Research
At present, I am mainly spending my research time on various aspects of Ergodicity Economics, some practical questions on renewable energy and some more fundamental problems in Machine Learning. At present, this is just a list of questions which will give you some idea. Some of these questions I am actively working on, usually with collaborators, but some are still exactly that: questions.
Ergodicity Economics
-
How do time and ensemble averages compare for stochastic processes that are intermediate between additive and multiplicative?
-
Can growth-rational decision strategies for repeated gambles be approximated by simpler heuristic strategies (in the sense of Gigerenzer et al.)
-
What can be understood quantitatively about the Reallocating Geometric Brownian Motion model when the reallocation parameter is negative?
-
How can we design mathematically consistent models of risk preferences and how can we design experiments to infer their parameters?
Renewable energy
-
Why do wind and solar energy systems exhibit the statistical fluctuations that they do?
-
How should the statistical fluctuations from individual devices be aggregated in space and in time and how do these aggregation processes differ (ergodicity again...)
-
Can we combine mathematical tools with data to make "better" operational predictions at the grid scale? Can we be clarify what "better" means in this context?
Machine learning and optimisation
-
Why does regularisation work?
-
How can reinforcement learning algorithms be adapted to maximise time averaged rate of reward rather than expected future discounted reward? Do such algorithms work better?
-
How do we incorporate multi-objective criteria into algorithmic decision making?