GPJL incredibly slow at prediciton when noise_learn = true

CliMA / CalibrateEmulateSample.jl

Stochastic Optimization, Learning, Uncertainty and Sampling

https://clima.github.io/CalibrateEmulateSample.jl/dev

Apache License 2.0

81 stars 14 forks source link

GPJL incredibly slow at prediciton when noise_learn = true #267

Open odunbar opened 7 months ago

odunbar commented 7 months ago

It seems like when GPJL predicts with noise_learn = true it is incredibly slow, but noise_learn = false is fast. This should be investigated as to whether it is something that can be fixed in CES.jl or is on the GaussianProcesses.jl side.