Sorry if this is an insultingly obvious concept, but it's something I haven't done before and I've been unable to find any material discussing the best way to approach it.

I'm wondering what's the best data structure for holding a 2D grid of unknown size. The grid has integer coordinates (x,y), and will have negative indices in both directions.

So, what is the best way to hold this grid? I'm programming in c# currently, so I can't have negative array indices. My initial thought was to have class with 4 separate arrays for (+x,+y),(+x,-y),(-x,+y), and (-x,-y). This seems to be a valid way to implement the grid, but it does seem like I'm over-engineering the solution, and array resizing will be a headache.

Another idea was to keep track of the center-point of the array and set that as the topological (0,0), however I would have the issue of having to do a shift to every element of the grid when repeatedly adding to the top-left of the grid, which would be similar to grid resizing though in all likelihood more frequent.

Thoughts?