Just found this package for computationally expensive climate downscaling.. in R!

Offtopic
1

From the readme of ClimDown, a Nature-published package to downscale climatic data:

“On our Linux systems, such runs can take up to 7 days to complete. […]
In general, this downscaling technique is very expensive for large spatiotemporal domains. The more you can limit your domain, the faster your runtime will be.”

Well… the package is written in R! I had a look at the code, it seems the algorithms itself is coded in R, not just a wrap…
Even if indeed the project is a bit old, I couldn’t really stop thinking why they have chosen R for a computationally intensive task…

1 Like
2

I would guess that the authors of ClimDown chose R because it was the language they were very comfortable with.

3

Nice I’ve also seen some packages in mostly-R that do Monte Carlo simulation things and its just very slow. But hey if it’s feasible, a week or two is basically nothing just working on other things in the meantime.