Adjusting the parameters for QuantumMDM (class and pipeline)

[toncho11]

This is PR is part of #156

• Removing un unused parameter "gen_feature_map" for QuantumMDM
• Adding a new parameter "upper_bound" for QuantumMDM

[gcattan]

Thanks @toncho11 ! Feel free to merge when you are ready

[toncho11]

There is some kind of problem. It prevents classify_P300_bi_quantum_mdm.py from completing.

 File "C:\Work\PythonCode\pyRiemann-qiskit\pyq_env\Lib\site-packages\docplex\mp\mfactory.py", line 362, in _expand_bounds
    fn_bounds = [var_bound(k) for k in keys]
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Work\PythonCode\pyRiemann-qiskit\pyq_env\Lib\site-packages\docplex\mp\mfactory.py", line 362, in <listcomp>
    fn_bounds = [var_bound(k) for k in keys]
                 ^^^^^^^^^^^^
  File "C:\Work\PythonCode\pyRiemann-qiskit\pyq_env\Lib\site-packages\pyriemann_qiskit\utils\hyper_params_factory.py", line 132, in <lambda>
    return lambda n_features: ZZFeatureMap(
                              ^^^^^^^^^^^^^
  File "C:\Work\PythonCode\pyRiemann-qiskit\pyq_env\Lib\site-packages\qiskit\circuit\library\data_preparation\zz_feature_map.py", line 99, in __init__
    if feature_dimension < 2:
       ^^^^^^^^^^^^^^^^^^^^^
TypeError: '<' not supported between instances of 'tuple' and 'int'

It a feature map function is always assigned in the base class even if not needed.

[gcattan]

Ok, I see. There are some leftovers in the pipelines.py script for QuantumMDM pipelines.
Can you try to replace them with the upper-bound coefficient?

[gcattan]

Sorry, it am mistaking. My precedent comment does not apply.

[toncho11]

It seems the problem is in the voting classifier. It still sets "gen_feature_map" for QuantumMDMWithRiemannianPipeline while this is indeed the "upper_bound" parameter now.