Optimas exits if a history file is given, but not found

[delaossa]

Optimas exits if a history file is given, but not found.

optimas/optimas/explorations/base.py

Lines 133 to 148 in ed14c6e

	def _load_history(self, history: Union[str, np.ndarray, None]) -> None:
	"""Load history file."""
	if isinstance(history, str):
	if os.path.exists(history):
	# Load array.
	history = np.load(history)
	# Only include runs that completed
	history = history[history['sim_ended']]
	else:
	raise ValueError(
	'History file {} does not exist.'.format(history))
	assert history is None or isinstance(history, np.ndarray), (
	'Type {} not valid for `history`'.format(type(history))
	)
	return history

This current behaviour is problematic with my usual post processing scripts, as sometimes the optimas run script is accessed from a location that does not follow the relative path to the history file.
Previously in libE_opt, if the history file was not found, it would continue as if not given.

[AngelFP]

Hi @delaossa, thanks for reporting this. I think we could raise a warning instead of an error here.

[delaossa]

Ok, I would prefer it.. :)
Would be ok with just replacing ValueError by Warning?
By the way, why does the history argument in _load_history accept a np.darray?

[delaossa]

Raising Warning still exits the program.
#81 implements a solution using the warning package.

[AngelFP]

This should be solved now that #120 is merged. The recommended approach to analyze an optimas run is to use the new ExplorationDiagnostics class. (see https://optimas.readthedocs.io/en/latest/user_guide/basic_usage/exploration_diagnostics.html and https://optimas.readthedocs.io/en/latest/api/_autosummary/optimas.diagnostics.ExplorationDiagnostics.html#optimas.diagnostics.ExplorationDiagnostics)