The AMS Conference 2018 brought together applied mathematicians, scientists, and industry leaders to explore cutting-edge computational methods. This gathering highlighted advances in modeling, simulation, and algorithms that address real-world challenges across multiple sectors.
Attendees engaged with emerging techniques for data analysis, uncertainty quantification, and high-performance computing. The event reinforced connections between theory, software tools, and deployment in science and engineering applications.
| Key Topic | Description | Notable Speakers | Impact Focus |
|---|---|---|---|
| Scientific Machine Learning | Integration of data-driven models with traditional numerical methods | Catherine B. Pelizzon, Michael P. Brenner | Accelerating discovery in physics and biology |
| Uncertainty Quantification | Rigorous treatment of variability in models and measurements | Alexandre Tartakovsky, Omar Ghattas | Improved risk assessment for infrastructure and climate |
| High-Performance Computing | Scalable algorithms and architectures for extreme-scale problems | Jack Dongarra, Mary W. Gray | Enabling exascale simulations worldwide |
| Computational Biology | Modeling biological systems at multiple scales | Michele Fezio, Ruth Ley | Advancing drug discovery and personalized medicine |
Scientific Modeling and Applied Mathematics
Advanced Numerical Methods
Speakers discussed robust numerical schemes for differential equations, mesh adaptation, and high-order discretizations. These methods underpin accurate simulations in climate, aerodynamics, and materials science.
Multiscale and Multiphysics Approaches
Techniques bridging atomistic and continuum models were a central theme. Delegates explored coupling strategies that preserve consistency across scales, improving predictive capability for complex systems.
Data-Driven and Machine Learning Topics
Integration with Traditional Solvers
Machine learning components were embedded into established solvers to accelerate convergence and reduce parameter tuning. Demonstrations showed gains in optimization and surrogate modeling tasks.
Uncertainty-Aware Learning
Presentations emphasized quantifying uncertainty in neural network outputs. Methods linked Bayesian inference with scientific simulations to support decision-making under incomplete information.
High-Performance Computing and Infrastructure
Scalable Algorithms
New algorithms designed for heterogeneous architectures were showcased. Emphasis was placed on communication-avoiding techniques and load balancing for extreme-scale systems.
Software Ecosystems
Toolkits for reproducible research, including automatic differentiation and symbolic manipulation, were highlighted. Open-source frameworks lowered barriers for broader adoption in academia and industry.
Applications and Impact
Energy and Environmental Systems
Modeling subsurface flow, battery design, and climate projections illustrated the societal value of computational advances. Improved fidelity helps policymakers evaluate long-term strategies.
Healthcare and Biomedical Engineering
Simulations of blood flow, tissue mechanics, and epidemiology demonstrated how applied mathematics supports clinical innovation. Personalized treatment planning benefited from rigorous quantification of variability.
Looking Ahead to Computational Research
- Embrace hybrid modeling that blends data and first principles for greater reliability.
- Invest in scalable algorithms that leverage exascale hardware and emerging architectures.
- Promote open standards to ensure reproducibility and broader collaboration.
- Strengthen ties between applied mathematics, domain scientists, and industry.
- Develop educational programs that train the next generation of computational scientists.
FAQ
Reader questions
What types of problems were showcased at AMS Conference 2018?
Presentations covered problems in fluid dynamics, materials design, biology, and climate science, with a focus on methods that combine rigorous mathematics with modern computing.
How did the conference address uncertainty in real-world models?
Uncertainty quantification sessions emphasized statistical and computational tools to characterize variability, enabling more reliable predictions for infrastructure and environmental planning.
Which software tools and libraries were highlighted for attendees?
The event featured toolkits for mesh generation, linear algebra, machine learning, and reproducible workflows, encouraging open science and easier adoption of advanced numerical methods.
What opportunities existed for early-career researchers and students?
Dedicated workshops, poster sessions, and networking events connected students with leaders in the field, supporting mentorship and visibility for emerging talent.