Propose a minisymposium

Proposals for minisymposia will be accepted until July 15, 2025.   A successful proposal should include a descriptive title and abstract, list of organizers, and number of potential speakers.  The abstract should describe the topic of the session clearly and is meant to attract both audience members and speakers.  

Minisymposium proposals can be submitted through the form. 

Sessions will consist of up to six 20 minute talks with 5 minutes of of questions per talk (25 min in total for one talk). Sample sessions from the 2023 Conference can be found at that meeting's website.  

Proposals are subject to approval by the 2025 SIAM PNW Organizing Committee.  

Accepted Minisymposia

• Recent Advances in Numerical Methods for Hyperbolic and Related Problems

Organizers: Jingwei Hu (University of Washington), Donsub Rim (Washington University in St. Louis), Christiane Helzel (Heinrich-Heine-University Dusseldorf)

Numerical methods for hyperbolic and related problems have advanced rapidly over the past decades. Recent developments have emphasized structure preservation, high dimensionality, multiscale and inverse modeling, and the computational efficiency of these problems. These methods and their generalizations have become essential tools in scientific inquiries in other disciplines, for example, in geosciences or plasma physics. This minisymposium will highlight recent progress in the field and showcase the impact of Randy LeVeque’s contributions to the broader hyperbolic community as well as to scientific domains beyond numerical analysis.

• Water Waves

Organizers: Eleanor Byrnes (University of Washington), John Carter (Seattle University)

This minisymposium focuses on nonlinear water waves. The speakers will present analytical and numerical results from mathematical models of waves on shallow and/or deep water.

• New Trends in Optimization

Organizers: François Clément (University of Washington), Stefan Steinerberger (University of Washington)

This minisymposium brings together together seasoned and emerging scholars to explore recent advances in optimization. Topics include, but are not limited to, discrepancy, kernel methods in data analysis, vector coloring, and discrete optimization in combinatorial structures.

• Applications of Neural Network Based Empirical Kernels

Organizers: Saad Qadeer (Pacific Northwest National Laboratory), Panos Stinis (Pacific Northwest National Laboratory)

Even while the use of neural network (NN) architectures has seen a meteoric increase, their learning properties, e.g., feature learning, double-descent, and transfer learning, still await a satisfactory treatment. The Neural Tangent Kernel (NTK), induced by the feature map given by the Jacobian of an NN with respect to its trainable parameters, controls the training dynamics and, in certain limiting regimes, encodes the learned NN. Due to these deep connections, the empirical NTK offers an attractive potential route towards understanding and possibly enhancing NN performance. This minisymposium aims to bring together researchers and practitioners to explore recent advances in the use of NN based empirical kernels, such as the NTK or its variants. 

• Data-driven Mathematical Models for Prediction and Control of Biology

Organizers: Obinna Ukogu (University of Washington), Ruibo Zhang (University of Washington)

Biological phenomena are complex in their nature, spanning multiple scales and involving many interacting components. Recent technological advances have enhanced our ability to observe complexity in the fields of ecology, oncology, infectious disease, and molecular biology. Faced with a growing stream of data, mathematical models are crucial for interpreting biological phenomena by building predictive and, sometimes, causal models. This minisymposium will highlight the use of mathematical models to connect data to biological mechanisms in complex systems. Areas of interest include optimal control, stochastic processes, differential equations, graphical models, and their applications in prediction, parameter inference, or hypothesis testing.

• Scientific Machine Learning

Organizers: 

Alex Johnson-Vázquez,  Alexander Hsu,  Juan Felipe Osorio Ramírez, Michele Martino, Bamdad Hosseini (University of Washington)

Developing machine learning models for use in the sciences often requires special attention to attain high precision and quantify uncertainty. Likewise, understanding machine learning itself raises interesting theoretical problems, such as understanding the dynamics of learning systems. This minisymposium will highlight recent topics in the field, such as reduced-order modeling, probabilistic modeling, measure transport, and data-driven modeling of differential equations. Both the mathematical theory and scientific applications will feature prominently in this session.

