pfmc-assessments / geostatistical_delta-GLMM

Tool for geostatistical analysis of survey data, for use when estimating an index of abundance

20 stars 17 forks source link

Closed EllenY closed 6 years ago

EllenY commented 6 years ago

I am running VAST with 1000 knots for 11 species in RStudio. How can I speed up the Optimize process for estimating the final model?

5.2 Estimate fixed effect and predict random variables

Opt = TMBhelper::Optimize( obj=Obj, lower=TmbList[["Lower"]], upper=TmbList[["Upper"]], getsd=TRUE, savedir=DateFile, bias.correct=FALSE )

James-Thorson commented 6 years ago

Easiest fix: Install and use Revolution R, which will speed up by factor approx. equal to number of processors on machine.

Other fixes:

Decrease number of knots
Decrease number of species
Decrease number of years
Decrease / change number of factors, e.g., to start with 1 factor for each component and increase slowly until new factors explain small proportion of total variance, as I did in Thorson et al. 2016 Global Ecology and Biogeography
Use bigger machine
Use amazon cloud computer