3. The reconstruction method
The eclipse mapping method (Horne 1985) employs the maximum entropy fitting package MEMSYS (Skilling et al. 1984) to interprete observed eclipse light curves in terms of intensity and brightness temperature maps of accretion disks. A cartesian two dimensional grid, located in the orbital plane and centered on the white dwarf undergoes an eclipse by the secondary star to produce an artificial lightcurve, whereby all pixels are assumed to have the same brightness at all phases. The intensities of the pixels are adjusted until the artificial light curve fits the observed one within a maximum allowed deviation. Since such a -fit does not produce a unique solution, an image-entropy is defined on the grid of pixels and maximized with respect to a default image. As a result a map which is as close as possible to the default map and which fits the data within the allowed -deviation is obtained.
The eclipse mapping method used by us is different with respect to the adopted geometry for the reconstructions. Instead of a flat grid, a three dimensional disk with a constant disk opening angle and an outward facing "ribbon" at a certain disk radius was used. Our improved method addresses two specific problems with the original eclipse mapping method. First, light curve variations due to anisotropic light sources like the so called "hump" which is caused by the hot spot where the gas stream hits the disk rim cannot be mapped due to the assumed isotropic radiation of the pixels. Second, there is only one-dimensional information available from the observed light curve for the reconstruction of the two-dimensional disk. Therefore additional constraints have to be introduced. This is done by manipulating the default image in a special manner, leading to the fact that the final result depends on the default image and the way it is created. For further details see Appendix A and B.
© European Southern Observatory (ESO) 1997
Online publication: April 6, 1998