Convex hull of my caches

Question

Hi, I have this idea, that it shouldn't be too difficult to compute the convex hull of all the caches found. That could be an interesting number on it's own, but would be even more awesome to have it as a badge, where one would need to get some percentage of globe's surface (2D in coordinates or any reasonable projection), or volume (taken as 3D convex hull). What do you think about it?

Answer 1

The idea to calculate a convex hull sounds interesting. But I am not sure how easy it would be to solve technically, and efficiant. I am sure it would be quite fast if there was better GIS-support in the database server where we have the cache-data, but that's pretty limited this far.

But I would love to experiment with it a bit when I get the time.

 

It won't be seen as another badge in the near future though. The badges we use, are from http://badgegen.com/ and our agreement with the author is that we do not add or change anything.

I will definitely keep this in mind.

EDIT:

Also the whole thing that the earth is a sphere makes it quite complex to do correct. It has to be a convex hull formula made for latitudes and longitudes. Which most likely would have been solved if our main database server engine had support for it, but it doesn't.

Comments

I'm glad you like the idea. Maybe after it would work here I can go lobbying to badgegen :)

I'm not sure which coordinate system would be the best, my ideas were:
1) 2D longitude-latitude - the simplest thing, there are simple and efficient algorithms to do that;
2) 2D in some other projection - partially solving the problem of unequal distances in 1), similarly, but there would need the projection to be done, a bit more of code
3) 3D using the real volume of the Earth - a bit more complicated, but I'm still sure efficient algorithms exist. There would be some coordinates transformation, but probably simpler than that of 2). This would have the great advantage of removing the singularity at W/E180 (e.g. doing Alaska+Siberia wouldn't generate much different score from Alaska+Canada with the same distances).

I understand you don't have these things in your engine, can you add it as an "external subroutine"? I would volunteer to implement it, if it uses some reasonable language...