1997A&A...324..843A

M 31 is a very tempting target for a microlensing search of compact objects in galactic haloes. It is the nearest large galaxy, it probably has its own dark halo, and its tilted position with respect to the line of sight provides an unmistakable signature of microlensing. However most stars of M 31 are not resolved and one has to use the "pixel method": monitor the pixels of the image rather than the stars. AGAPE is the implementation of this idea. Data have been collected and treated during two autumns of observation at the 2 metre telescope of Pic du Midi. The process of geometric and photometric alignment, which must be performed before constructing pixel light curves, is described. Seeing variations are minimised by working with large super-pixels (2.1") compared with the average seeing. A high level of stability of pixel fluxes, crucial to the approach, is reached. Fluctuations of super-pixels do not exceed 1.7 times the photon noise which is 0.1% of the intensity for the brightest ones. With such stable data, 10 microlensing events are expected for a full "standard halo". With a larger field, a regular and short time sampling and a long lever arm in time, the pixel method will be a very efficient tool to explore the halo of M 31.