Semi-Analytical Models (SAMs) are tools that allow us to study how galaxies form and evolve in a simulated Universe. Because they implement a big number of physical processes, with various degrees of freedom, they allow astronomers to explore how this formation and evolution of galaxies varies as different parameters change.
This big number of parameters also means that finding the “optimal” set of parameters that produces the “best looking” SAM outputs is impossible (because the parameter space is too big). A common technique in these cases is to use heuristics (e.g., MCMC, PSO, genetic algorithms, etc.) to explore the parameter space more efficiently and find local optima.
During this project, a tool will be developed to execute such heuristics over shark, a new SAM developed at ICRAR. This tool needs to run shark, read its outputs, assess how “good” they look, calculate a new set of parameters (using some heuristic) that will improve the results, and run again. The tool will be written in Python, so some basic python coding skills are necessary. It also needs to correctly integrate itself with HPC centers, like Pawsey and our local pleiades cluster, but previous HPC knowledge is not required.
