yosh

Mikola is now going through how k-d trees work. This is going pretty deep haha.

"So far we've only been talking about basic data structures. This is because IPFS itself is a functional data structure" - making sense here

Mikola now touching on the route planning aspect. Points out that systems already exist that run on OSM - it's a solved problem

And now we're talking Dijkstra / A* path finding algorithms. From the person that wrote one of the fastest pathfinding algorithms ever

"2 hop hub labelling" - hardest word in two days, and we've talked about a lot of complex things lol

So fast path finding can be done by pairing "2 hop hub labelling" and A* - scales linearly in mem, sub ms query times

Only downside is that you need about a TeraByte of RAM to build the index - hahahaha; "we should just buy this computer" - lol Mikola

Apparently that's how Microsoft got the 2HHL algorithm to run - 1 terabyte of RAM