[Kstars-devel] A couple more thoughts on proper motion

[James Bowlin]

It would be interesting to see how much the duplication changes if we 
reduce the time span to, say, +/- 5,000 years.  I think the dependence 
on timespan would have a quadratic component because of the +/- 25 
arcminute safety margin but I don't know how strong it is.  I leave it 
to the astronomers among us to decide where to set the timespan we 
finally use.

Another thought is that we might want to try putting some of the most 
duplicated stars into the named star index.  I don't think this will 
create a problem with the mag limit like when we tried to move named 
stars into the unnamed star index (or something like that).  Could we 
significantly reduce the duplication if we put, say, the 1,000 most 
duplicated stars into the named star list?   I'm not certain this is a 
winning strategy because the cost of periodic re-indexing may exceed 
the cost of the duplicate drawing.  But the drawing is done all the 
time and the re-indexing is only done as needed.

Also, another way to possibly reduce some of the cost of duplication is 
to give each star a drawID field so we can ensure that each star is 
only drawn once per draw cycle.  With a lot of duplication this is
a very good strategy but with little duplication it is possible that
the cost of checking the drawID for every star is greater than the cost
of simply drawing the duplicates.


-- 
Peace, James