Try to tune spark configuration using Sandia labs Dakota
Dakota needs several inputs.
- an input .in file which is documented here. This defines the optimization algorithms that Dakota will use to search the parameter space.
- An executable script that takes a parameter file as input and writes to standard out
- A parameter file to use as an example/template. The fields can be specified to Dakota by position
- A driver script, which can be a python2.7 file that the Dakota UI can generate using a wizard
- An example output file, or specification of the fields that are under analysis