PhD Scholarship in Operations Research: Acceleration Methods for Stochastic Optimization Problems – DTU Management

The Department of Technology, Management and Economics at DTU invites applications for a PhD student position focused on leveraging AI to mitigate stochastic problems.

Job Responsibilities

As a PhD student, you will develop novel aggregation and scenario reduction methods in stochastic optimization using AI. Additionally, you will collaborate with the team to develop solvers for stochastic optimization problems and test these methods using real-world data.

Academic Development

As part of your PhD studies, you will engage in advanced coursework to enhance your skills, implement and test algorithms, and learn to author scientific papers.

About the Research Area and Project

The ERC-funded project DECIDE aims to create leading-edge heuristic and exact methods for two-stage stochastic optimization problems. The objective is to solve significantly larger problems, encompassing numerous scenarios for the future, using AI-based methods to generate scenarios and guide the search.

Stochastic optimization problems often have extended time horizons with numerous low-level decision variables, making them complex to solve. Data aggregation or scenario reduction can accelerate the solution process by reducing problem size. Data aggregation leverages periodicity in problems, while scenario reduction removes similar scenarios.

Both approaches utilize clustering in time and scenario dimensions, resulting in faster solution times with minimal decline in solution quality. This PhD project will explore the potential of AI to reduce stochastic problems without compromising solution quality, focusing on data aggregation and scenario reduction interplay.

The Team and Leadership

You will join the robust academic team at the Management Science division of DTU Management, which includes three PhD students, four Postdocs, and several international operations research experts. The project is led by Professor David Pisinger, a distinguished author in the field who has supervised over 30 PhD students. The DECIDE project is funded by the European Research Council (ERC-2022-ADG).

Responsibilities and Qualifications

Within the DECIDE project, your key responsibilities will include:

• Developing next-generation decision-making tools under uncertainty.
• Collaborating on the creation and testing of algorithms for complex combinatorial optimization problems.
• Investigating AI applications in stochastic optimization and metaheuristics.

Required qualifications:

• Experience with AI models such as Variational Autoencoders, Normalized Flows, and Generative Adversarial Networks.
• Proven experience in developing fast algorithms for hard combinatorial optimization problems.
• Knowledge of stochastic optimization and metaheuristic implementation.
• Strong research passion, excellent programming skills, collaboration capabilities, and fluency in spoken and written English.
• A two-year master’s degree (120 ECTS points) or an equivalent qualification.

Approval and Enrolment

The PhD scholarship is contingent on academic approval, and the selected candidate will be enrolled in a DTU general degree programme. For further details on enrolment requirements and the PhD study programme, please refer to .

Assessment Process

Qualified candidates will be assessed by two representatives from DTU Management.

Our Offer

DTU provides a stimulating and rewarding international environment, fostering academic excellence, collegial respect, and academic freedom. We are committed to excellence in research, education, innovation, and scientific advice.

Salary and Terms

The position is full-time for a period of 3 years, starting on 1 October 2024 (or as per mutual agreement). The salary and appointment terms will be based on the collective agreement with the Danish Confederation of Professional Associations, and any allowances will be determined with the relevant union.

For more information about career paths at DTU, please refer to our .

Further Information

For more details, please contact Professor David Pisinger at +45 4525 4555 or via email at . Additional information about DTU