Try presimulation with steady state algorithms.

Also, better error message for failing because of NaN's.

biosimulators_tellurium/core.py

@@ -273,12 +273,39 @@ def exec_sed_task(task, variables, preprocessed_task=None, log=None, config=None
 
  results = numpy.array(road_runner.simulate(sim.output_start_time, sim.output_end_time, sim.number_of_steps+1).tolist()).transpose()
  else:
- road_runner.steadyState()
- results = road_runner.getSteadyStateValues()
+ results = None
+ simdists = [0, 0.1, 1, 10, 100, 1000]
+ sd = 0
+ lasterr = ""
+ while sd < len(simdists) and results is None:
+ try:
+ if simdists[sd] > 0:
+ road_runner.resetAll()
+ if model.changes:
+ for change in model.changes:
+ component_id = preprocessed_task.model_change_target_tellurium_id_maps[task.id][(change.model, change.target, change.symbol)]
+ new_value = float(change.new_value)
+ road_runner[component_id] = new_value
+ road_runner.simulate(end=simdists[sd])
+ road_runner.steadyState()
+ results = road_runner.getSteadyStateValues()
+ except Exception as e:
+ lasterr = str(e)
+ sd += 1
+ if results is None:
+ msg = 'Steady state analysis failed with algorithm `{}` ({}):'.format(
+ preprocessed_task.algorithm_kisao_ids[task.id],
+ KISAO_ALGORITHM_MAP[preprocessed_task.algorithm_kisao_ids[task.id]]['id'])
+ msg += "\n '" + lasterr + "'"
+ for i_param in range(preprocessed_task.solvers[task.id].getNumParams()):
+ param_name = preprocessed_task.solvers[task.id].getParamName(i_param)
+ msg += '\n - {}: {}'.format(param_name, getattr(preprocessed_task.solvers[task.id], param_name))
+ raise ValueError(msg)
 
  # check simulation succeeded
  if config.VALIDATE_RESULTS and numpy.any(numpy.isnan(results)):
- msg = 'Simulation failed with algorithm `{}` ({})'.format(
+ msg = 'Simulation failed: ' + str(numpy.count_nonzero(numpy.isnan(results))) +\
+ ' nan value(s) found in results with algorithm `{}` ({})'.format(
  preprocessed_task.algorithm_kisao_ids[task.id],
  KISAO_ALGORITHM_MAP[preprocessed_task.algorithm_kisao_ids[task.id]]['id'])
  for i_param in range(preprocessed_task.solvers[task.id].getNumParams()):