Research Overview
SECQUOIA's research spans connected themes represented in David E. Bernal Neira's publication record and in the group's current projects: mathematical optimization, process systems engineering, quantum optimization and simulation, privacy-aware learning, and open scientific software. Across these areas, we develop models, algorithms, and benchmark problems that connect rigorous theory with deployable tools for chemical, energy, biomedical, and other industrial applications.
Mathematical Optimization, GDP, and MINLP
We develop theory, algorithms, and modeling tools for generalized disjunctive programming, mixed-integer nonlinear programming, and other nonlinear discrete optimization models. Representative directions include outer-approximation methods, convexification, regularization, decomposition, and logic-based formulations for planning, scheduling, design, and dynamic optimization problems.
Process Systems Engineering and Sustainable Operations
Our optimization methods are motivated by process systems engineering applications, including process synthesis, design and control, refinery planning and scheduling, manufacturing networks, process intensification, water systems, and supply chains. These projects target sustainability and operational reliability by improving resource efficiency and the operation of complex chemical and energy systems.
Quantum Optimization, Simulation, and Benchmarking
We study how emerging quantum and quantum-inspired methods can help solve hard optimization and simulation problems. Research directions include hybrid quantum-classical optimization, QUBO and Ising formulations, routing and network problems, variational algorithms, Hamiltonian simulation benchmarks, and practical assessments of quantum hardware and heuristic solvers on scientifically meaningful test cases.
Federated Learning and AI for Engineering and Healthcare
We investigate machine learning methods that respect privacy, distribution, and scientific structure in the data. Representative efforts include federated learning for chemical engineering and healthcare, tensor-network-based learning architectures, privacy-aware distributed collaboration, and learning-enhanced optimization workflows.
Open-Source Software, Libraries, and Reproducibility
Reproducible software is a core part of our research. We contribute to open-source tools and benchmark libraries for optimization and quantum computing, including modeling ecosystems, solver toolkits, curated problem libraries, and reusable benchmark instances. Representative efforts in the group's software and publication record include MindtPy, Pyomo.GDP, QUBO.jl, GDPLib, and quantum benchmark libraries such as HamLib.