Improve error message for probability constraint

[ChristianZimpelmann]

• optimagic version used, if any: 0.5.0

What would you like to enhance and why? Is it related to an issue/problem?

The following code fails with the message TypeError: only integer scalar arrays can be converted to a scalar index.

import numpy as np
import optimagic as om


def fun(params):
    offset = np.linspace(1, 0, params.shape[0])
    x = params["value"] - offset
    return x @ x


params = pd.DataFrame(
    {
        "value": [0.3, 0.2, 0.5],
        "lower_bound": [0, 0, 0],
        "upper_bound": [1, 1, 1],
    },
    index=["pi_1", "pi_2", "pi_3"],
)
res = om.minimize(
    fun=fun,
    params=params,
    algorithm="scipy_lbfgsb",
    constraints=om.ProbabilityConstraint(selector=lambda params: params[:3]),
)

The problem seems to be that the bounds in the params DataFrame are not allowed (although they are also implemented by ProbabilityConstraint).

Building params as follows works fine:

params = pd.DataFrame(
    {
        "value": [0.3, 0.2, 0.5],
        "lower_bound": [np.nan, np.nan, np.nan],
        "upper_bound": [np.nan, np.nan, np.nan],
    },
    index=["pi_1", "pi_2", "pi_3"],
)

Describe the solution you'd like

Catch exception and improve error message.

Alternative: allow for specifying the bounds. This would require some thought on how to handle cases in which other bounds than 0, 1 are specified in the DataFrame.