Run configuration

Information that is shared across the whole simulation is held in the RunConfig class, a singleton initialised by RunConfigInit:

from exosim.utils import RunConfig

Parallel processing

ExoSim simulations are demanding, so parallel processing matters. Set the number of parallel processes with:

from exosim.utils import RunConfig

RunConfig.n_job = N

The value is applied to both joblib and numba.

Chunk size

The chunk size is the size of a chunk of a cached dataset (see Cached signals). Set it with:

from exosim.utils import RunConfig

RunConfig.chunk_size = N

where N is the chunk size in MB, applied for the whole environment.

Random seed and random generators

Set the initial random seed with:

from exosim.utils import RunConfig

RunConfig.random_seed = N

where N is the seed. By default the seed is None, so each simulation is unique.

ExoSim also provides a default random generator (numpy.random.Generator), already initialised with the current seed:

from exosim.utils import RunConfig

rng = RunConfig.random_generator

It is used like any other NumPy generator:

from exosim.utils import RunConfig

# uniform distribution:
RunConfig.random_generator.uniform(-1,0,1000)

# normal distribution:
RunConfig.random_generator.normal(0,1,1000)

# Poisson distribution:
RunConfig.random_generator.poisson(5, 1000)

More examples are in the numpy.random.Generator documentation.

ExoSim works on chunks of data and the generator may be called inside loops, so when the seed is not None, random_generator adds 1 to the seed at every call. This keeps the draws independent between chunks while staying reproducible.