Further Investigations of Awmp Strike Limit Algorithms and Their Optimisation

We review and enhance several potential Aboriginal Whaling Management Procedure (AWMP) Strike Limit Algorithms (SLAs), and examine their performance on the extreme B3 and B7 shery type 2 Initial Exploration Trials. We focus on optimisa-tion methods described by Givens (1997b) for improving the performance of existing SLAs and creating enhanced, merged SLAs from existing candidates. The statistical motivation for this approach is reviewed. Optimisation of existing SLAs was very eeective. For example, by applying this approach to the Punt and Butterworth (1997) SLA, we created a new variant which is superior in the B3 trial and which provides 50% more need satisfaction on the B7 trial while reducing the nal stock size by only 0.01K below what the original SLA achieved, based on 5th percentile performance. In an example we considered for merging two SLAs, the approach permitted us to re-balance extinction risk and need satisfaction. However, the H-optimisation of the individual SLAs had been so successful that their subsequent merging was not signiicantly better, unlike a past example (Givens, 1997b) where more diverse SLAs were greatly improved by merging. The optimisation method examined here isolates IET simulation outside the opti-misation loop. It uses an objective function that can be within 4% of the one used in a standard approach; but to get the exact answer, the standard approach can require 600,000% more computing time. The method described here therefore allows preliminary research eeort to be concentrated on a wide-ranging and intensive search for optimal SLA performance.