Weird parameter vector evaluation failure

I would like to report a weird behaviour of the pypesto objective.

Minimal example code

I attach a minimal example in which this occurs:

• Executing success.py all runs fine
• Executing error.py an error occurrs in the last objective evaluation (which was successful in the previous script).

The error that occurs is

Traceback (most recent call last):
  File "/home/elba/PhD/IntegrationScaling/PYTHON/GRID/error.py", line 80, in <module>
    res = problem.objective.call_unprocessed(next_x, [0], 'mode_fun')
  File "/usr/local/lib/python3.9/dist-packages/pypesto-0.2.7-py3.9.egg/pypesto/objective/aggregated.py", line 87, in call_unprocessed
    return aggregate_results([
  File "/usr/local/lib/python3.9/dist-packages/pypesto-0.2.7-py3.9.egg/pypesto/objective/aggregated.py", line 88, in <listcomp>
    objective.call_unprocessed(x, sensi_orders, mode, **kwargs)
  File "/usr/local/lib/python3.9/dist-packages/pypesto-0.2.7-py3.9.egg/pypesto/objective/amici.py", line 337, in call_unprocessed
    self.apply_steadystate_guess(data_ix, x_dct)
  File "/usr/local/lib/python3.9/dist-packages/pypesto-0.2.7-py3.9.egg/pypesto/objective/amici.py", line 385, in apply_steadystate_guess
    (x_sim - guess_data['x'])[
IndexError: arrays used as indices must be of integer (or boolean) type

And only occurs when the first objective is evaluated with the gradients, then the second fails (just execute the notebooks). The only difference between success.py and error.py lies in line 77 and the failure comes in the second objective evaluation.

Thanks!

EDIT: This next_x parameter vector is a new guess that is proposed during parameter profile and leads to failure.

[elbaraim]

solved

[elbaraim]

okay… I have

>>> import pypesto
>>> pypesto.__version__
'0.2.7'
>>> import amici
>>> amici.__version__
'0.11.13'
>>> import petab
>>> petab.__version__
'0.1.14'

I will update amici and petab and see if that solves the issue