verbs.batch_runner.batch_run#

batch_run(sim_func: Callable, *, n_steps: int, n_samples: int, parameters_samples: List[Dict], snapshot=None, cache=None, gas_priority: bool = False, n_jobs: int = -2, verbose: int = 10, **sim_kwargs)[source]#

Run a batch of simulations across a set or parameters

Convenience function to run simulations across a range of parameters and random seeds. Uses joblib to run simulation samples in parallel.

Will generate n_sample Monte-Carlo samples for each of the provided parameter samples.

The simulation environment for each sample can optionally be initialised from either a provided snapshot (generated using the verbs.envs.ForkEnv.export_snapshot() method), or a cache (generated using the verbs.envs.ForkEnv.export_cache() method).

Parameters:

sim_func –
Simulation execution function, should have the signature
```
def sim_func(
    env, seed, n_steps, **params, **sim_kwargs
) -> typing.Any:
    ...
```
where arguments are:
- env a simulation environment
- seed a random seed
- n_steps number of simulation steps
- **params keyword simulation parameters
- **sim_kwars keyword simulation fixed arguments/parameters
n_steps (int) – Number of simulation steps.
n_samples (int) – Number of monte-carlo samples to create for each parameter sample.
parameters_samples (list[dict]) – List of dictionaries containing simulation parameters. These parameters will be passed as keyword arguments to the sim_func function.
snapshot (tuple , optional) – Optional snapshot used to initialise the simulation environment for each execution.
cache (verbs.types.Cache, optional) – Optional cache used to initialise the simulation environment for each execution.
gas_priority (bool, optional) – If True gas-priority sorting will be used to order transactions in each new simulation step/block. Otherwise transactions will be randomly shuffled. Default value if False.
n_jobs (int, optional) – Number of jobs to run simultaneously, default is -2 i.e. 1 less than the number of available processors.
verbose (int, optional) – Verbosity of joblib logging, default is 10 ( full logging).
**sim_kwargs – Any additional keyword arguments passed directly to the sim_func function (i.e. they are shared across all the executions).

Returns:

list[dict] – List of results, grouped by their parameters. Each entry is a dictionary containing "params" the parameters to produce the samples, and "samples" the list of data produced by each Monte-Carlo sample.