model response vector and experiment vector shape do not match with multiple sensors

[shangwjen]

For probeye 3.0.4, in the situation of multiple output sensors with a single correlation results in a shape mismatch error:

This occurs due to the list concatenation of the values from the multiple output sensors.

Is this intended? And if yes, what would be a solution to make this situation work?

To reproduce the error edit test_1D_correlation_multiplicative_model_error from line 171 to 187 with:

for i in range(n_experiments):
    y_test = np.random.multivariate_normal(mean=y_true, cov=cov)
    problem.add_experiment(
        name=f"Test_{i}",
        sensor_data={
            "x": x_test,
            "y": y_test,
            "y1": y_test+1,
        },
    )
    if plot:
        plt.scatter(
            x_test,
            y_test,
            label=f"measured data (test {i + 1})",
            s=10,
            zorder=10,
        )

where "y1": y_test + 1 is added.

and for line 205 replace with:

self.output_sensors = [Sensor("y", std_model="sigma"), Sensor("y1", std_model="sigma")]

where Sensor("y1", std_model="sigma") was added to the line.

[joergfunger]

What exactly do you expect? If the two sensors are supposed to be uncorrelated then this is somehow similar to this. In case they are correlated, how would you like to define the covariance matrix, something like ( [sigma_y1, sigma_y2]^T cor[cor_x, cor_x; cor_x,cor_x] [sigma_y1, sigma_y2]), where cor_x is the correlation between a single set of measurements based on x?