Introduction

 Long baseline optical interferometry is still in its infancy, with aperture count currently limited to 3. The feasibility of using numerous apertures, spaced kilometers apart, to obtain high-resolution snapshot images was initially debated ([Traub 1986,Beckers 1986,Beckers 1997]). Fizeau interferometry, the only beam-combining scheme then known to provide snap-shot images did not appear usable with highly-diluted apertures. Only recently was a solution found ([Labeyrie 1996]), in the form of ``densified-pupil imaging'', a beam-combining scheme which can achieve efficient imaging on compact objects.

With this scheme, ways of using diluted apertures spanning up to 100,000 km are foreseen ([Labeyrie 1998]). It will require active or adaptive phasing, like in monolithic telescopes.

The closure phase method ([Jennison 1958]), widely used in the radio interferometry domain and also recently used in the COAST optical interferometer ([Baldwin et al. 1996]) for observing the binary star Capella, is suitable for phasing diluted arrays but requires individual phase measurements on all the baselines taken into account for the closures. In our case, this would imply a pair wise recombining technique for a number of baselines much larger than the 13 groups of apertures used for our approach to cophasing the array.

We have developed a different algorithm, suggested by one of us (A.L.), and possibly well suited for this type of instrument. We describe the algorithm in section  4, results of numerical simulations in section 5, and discuss the limitations in section 6.