6 Conclusions

Using the theory of model parameter determination, we developed a general method which permits a quantitative comparison of observational techniques and optimization of complex observations through the relative figure of merit, defined as a generalization of the ratio of random errors of model parameters.

The method was applied to compare the outcome of OLBI and classical spectroscopic observations, both used to determine the parameters of the outflow of the P Cyg star. The observable quantities were computed using an efficient radiative transfer code and realistic grids of the envelope model. The OLBI and spectroscopic measurements errors were simulated assuming that the only source of errors is the counting statistics, i.e. we considered a nearly-ideal interferometer. Other main assumptions and simplifications of the present work were as follows: (1) the P Cyg wind is stationary and spherically symmetric; (2) the only OLBI observable considered here is the visibility modulus; (3) only the hydrogen H$\alpha$ line is considered.

The main conclusions of the present work are as follows:

1.

A meaningful evaluation of the OLBI vs. spectroscopy relative figure of merit requires exploration of a large domain of the model parametric space (about 10⁵ points computed in the present work);

2.

If the P Cyg outflow can be accurately described by a simple model of a stationary spherically symmetric isothermal envelope, then the use of interferometry does not substantially reduce the random error in the model parameter determination;

3.

If the P Cyg outflow requires more complex models, e.g. with a priori unknown dependence of the envelope temperature on radial distance, use of a nearly-ideal interferometer can reduce errors in model parameters by an order of magnitude as compared to the spectrometer of the same collecting power and noise level;

4.

The study of the relative robustness, i.e. the stability of the obtained model parameter determination with respect to systematic biases in the model, indicates that when the OLBI and the physical model reduce random errors, they also tend to reduce the systematic error;

5.

The optimal projected baseline lengths for observation of P Cyg with a nearly ideal interferometer lie in the range 45-90 m;

6.

If several model parameters are to be determined simultaneously, the OLBI observations at a single projected baseline cannot provide a substantial error reduction.