擒数网 （随信APP) | 从学术界到金融科技：一位具有计算力学基础的创新数据科学家的成长之路

文章目录[隐藏]

• 脆冰受垂直结构挤压的Peridynamic模拟
• 聚脲弹性体微相结构的粗粒子分子建模
• 聚脲固化速率对结构和强度的粗粒子分子模拟
• Peridynamic simulation of brittle-ice crushed by a vertical structure
• Coarse-grained molecular modeling of the microphase structure of polyurea elastomer
• Coarse-grained molecular simulation of the role of curing rates on the structure and strength of polyurea

明浩刘提供的照片

《数字新闻》的贡献者表达的观点属于他们自己。

刘明浩的职业生涯证明了计算科学在解决复杂实际问题中的力量。从他在计算力学方面的开创性研究开始，明浩一直展现出推动理解动态系统边界的激情，特别是在传统实验无法涵盖的极端条件下。他在开发先进模拟和计算模型方面的专业知识不仅填补了科学知识的关键空白，还为他顺利过渡进入金融科技领域奠定了基础。

他的博士研究发表了许多有影响力的论文，反映了通过模拟推动理解边界的雄心。

脆冰受垂直结构挤压的Peridynamic模拟

在这项研究中，明浩运用Peridynamic理论来模拟脆冰在受到冲击力作用下的断裂力学，这是物理实验难以复制的条件。这项工作为在极端寒冷条件下提高结构设计提供了至关重要的洞见。

聚脲弹性体微相结构的粗粒子分子建模

在这篇论文中，明浩开发了一种聚脲弹性体结构匹配的粗粒子模型。该模型隐含地表示了氢原子，简化了分子模拟的复杂性。通过使用迭代Boltzmann反演和新颖的距离依赖缩放函数进行训练，该模型显著减少了迭代时间。模拟捕捉了聚脲的微相分离，揭示了五纳米硬领域间距，与类似弹性体的X射线散射数据一致。分析了两种不同的模型系统，研究表明多块系统形成大型互连硬结构域，而二块系统产生较小、带状的结构域。研究进一步揭示，软段形成桥型和环型结构，有助于在各种条件下更深入理解材料的微结构。

聚脲固化速率对结构和强度的粗粒子分子模拟

这项研究探讨了固化过程如何影响聚脲的力学强度，这种材料在需要高性能材料的行业中发挥着关键作用，如国防和汽车行业。通过在硅中模拟这些效应，更深入地了解材料在极端制造条件下的属性变化。

科学计算在明浩的所有研究工作中都是核心。通过使用计算模拟，明浩能够模拟不同尺度下复杂的物理系统，从在冰中的断裂力学到聚脲中的分子动力学。高性能计算的力量使得可以探索在现实实验中难以观察到或不可能实现的系统行为。这些计算技术，如Peridynamics，粗粒子建模和晶格Boltzmann方法，不仅减少了对物理样机的需要，还通过深入洞察在极端条件下材料性质提供了加速发现过程的优势。

这种科学计算的专业知识无缝转化为明浩在金融科技行业的数据科学家角色。在大规模数据分析和机器学习推动创新的领域中，明浩应用类似的计算方法解决金融问题。就像用模拟预测材料行为一样，现在也可以通过部署预测模型和算法来优化金融产品，分析市场趋势，并增强客户体验。

具体来说，明浩在他的数据科学角色中专注于信用风险建模和推荐系统。他在学术研究中发展的技能 - 模拟动态系统，分析复杂数据，并基于这些模型进行预测 - 在金融背景下直接适用。例如，信用风险建模需要理解在不同条件下贸易线数据的金融行为模式，就像明浩曾经在极端压力下模拟分子结构的材料属性一样。同样，构建推荐系统涉及到预测建模和模式识别，这些技能通过他在分子和材料模拟方面的工作得到了锻炼。在这两种情况下，明浩带来了科学严谨精神与创新金融解决方案相结合的独特方法。

通过将科学精确性与创新金融解决方案相结合，刘明浩展示了先进计算方法如何超越学科限制，推动技术和商业创新。他从学术到金融科技的旅程突出了数据科学和计算专业知识在各行业中应对最具挑战性问题中的转变力量。

Photo courtesy of Minghao Liu

Opinions expressed by Digital Journal contributors are their own.

