Not a current user.

Project

Starrfelt and Kokko (2012) were the first to recognize the impact of spatial and temporal structure of the environment on bet-hedging, demonstrating that different schemes of structure favored different strategies. While their model was conceptually important, it was not able to truly diagnose optimal strategies because the some strategies they considered had higher fitness across the entire range of environments than others. This gave an edge to the former strategies in outcomes of their model. Crowley et al. (unpublished model) formulated an improved model by normalizing the fitnesses of strategies across environments, so that the optimal strategy was only based off of spatiotemporal structure. Both of these models are built around a binary environment, where the environment takes on one of two values. Most real environments are not binary, but continuous. We are constructing a model based on a continuous environment for two reasons. First, it allows for more realism without sacrificing much interpretability. Second, a continuous model is likely to favor different strategies, namely conservative bet-hedgers, which is important specifically because the continuous model is more realistic. We coded our model in R. A crucial step in our model is optimizing (optimx R package) a complicated function with respect to several variables, which is too computationally intensive for a high-powered desktop. Instead, we will perform our calculations on the DLX cluster.

PI

Dr. Philip Crowley (PI, UK)

Students

Aviv Brokman (graduate student, UK)

Software

R

Funding

None needed

Publications from DLX

None yet

Browser not supported