can automatically identify sensitive regions — such as connector contacts and high-speed traces — and perform mesh refinement. In the design of a 10G network card with PCIe 4.0, this technology reduced the simulation time by 30% while maintaining a 5% error tolerance.

· Intelligent Breakthrough in Solvers: Faced with the computational challenge of large-scale sparse matrices, Julin aims to break through the memory and speed bottlenecks of traditional algorithms through pre-research on parallel algorithm optimization and hardware adaptation, combined with AI preprocessing technology.

· System-Level Dimensionality Reduction Acceleration: Using Low-Rank Approximation and SVD (Singular Value Decomposition) technologies, it achieves Model Order Reduction (MOR). This makes it possible to reduce the complexity of electromagnetic models with millions of degrees of freedom by 10 to 100 times, thereby supporting real-time simulation for digital twins and Hardware-in-the-Loop (HIL).