Can I use r-tree indexes to find MAX(), MIN() values in arbitrary intervals quickly?

As far I can see, r-tree indexes are used today mostly for geographic systems, for spatial searches. I need some very similar, but not in space, but in 1-D time series. Imagine a *huge* time series, for example floating point data of a billions of point. *I need to answer quickly questions like* > *"what was the maximal value in the interval (1345223, 23234235)"?* Note, the DB is huge, a simple SELECT MAX(value) WHERE id BETWEEN 1345223 AND 23234235 is not acceptable - at least not if it needs a linear scanning of the interval. It must happen in same type of tricky index, most likely r-tree. Maybe it helps (but more likely not): as it is a time series table, it is inherently single-column. Other index types or other database engines are okay as well. I am investigating postgresql only because it is the smartest open source engine I know.