On an Efficient Implementation of Tierra

In a project the author is involved with, the run time eeciency of Tierra has become of paramount importance. Whilst some optimisations can be applied to the Tierra code itself, the speedup is at most 2{3 times, with no beneets from vectorisation or parallelisation being realised due to the structure of the code. A new formulation of Tierra is proposed which gains the computational eeciency of a population model, without sacriicing the advantages of an individual model. This new formulation can take advantage of parallelism when available. In some preliminary tests, the new code is up to three orders of magnitude faster than Tierra.