Using RTrees for Faster Search

We do a lot of searching in Grasshopper and I mean a lot. Finding points that are closest to other points is one of the most common searches that I do. Say I have two geometry files, one is an IFC and another is a Rhino model (Don't worry, you don't have to know what these files are). I want model something in between the two.

How do I know which beam from the IFC "belongs" to which Rhino line ? Well, you have to do a . You have to figure out which IFC beam is closest to the Rhino line. That way for every IFC line, you can get the closest Revit line which will give you the pairs and the right data structure to start modelling.

Luckily for us, we don't have to think too hard about the math of this. Grasshopper gives us "Closest point" or "Curve closest point" to help us with this. The only problem is, if you have a lot of elements, it can be really slow to run these components.

But just by knowing a bit more about how computers work, there are other components that make this a lot quicker. If we stored the points in something called an RTree, the computer can compute closest points for us much quicker.

It changes how you manage your data in Grasshopper but it makes searching for the closest point exponentially faster. Especially if you have >5000 points that you need to look through. Checkout how I've implemented it here : Use Rtrees For Faster Closest Point Operations

If you're interested in reading more, try Introduction to Tree Data Structure - GeeksforGeeks. It help me understand more about it. But you don't actually need to know how the data structure works to use these components. Although, it does help if you need to build something on your own later.

Thanks for reading,

I'll see you next week.