MINNPACK 2026

MINNPACK 2026: Minneapolis Hub for Computational Science & Optimization (Oct 28-29, 2026)

Scheduled for October 28-29, 2026, in Minneapolis, Minnesota, MINNPACK 2026 is poised to be the premier regional gathering for researchers, practitioners, and students in computational science, optimization, and numerical analysis. Building on the legacy of the original MINPACK software library developed at Argonne National Lab and the University of Minnesota, this workshop series (often hosted by the University of Minnesota) focuses on the cutting-edge algorithms, software, and applications driving progress in solving complex scientific and engineering problems.

Core Content & Themes:

MINNPACK 2026 will center on the development, analysis, and application of sophisticated numerical methods, with a strong emphasis on nonlinear optimization and large-scale computational problems. Expect a blend of theoretical foundations, algorithmic innovations, practical software implementations, and real-world case studies. Key thematic areas likely include:

1. Nonlinear Optimization Algorithms: Deep dives into state-of-the-art methods for unconstrained and constrained optimization. Topics will cover:
Derivative-Free Optimization (DFO): Algorithms for problems where gradients are unavailable or unreliable (e.g., complex simulations, expensive experiments).
Large-Scale Optimization: Techniques (e.g., limited-memory quasi-Newton methods, interior-point methods, subspace methods) tailored for problems with millions of variables.
Nonlinear Least Squares: Specialized algorithms (e.g., variants of Levenberg-Marquardt) prevalent in data fitting and inverse problems.
Global Optimization: Heuristics and deterministic methods for finding global minima in non-convex landscapes.
Optimization under Uncertainty: Incorporating stochasticity or robustness into optimization models.

2. Software & High-Performance Computing (HPC):
Modern Software Ecosystems: Presentations on and comparisons of leading libraries (e.g., SciPy, NLopt, IPOPT, KNITRO, Pyomo, CVXPY) and their capabilities.
Parallel & Distributed Computing: Strategies for scaling optimization algorithms efficiently on multi-core CPUs, GPUs, and clusters.
Software Design & Maintenance: Best practices for developing robust, efficient, and maintainable numerical software.

3. Applications & Interdisciplinary Connections:
Machine Learning & Data Science: Optimization's critical role in training models (e.g., neural networks via SGD, support vector machines, hyperparameter tuning).
Engineering & Physics: Solving systems of nonlinear equations, parameter estimation, optimal control, and design optimization.
Computational Biology & Chemistry: Molecular modeling, drug design, parameter fitting in biological systems.
Finance & Economics: Portfolio optimization, risk management, and econometric modeling.

4. Emerging Trends & Interfaces:
Optimization for AI: Novel algorithms leveraging AI techniques (e.g., neural networks for surrogate modeling) and vice-versa.
Differentiable Programming: The intersection of automatic differentiation and optimization algorithm design.
Quantum Computing & Optimization: Exploring hybrid approaches and the potential impact of quantum hardware on optimization landscapes.

Format & Audience:

The workshop will feature a mix of invited plenary talks by leading international experts, contributed talks presenting novel research, and poster sessions showcasing ongoing work. Panel discussions on hot topics and software tutorials are also common elements. The audience is expected to be diverse, including academic researchers, industrial scientists and engineers, software developers, and advanced graduate students from mathematics, computer science, engineering, physics, and related fields. MINNPACK 2026 aims to foster collaboration, disseminate new knowledge, and strengthen the computational science community in the Upper Midwest and beyond. (Note: Specific speakers, exact topics, and detailed schedules will be announced closer to the date; this overview reflects the typical scope and focus of the MINNPACK series.)