5 Conclusion

A new method was introduced to solve the configuration problem. It is based on the computation of pressure forces emanating from the discrepancies between the model and actual distribution of Fourier samples. The array elements submitted to these forces move toward optimal positions. The efficiency, rapidity and flexibility of the method was demonstrated. Its implementation in a C++ library able to handle most of the observational situations like synthesis, mosaicing, multi-configuration as well as terrain constraints constitutes a powerful tool for array design. A procedure to manage multiple scientific goals and source positions was given in that purpose.

Such an approach centered on the scientific purposes and able to handle a large number of antennas can contribute to the design of new generation interferometers like ALMA (64 antennas and 256 pads) or ATA (350 antennas). In the case of ALMA, even after the construction of the pads the degree of freedom in positioning the antennas for an observation will be high and it might be useful to have a software able to find out which pads should be optimally used for a single observation and a single scientific purpose. APO could also achieve this task.

Acknowledgements

I am grateful to the referee T. Cornwell for recommendations that helped to considerably improve the clarity of this paper. I thank J. Lequeux and A. L. Melchior for careful reading of the manuscript and encouraging comments.