Error in pmax(t(mm), object$featuremins[names(object$betas)])

[nhelm]

Hi,

I am having an issue when attempting to project one of my models to new areas.

I used ENMeval and Maxnet (not maxent.jar) to develop multiple models using different spatially structured cross validation methods to evaluate transferability. I used the same occurrence data, background points, and environmental variables for each cross validation method. My background selection came from buffered areas around location points (hence the need to project to new areas, in this case the historical range of my species of interest).

I ran a grid search on the regularization parameter and examined different feature classes (using ENMeval). I then selected the optimal model based on avg. test AUC (for each spatially structured cross-validation method).

After selecting the optimal model I projected the results to new areas For some reason, all of my models projected fine with the exception of one. Each time I try to project using:

project<-maxnet.predictRaster(opt_model,variables,
type=c('exponential'),clamp=TRUE)

I get this error:

Error in pmax(t(mm), object$featuremins[names(object$betas)]) :
long vectors not supported yet: ../../../../R-3.5.3/src/include/Rinlinedfuns.h:519

I have checked and rechecked the optimal model object to ensure I am using the correct betas. This particular optimal model was the most complex and so it used all of the environmental variables. Any thoughts on what may be happening?

Thanks,

Nolan