ameisner
commented
8 years ago Hi Jeff,
Great questions. I think the answers have some subtleties/caveats, but I can give you some numbers anyway.
number of exposures --- Well, we'd ultimately like to run on ~10 million exposures (that number includes both W1 and W2). But for a single asteroid and for a timespan over which the apparent motion can be considered roughly linear, the number is ~12 exposures.
asteroids --- on average there are 5 already known asteroids that land within a given WISE exposure. However, just because a known asteroid falls within an exposure does not mean it will be detected by WISE, so the number of detected known asteroids per exposure could be somewhat less. Also, the number of asteroids per exposure is an extremely strong function of where the telescope is looking relative to the plane of the solar system. Way off the plane of the solar system, there could very well be one or zero known asteroids in a given exposure, and there could be a dozen near the plane.
quasars --- the number of quasars detectable in a WISE stack covering the footprint of a typical exposure will be ~30. However, most of these will be too faint to detect in a single exposure. So that brings up an interesting question of what we want to do about sources that are known to exist but won't be detectable in a single exposure. This same issue applies to stars and galaxies as well. I think we'll still want include them in the single-exposure modeling, right? Quasars will look morphologically exactly like stars in the WISE data, but will have different colors.
stars/galaxies --- The entire WISE source catalog (made from detections on stacks) has 747 million objects. Based on my experience, a substantial fraction of those will be bogus sources, but probably not more than a couple hundred million. So the nominal average number of static sources per exposure, based on the angular size alone, would be about 7000 stars+galaxies+quasars. But this is biased high to a large extent by the huge number of sources near the plane of the Galaxy, so I'd estimate the total number of static sources to be a couple thousand at high Galactic latitude. Again, most of these won't be detectable in a single exposure. I'd guess that 99% or more of the static sources have morphology indistinguishable from a point source, so in that respect it doesn't matter whether we know their star/galaxy/quasar classification. Of course, the expected source colors and photometric variability will depend on star/galaxy/quasar classification, so it would be great to take into account the source types as much as possible.
-Aaron
The WISE images in W1--W3 are 1016x1016 pixels and 508x508 pixels in W4.
I anticipate that current code won't run that quickly with full size images. Probably the outer loop in
compute_log_probabilityshould be over asteroids and the inner loop should be over pixels. I can do some other things to make it fast too if necessary.@ameisner , How many images might we want to process at once eventually, with
compute_log_probability? (How many images might a single asteroid appear in?) On average, roughly how many stars, galaxies, asteroids, and quasars appear in each of these image?