[geos-devel] Precision/tolerance questions

[Hugo Mercier (ML)]

On 30/07/2015 12:03, Sandro Santilli wrote:
> On Thu, Jul 30, 2015 at 11:03:55AM +0200, Hugo Mercier (ML) wrote:
>> Hi all,
>>
>> I am not very familiar with the way GEOS can handle precision models ...
>>
>> If I am correct, having a "fixed" precision model, i.e. a regular grid
>> allows to define a tolerance where (point) coordinates will be snapped
>> before computations.
> 
> This is actually not always true. Input coordinates are tipically NOT
> snapped/rounder, while coordinates that are generated by internal
> operations will be. This is to avoid duplicating the work when the
> input is already rounded.

Ok, good to know.

>> I can read "Adding a PrecisionModel to the GEOS functionality uses a
>> precision grid to "snap" all coordinates, which could eliminate some
>> rounding and robustness issues with some functions. "
>> What rounding and robustness issues are they ? and which functions ?
> 
> Constructive functions can raise an exception when unable to deal with
> numerical instabilities. Intersection/Difference/Union...

And with a floating model, I guess there is no way to detect such
instabilities.

Thanks