Coral bleaching is poorly understood but devastating, killing reefs by making their coral expel the symbiotic algae that feed them. Scientists are still probing its causes. Were an accurate model of the phenomenon ever developed, the combinations of environmental variables would be too many to try manually, but the optimization library we developed for APES automatically evolves an initial guess about them into the right one by simulating a batch of parallel reefs across a range of conditions, comparing their growth to the target, and repeating with a new batch seeded with the initial conditions of the closest-scoring reefs until one gets within tolerance.
The most important influence on coral reef growth is ocean temperature, which fluctuates with some amplitude around a mean over each year. Coral has adapted to these fluctuations over countless millions of years, but global warming is heating the oceans, raising their mean temperature and therefore their peak temperature, which has become so hot that coral reefs worldwide bleach every summer. During literature review about the metabolic pathways responsible, we turned the problem on its head and asked, "What ocean temperature mean and amplitude does coral need to survive?", and our parallel optimization library delivered the answer instantly and easily for any timeframe and growth amount.
With just a few small CSV files and the lines of code to import the library and read them in, we had the parallel optimization library up and running, delivering answers as fast as we could ask questions. We had to code a simulation for the coral, of course, but that's a job for the automatic model fitting part of our software, which will require us to finish our data gathering and literature review first.