Minghao Liu's career is a testament to the power of computational science in solving complex, real-world problems. Beginning with his groundbreaking research in computational mechanics, Minghao has consistently demonstrated a passion for pushing the boundaries of understanding dynamic systems, particularly under extreme conditions where traditional experiments fall short. His expertise in developing advanced simulations and computational models not only filled critical gaps in scientific knowledge but also laid the foundation for his seamless transition into the world of FinTech.

His Ph.D research, which resulted in many impactful publications, reflects this ambition to push the boundaries of understanding through simulations.

Peridynamic simulation of brittle-ice crushed by a vertical structure

In this study, Minghao applied peridynamic theory to simulate the fracture mechanics of brittle ice subjected to impact forces, conditions that are difficult to replicate in physical experiments. The work provided crucial insights into crack propagation patterns under extreme cold, helping to improve the design of structures exposed to harsh arctic conditions.

Coarse-grained molecular modeling of the microphase structure of polyurea elastomer

In this paper, Minghao developed a structure-matching coarse-grained model of polyurea elastomers. The model implicitly represented hydrogen atoms, streamlining the complexity of molecular simulations. Trained using iterative Boltzmann inversion and a novel distance-dependent scaling function, the model significantly reduced iteration times. The simulations captured the microphase separation of polyurea, revealing a hard domain spacing of five nm, consistent with x-ray scattering data from similar elastomers. Analyzing two different model systems, the research demonstrated that multiblock systems form large, interconnected hard domains, while diblock systems create smaller, ribbon-shaped domains. The study further revealed that the soft segments formed bridge-like and loop-like structures, contributing to a deeper understanding of the material's microstructure under various conditions.

Coarse-grained molecular simulation of the role of curing rates on the structure and strength of polyurea

This research explored how the curing process influences the mechanical strength of polyurea, a material that plays a key role in industries requiring high-performance materials, such as defense and automotive. Simulating these effects in silico allowed for a more nuanced understanding of how material properties change under extreme manufacturing conditions.

The use of scientific computing was central to all of Minghao's research efforts. By employing computational simulations, Minghao was able to model complex physical systems at different scales — ranging from fracture mechanics in ice to molecular dynamics in polyurea. The power of high-performance computing enabled the exploration of system behaviors that are difficult or impossible to observe in real-world experiments. These computational techniques, such as peridynamics, coarse-grained modeling, and lattice Boltzmann method, not only reduced the need for physical prototypes but also accelerated the discovery process by providing deeper insights into material properties under extreme conditions.

This expertise in scientific computing seamlessly translated into Minghao's current role as a data scientist in the FinTech industry. In a field where large-scale data analysis and machine learning drive innovation, Minghao applies similar computational methodologies to solve financial problems. Just as simulations were used to predict material behaviors, now predictive models and algorithms are deployed to optimize financial products, analyze market trends, and enhance customer experiences.

Specifically, Minghao focuses on credit risk modeling and recommendation systems in his data science role. The skills developed in his academic research — modeling dynamic systems, analyzing complex data, and making predictions based on those models — are directly applicable in financial contexts. Credit risk modeling, for example, requires understanding patterns of financial behavior of tradeline data under varying conditions, much like how Minghao once modeled material properties of molecular structures under extreme stresses. Similarly, building recommendation systems involves predictive modeling and pattern recognition, skills honed through his work with molecular and material simulations. In both cases, Minghao brings a unique approach that blends scientific rigor with innovative financial solutions.

By blending scientific precision with innovative financial solutions, Minghao Liu exemplifies how advanced computational methodologies can transcend disciplines, driving both technological and business innovation. His journey from academia to FinTech highlights the transformative power of data science and computational expertise in tackling the most challenging problems across industries.