verify_convergence option in optimize_yaw_angles not working

[fwasilcz]

Hi,
I run into some problems during yaw optimization excercise and to get to the bottom of it I tried turning on "verify_convergence" option in optimize_yaw_angles method.
My first question is - how is the convergence verified, I did not find anything about it.
Secondly, when I turned the option on, the optimizer returns an error (posted below). The error is generated only by the "verify convergence" option, with it turned off the code works fine, so I'm guessing this may be a bug.

Optimizing yaw angles with 16 workers.
_RemoteTraceback:
"""
Traceback (most recent call last):
File "C:\Users\Filip_Win\miniconda3\envs\floris4.0\Lib\concurrent\futures\process.py", line 263, in _process_worker
r = call_item.fn(*call_item.args, **call_item.kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Filip_Win\miniconda3\envs\floris4.0\Lib\concurrent\futures\process.py", line 212, in _process_chunk
return [fn(*args) for args in chunk]
^^^^^^^^^
File "C:\Users\Filip_Win\miniconda3\envs\floris4.0\Lib\site-packages\floris\parallel_floris_model.py", line 48, in _optimize_yaw_angles_serial
df_opt = yaw_opt.optimize(print_progress=print_progress)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Filip_Win\miniconda3\envs\floris4.0\Lib\site-packages\floris\optimization\yaw_optimization\yaw_optimizer_sr.py", line 310, in optimize
df_opt = self._finalize()
^^^^^^^^^^^^^^^^
File "C:\Users\Filip_Win\miniconda3\envs\floris4.0\Lib\site-packages\floris\optimization\yaw_optimization\yaw_optimization_base.py", line 388, in _finalize
self._verify_solutions_for_convergence(
File "C:\Users\Filip_Win\miniconda3\envs\floris4.0\Lib\site-packages\floris\optimization\yaw_optimization\yaw_optimization_base.py", line 566, in _verify_solutions_for_convergence
jj = (ids_max_loss[0][0], ids_max_loss[1][0])
~~~~~~~~~~~~^^^
IndexError: tuple index out of range
"""

The above exception was the direct cause of the following exception:

Traceback (most recent call last):

File ~\miniconda3\envs\floris4.0\Lib\site-packages\spyder_kernels\py3compat.py:356 in compat_exec
exec(code, globals, locals)

File d:\wyniki\wiatr\analizy_ekonomiczne\definicja_farmy\floris4.0\calculate_and_save_optimal_yaw_10deg_simple.py:118
df_opt = pfmodel.optimize_yaw_angles(

File ~\miniconda3\envs\floris4.0\Lib\site-packages\floris\parallel_floris_model.py:509 in optimize_yaw_angles
df_opt = pd.concat(df_opt_splits, axis=0).reset_index(drop=True).sort_values(

File ~\miniconda3\envs\floris4.0\Lib\site-packages\pandas\core\reshape\concat.py:382 in concat
op = _Concatenator(

File ~\miniconda3\envs\floris4.0\Lib\site-packages\pandas\core\reshape\concat.py:445 in init
objs, keys = self._clean_keys_and_objs(objs, keys)

File ~\miniconda3\envs\floris4.0\Lib\site-packages\pandas\core\reshape\concat.py:504 in _clean_keys_and_objs
objs_list = list(objs)

File ~\miniconda3\envs\floris4.0\Lib\concurrent\futures\process.py:642 in _chain_from_iterable_of_lists
for element in iterable:

File ~\miniconda3\envs\floris4.0\Lib\concurrent\futures_base.py:619 in result_iterator
yield _result_or_cancel(fs.pop())

File ~\miniconda3\envs\floris4.0\Lib\concurrent\futures_base.py:317 in _result_or_cancel
return fut.result(timeout)

File ~\miniconda3\envs\floris4.0\Lib\concurrent\futures_base.py:449 in result
return self.__get_result()

File ~\miniconda3\envs\floris4.0\Lib\concurrent\futures_base.py:401 in __get_result
raise self._exception

IndexError: tuple index out of range

Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.
Verifying convergence of the found optimal yaw angles.

[misi9170]

Hi @fwasilcz ,

Thank you for your question. If I'm understanding correctly, you are using the optimize_yaw_angles() method on the ParallelFlorisModel (with verify_convergence=True) when you run into this issue. This does look like a bug: verify_convergence is simply passed on to the YawOptimizationSR class and finally the base YawOptimization class, which then runs the verification procedure here when finalizing the yaw angles. It seems that, for some reason when this is running in parallel, there is a bug in the indexing.

All that being said, we have recently deprecated the ParallelFlorisModel class and recommend that you try using its replacement, ParFlorisModel, instead. This class is simply a drop-in replacement for the standard FlorisModel that can be passed directly to a yaw optimizer; that is, rather than having a special method optimize_yaw_angles, you can pass an instantiated ParFlorisModel directly to YawOptimizationSR. See here for an example of how to do this.

Because our support has shifted to the more robust and interchangeable ParFlorisModel, we may not fix the bug you have identified. However, please do let us know if you run a YawOptimizationSR optimization using the ParFlorisModel with verify_convergence=True and still see the issue!

Best,
Misha

[fwasilcz]

Hello Misha,
Thank you very much for your reply! I was unaware that there is a new class for parallel simulations. I'll try it out, see if problems persist and get back to you.
Regards
Filip