[RTR] fix: parameter objective need parameters decision variables in any scenario right ?

[Ipuch]

I've built an example on top of your example @EveCharbie. The Idea is to have an objective with the parameter in it to see it evolve through the iterations of IPOPT, to see if the sol output is the same as the one of IPOPT.

While I had only one objective, the objective was not taken into account. I manage to do a quick fix, with no ideas of the consequences. Some choices may have some rational and need to be discussed in the next PR Review @pariterre

@p-shg, you can build on top of this PR to see through ipopt how the scaling affect the parameter.

On top of :

• [WIP] Made a proper parameter scaling example #929

---

This change is 

[codecov]

Codecov Report

Attention: Patch coverage is 76.47059% with 4 lines in your changes missing coverage. Please review.

Project coverage is 77.75%. Comparing base (e7494ca) to head (55c410e).
Report is 19 commits behind head on master.

Files with missing lines	Patch %	Lines
bioptim/optimization/parameters.py	81.25%	3 Missing ⚠️
bioptim/interfaces/interface_utils.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #930      +/-   ##
==========================================
+ Coverage   77.62%   77.75%   +0.12%     
==========================================
  Files         156      156              
  Lines       18035    18040       +5     
==========================================
+ Hits        14000    14027      +27     
+ Misses       4035     4013      -22     

Flag	Coverage Δ
unittests	77.75% <76.47%> (+0.12%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[pariterre]

Reviewed 3 of 4 files at r1, 3 of 3 files at r2, all commit messages.
Reviewable status: all files reviewed, 4 unresolved discussions (waiting on @Ipuch)

---

bioptim/optimization/parameters.py line 304 at r2 (raw file):

    def unscaled(self):
        """
        This method allows to intercept the scaled item and return the current node_index

Add a comment:

The return parameters is actually the scaled version of it. But for internal reasons, it is expected. This can be confusing for the user who expects an unscaled version though. Therefore, .unscaled should only be call by the programmer who knows what there are doing

---

bioptim/examples/getting_started/example_parameter_scaling.py line 183 at r2 (raw file):

    g_scaling = VariableScaling("gravity_xyz", np.array([1, 1, 10]))  # Does not converge to the right place
    #  "Optimal parameters unscaled: {'gravity_xyz': array([ 0.        ,  5.00000002, -4.9999999 ])}"
    #  "Optimal parameters scaled: {'gravity_xyz': array([ 0.        ,  5.00000002, -0.49999999])}"

remove this

---

bioptim/examples/getting_started/example_parameter_scaling.py line 213 at r2 (raw file):

    dynamics = Dynamics(
        DynamicsFcn.TORQUE_DRIVEN,
        # state_continuity_weight=100,

Reinstate

---

bioptim/interfaces/interface_utils.py line 388 at r2 (raw file):

        p = PenaltyHelpers.parameters(
            penalty, penalty_idx, lambda p_idx, n_idx, sn_idx: _get_p(ocp, p_idx, n_idx, sn_idx, scaled)
        )

test this