The NCCR Automation is looking for a doctoral student to study the problem of fair and efficient resource allocation in engineering systems. The student will be cosupervised by Andrea Censi (Institute for Dynamic Systems and Control, D-MAVT) and Saverio Bolognani (Automatic Control Laboratory, D-ITET). The goal of this collaboration is to develop mechanisms and protocols that allow agents to compete for a limited shared resource (the road space, the electric grid, etc.). As the allocation of these resources has high societal relevance, we aim to develop tools endowed with fairness and efficiency guarantees.
The problem of repeated resource allocation in a competitive setting, when agents have private and time-varying need for the resource, can be described as a population game endowed with Markovian dynamics. This formalism is practical for analyzing the agents' rational behavior and for designing mechanisms that ensure efficient resource allocation. One such mechanism is the "karma game," where an agent-by-agent karma counter guarantees bookkeeping of the resource allocation and allows agents to strategically disclose their preference with their bids but without any truthfulness assumption.
https://arxiv.org/abs/2207.00495
https://arxiv.org/abs/2104.14662
In this project, we want to extend this idea to a much broader setting in which a similar mechanism can be employed to automate the allocation of a variety of heterogeneous resources in a much more complex setting.
We envision multiple possible generalization steps:
From decisions on resource allocation to decisions over generic outcomes
The current Karma formalization is limited to deciding the allocation of one resource (thus the interpretation of Karma messages as “bidding”, and players “winning” or “losing”). We want to extend this principle to the case of n agents with preferences over m outcomes, similar to the formalism used in social choice theory, voting, and bargaining theory.
Scalability and core selection: We aim at designing mechanisms that do not require any notion of reputation, do not require truthfulness of the agents, and where coalitions/collusion are not favored, even when the decision affects multiple agents.
Externalities and invisible players: we expect the selection of an outcome to affect many more agents than the ones involved in the negotiation, because many decisions in complex socio-technical systems come with externalities that are not factored in the price associated with these decisions. In a well-designed karma economy, we expect the karma system to account for these externalities and to make the resulting game a non-negative reward game for society at large, including non-participating (invisible) agents.
This Ph.D. position is located at the intersection of multiple domains: we plan to employ tools from Game Theory and Control Theory to model resource allocation problems that belong to diverse engineering domains like traffic, energy, robotics, and more.
You will have to work towards a dual goal. On the one hand, to further develop the mathematical foundations of the "karma games", by deriving rigorous and principled generalizations to complex resource allocation schemes. On the other hand, you will have to "engineer" these mechanisms and adapt them to practical application domains where they can be impactful and relevant.
You will collaborate with experts on these different topics, both in the two laboratories involved in the collaboration and withing the network of partners that form the NCCR Automation. You will publish your results in peer-reviewed journals and present them at international conferences.
As member of the NCCR Automation, you will benefit from the many opportunities offered by the NCCR (training, professional development, collaboration with researchers in academia and industry) and are expected to contribute to the NCCR activities when relevant (supervision of students, communication, technology transfer, equal opportunities, outreach).
A solid training in the mathematical tools that are needed for the project, including dynamical systems and control systems, is necessary. Knowledge of game theory, multi-agent systems, Markov decision processes, and any of the engineering domains mentioned above are valuable additions. The project will include some numerical experiments, therefore some experience with computational methods and numerical simulations is also useful.
Because the project goals lie at the intersection of engineering and economics, we expect the student to be interested in the science of socio-technical systems. As frequent collaborations are expected, a positive attitude towards team work, open communication, and collaborative science is expected.
We are offering a multifaceted and challenging position in a modern research environment with excellent infrastructure. The ideal starting date is the end of 2024 or the first months of 2025, with a planned duration of 4 years.
The supervision will be joint between the two labs, although formally the student will be affiliated with either the group of Prof. Florian Dörfler at the Automatic Control Laboratory or with the group of Prof. Emilio Frazzoli at the Institute for Dynamics Systems and Control.
In line with our values, ETH Zurich encourages an inclusive culture. We promote equality of opportunity, value diversity and nurture a working and learning environment in which the rights and dignity of all our staff and students are respected. Visit the NCCR Automation Equal Opportunities page to find out how we ensure a fair and open environment that allows everyone to grow and flourish.
chevron_right Working, teaching and research at ETH Zurich
We look forward to receiving your complete online application including
Please note that we exclusively accept applications submitted through our online application portal. Applications via email or postal services will not be considered. Questions regarding the position should be directed to Saverio Bolognani (bsaverio@ethz.ch) or Andrea Censi (acensi@ethz.ch); applications sent to these email addresses will not be considered.
Bitte beziehen Sie sich in Ihrer Bewerbung auf AkademischeStellen.com