30 arcsecond max

[Pedro Rodriguez]

Is there a way around the dmax of 30 arscseconds? In other words can you query for dmax equal to 40 or 50 or 60? The docs say you can't. What is the software limitation on this? Are there any overrides that allow you to go greater?

[Mario Juric]

Hi Pedro,
LSD partitions the sky into approximately rectangular overlapping
chunks of ~12deg^2 (by default). They overlap by 30arcsec. This ensures
that neighbor discovery near the edges will be done correctly, but only
if the search radius is less than 30arcsec. If it's greater than
30arcsec, some objects that are close to partition edges may be missed.

To go up to ~60", you'd need to rebuild your database with the overlaps
increased, by explicitly passing the margin width to
build_neighbor_cache(), or just changing the default of that argument.
There's currently no way to do this via the command line, and no way to
take an existing database and repartition it with greater overlaps.

PS: Also, note that increasing the overlaps will increase the amount of
space your database requires, because of the increased duplication.

Cheers,
--
Mario Juric,
Data Mgmt. Project Scientist, Large Synoptic Survey Telescope
Web : http://www.cfa.harvard.edu/~mjuric/
Phone : +1 617 744 9003 PGP: ~mjuric/crypto/public.key

Time to DM Summer 2012 'Data Release Ready' Date: 40 days

[Pedro Rodriguez]

So just to confirm, if I were to do what you lay out, you could match for greater than 30 arcsecond, losing no accuracy/matches. The only side effect being that searches would take more data storage and/or computing time? As long as there are no leak of matches, I am not concerned about increased usage fo disk space or computation time.

Thanks,

Pedro

--
Pedro Rodriguez
UCBerkeley 2014 | Physics/AstroPhysics
SnowGeek.org Founder
ski.ro...@gmail.com
208-340-1703

[Mario Juric]

On 6/22/12 16:05 , Pedro Rodriguez wrote:
> So just to confirm, if I were to do what you lay out, you could match
> for greater than 30 arcsecond, losing no accuracy/matches. The only side
> effect being that searches would take more data storage and/or computing
> time? As long as there are no leak of matches, I am not concerned about
> increased usage fo disk space or computation time.
>

Correct.

--
Mario Juric,
Data Mgmt. Project Scientist, Large Synoptic Survey Telescope
Web : http://www.cfa.harvard.edu/~mjuric/
Phone : +1 617 744 9003 PGP: ~mjuric/crypto/public.key

Time to DM Summer 2012 'Data Release Ready' Date: 35 days