• Traditional vs surrogate modeling of complex phenomena

Organizers: Malgorzata Peszynska (Oregon State University), Lynn Schreyer (Washington State University)

The traditional way we do modeling of complex phenomena involves physically meaningful descriptions derived from first principles and their accurate computational approximations, combined with attempts to use realistic data and to conduct verification and validations. However, at some point the complexity  of such efforts is overwhelming and might not practical. 

In this session we discuss various ways to replace or supplement the traditional models by reduced order models, with modeling approximations such as compartment models or neural networks used as surrogates. These are useful especially in the context of multiple scales and large data uncertainty. 

• Geometry and Topology in the Analysis of Data

Organizer: Michael Perlmutter (Boise State University)

Modern data sets are often complex and high-dimensional. However, such data sets often have a hidden low-dimensional structure. This session will focus on techniques, rooted in topology and geometry which will harness this structure to provide simplified representations of the data and otherwise overcome the curse of dimensionality for machine learning and data science tasks.

• Celebrating Randy LeVeque and his contributions to tsunami modeling

Part I: mathematical and numerical advances; Part II: Applications

Organizers: 

Yong Wei (University of Washington), Breanyn MacInnes (Central Washington University), Alex Dolcimascolo (Washington Geological Survey)

Extreme natural hazards such as earthquakes and tsunamis present unique mathematical and computational challenges due to their multi-scale nature, complex interactions with coastal environments, and potential catastrophic consequences. This minisymposia aims to bridge theoretical developments with practical applications that capture the complex dynamics and far-reaching impact of these geophysical processes. We highlight recent progress in mathematical methods, numerical approaches, and innovative computational techniques. Topics include seismic and non-seismic origins of tsunamis, inundation modeling, wave-structure interactions, multi-fluid dynamics, high-performance 3D earthquake simulations, integration of deep-learning and artificial intelligence, inundation hazard mapping, tsunami warning and forecast, and GeoClaw development and education.

By this minisymposium, we honor Prof. Randy LeVeque’s commitment to open-source development and collaboration of the GeoClaw tsunami modeling software. His work has fostered a vibrant interdisciplinary community that supports both fundamental research and operational hazard mitigation. In addition, GeoClaw provides a valuable entry point for students to engage in research early in their academic careers.

• Recent advances in analysis of nonlinear PDE

Organizers: Elaine Cozzi, Radu Dascaliuc, Xueying Yu (Oregon State University)

A common feature of many distinct systems of partial differential equations (PDEs) arising in physical and biological contexts is the nonlinear interaction across multiple scales—a phenomenon that poses significant mathematical challenges. These systems include fluid dynamics equations, such as the Navier-Stokes and Euler equations, as well as dispersive PDEs like the nonlinear Schrödinger equation, among others. The methods used to study well-posedness, regularity, and qualitative behavior of these equations often differ substantially. The aim of this minisymposium is to bring together both early-career researchers and established experts to explore how recent methodological advances can be applied across different types of PDEs.

• Recent Developments on Hybrid Methods Combining Neural Networks with Classical Numerical Methods

Organizers: Yukun Yue (University of Wisconsin-Madison), Shukai Du (Syracuse University)

The integration of neural networks with classical numerical methods has emerged as a promising direction for solving complex partial differential equations, inverse problems, and other computationally intensive tasks. These hybrid approaches seek to combine the reliability, accuracy, and interpretability of traditional numerical schemes with the efficiency, flexibility, and approximation power of machine learning techniques. This minisymposium will bring together researchers working at this intersection to develop and analyze such frameworks. Talks will highlight theoretical advances, algorithmic designs, and applications, with a focus on fostering collaboration between the numerical analysis and machine learning communities.

