This fully funded PhD project at The University of Manchester focuses on uncertainty quantification for multi-physics simulations, a critical challenge in modeling complex systems such as composite materials, carbon sequestration, and electric power grids. Multi-physics systems are characterized by nested components operating at different time and spatial scales, often with non-linear interactions and a large number of elements. Simulating these systems is computationally intensive, and while computer emulation has advanced for individual components, its application to multi-physics scenarios remains underdeveloped. The project aims to advance the theory and practical tools for composing individual computer emulators into hierarchical structures, leveraging concepts from deep Gaussian processes (deep GPs) and linked emulators. The research will bridge the gap between generic model formulation and real-world multi-physics applications, with a particular focus on fusion energy. The ideal candidate will have an MSc in Physics, Computer Science, Statistics, Mathematics, or a related field, and will contribute to methods development in this emerging area. The program is part of the UKRI AI Decisions CDT and is supported by the UK Atomic Energy Authority (UKAEA) as an industry partner. Successful applicants will receive full funding, including home tuition fees and a tax-free stipend at the UKRI rate (£20,780 for 2025/26), with a start date in September 2026. The University of Manchester is committed to equality, diversity, and inclusion, encouraging applicants from all backgrounds and offering flexible study options. Applicants are required to submit transcripts, CV, a supporting statement, and referee details, and are encouraged to contact the project supervisor or the UKRI AI Decisions CDT Team for queries. The project offers an excellent opportunity to develop expertise in AI, machine learning, and statistical modeling for complex multi-physics systems, with direct industry engagement and a supportive research environment.

This PhD project at The University of Manchester aims to revolutionize computational fluid dynamics (CFD) by developing an Agentic AI system that automates mesh generation and synthesizes best practices from decades of CFD literature. Mesh generation and methodological best practices are major bottlenecks in CFD, often requiring expert intervention and extensive trial-and-error. The proposed research will create a multi-agent AI framework, where specialized agents handle user input, mine literature for proven strategies, evaluate mesh and solver performance, and use reinforcement learning to iteratively refine both meshing strategies and extracted best practices. The system will be built on open-source platforms such as Gmsh and OpenFOAM, ensuring transparency and reproducibility. Initial test cases will focus on canonical problems before advancing to complex flows, such as the reverse swing of a cricket ball. The reinforcement learning framework will enable adaptability, with reward functions guiding continuous improvement. Over time, the system will build a knowledge base of meshing strategies and CFD workflows that generalize across geometries and benchmark against industry standards. Expected outcomes include an open-source repository of literature-based strategies, a multi-agent AI framework for best practice synthesis and mesh automation, and demonstrated improvements in CFD workflow efficiency, reliability, and accessibility. The project is fully funded through the UKRI AI CDT program and Cummins Inc., offering home tuition fees and a tax-free stipend. Applicants should have a strong background in engineering, applied mathematics, physics, or computer science, and be motivated to bridge CFD and AI. The university encourages applications from diverse backgrounds and supports flexible study arrangements. The application process requires submission of transcripts, CV, a supporting statement, referee contact details, and an English language certificate if applicable. The deadline for applications is December 5, 2025, with a start date in September 2026.

This fully funded PhD project at The University of Manchester, in collaboration with Honda Research Institute (HRI) Europe, aims to develop an adaptive optimization framework for complex socio-technical systems, with a particular focus on energy distribution networks for electric vehicles (EVs). These systems present large, high-dimensional search spaces and require balancing multiple, often conflicting and expensive-to-evaluate objectives such as fairness, explainability, user satisfaction, and environmental impact. Current industry approaches are typically rule-based, slow, and costly, motivating the need for advanced optimization techniques. Building on prior research in expensive, multi-objective, and high-dimensional optimization, the project will explore guided local and global search strategies, intelligent variable subset selection, and the integration of socially derived criteria into optimization objectives. The research will leverage Bayesian Optimization and Gaussian Processes to efficiently navigate expensive search landscapes, exploiting correlations and problem properties to optimize under limited evaluation budgets. The successful candidate will have opportunities to visit HRI Europe and join their global PhD cohort, gaining exposure to industry-driven research and innovation. Applicants should have a strong background in optimization, ideally with experience in multi-objective and expensive optimization, and familiarity with Gaussian Processes. The program is part of the AI UKRI CDT, offering full funding including home tuition fees and a tax-free stipend at the UKRI rate (£20,780 for 2025/26). The start date is September 2026. The University of Manchester is committed to equality, diversity, and inclusion, encouraging applicants from all backgrounds and offering flexible study options. Application requires submission of transcripts, CV, supporting statement, and referee contact details via the university portal. For more information, visit the project and company webpages.