• Scientific Computing and Numerical Analysis

Organizers: Grady Wright (Boise State University), Heather Wilber (University of Washington)

This minisymposium highlights recent advances in scientific computing and numerical analysis, with a particular focus on high-order methods, numerical linear algebra, and approximation theory. Emphasis will be placed on the development of accurate, efficient, and robust algorithms for complex problems arising in science and engineering. Talks will cover theoretical foundations, algorithmic innovations, and applications that push the boundaries of modern computational methods.

• Partial Differential Equations: Modeling, analysis, and computation

Organizers: Jeffrey Ovall (Portland State University), Hannah Kravitz (Portland State University)

This minisymposium welcomes talks in which the computation of solutions to partial differential equations (source problems or eigenvalue problems) plays a central role.  Contributions may range from those that are more focused on modeling and/or application, to those that emphasize numerical analysis, to those that primarily concern scientific computing and simulation.  We will also consider talks that present novel algorithms, even if the associated analysis, coding and/or applications are not as well developed.

• Data Assimilation in Earth and Atmospheric Sciences

Organizers: Brian Kyanjo (Georgia Institute of Technology), Alex Robel (Georgia Institute of Technology)

This minisymposium highlights recent advances in data assimilation for ice sheet modeling, with a focus on techniques that enhance the initialization, prediction, and uncertainty quantification for ice sheet simulation. Speakers will present developments in ensemble Kalman filters, variational approaches, adjoint-based methods and other hybrid techniques applied to ice sheet models. The session will highlight the integration of observational data—such as surface velocity, thickness, and grounding line position—into forward models. It will also feature emerging open-source tools, including the Ice Sheet State and Parameter Estimator (ICESEE) (https://github.com/ICESEE-project/ICESEE), a Python-based framework designed for scalable ensemble data assimilation with models like ISSM and Icepack. This session aims to bring together researchers working at the intersection of glaciology, computational modeling, and data assimilation to foster collaboration and methodological innovation.

• Modeling in the Life Sciences

Organizers: Zhen Chao (Western Washington University), Tilmann Glimm (Western Washington University)

Quantitative modeling plays a central role in understanding complex biological systems, from the molecular to the ecological scale. The life sciences are increasingly reliant on mathematical models and computational tools to uncover mechanisms, predict behavior, and inform decision-making in biology, medicine, and public health.

This minisymposium will highlight recent advances in mathematical modeling across a broad range of life science applications and from groups across the Pacific Northwest. Topics may include, but are not limited to, infectious disease dynamics, cell biology, population biology, neuroscience, biomechanics, systems biology, and bioinformatics.  Novel mathematical results, both computational and analytical, and interdisciplinary approaches will be represented with emphasis on model development and analysis, integration with experimental data, and real-world applications. 

The session aims to showcase recent work and foster dialogue among researchers in applied mathematics, computational science, and life science from the Pacific Northwest and beyond, with the goal of advancing collaborative research and promoting innovative mathematical approaches to biological challenges. 

• Mathematical Modeling for Systems Neuroscience

Organizers: Saba Heravi, James Alexander Hazelden, Eli Shlizerman (University of Washington)

This minisymposium focuses on recent advances in mathematical and computational modeling of complex neuronal systems from the viewpoint of dynamical systems and data science. Topics span a variety of scales of neuronal systems, from individual neurons to whole-brain connectomics, covering parameter inference and connectomics, low-dimensional manifold learning, simulation frameworks and biophysical learning rules.

• Algebraic Methods in Mathematical Machine Learning

Organizers: Maksym Zubkov (University of British Columbia), Yulia Alexander (University of California, Los Angeles), Jiayi Li (Max Planck Institute)  

This minisymposium will focus on applications of algebraic methods for understanding the mathematical theory of machine learning and optimization. It will cover topics including applied algebraic geometry, group symmetries, data and optimization invariances. It will focus on the applications of algebraic techniques to analyze the training and structural aspects of neural networks. The session will allow researchers to discuss and share their latest results on geometry of neural networks, equivariant architectures, symmetries, and non-convex optimization.