The function gam::gam() allows you to adjust the shrinkage parameter for a random effect as such:
fit = gam::gam(y ~ 1 + random(fac, df=5))
Is there a similar way to adjust the shrinkage parameter for random effects in mgcv? I'm not interested in setting the penalty to a fixed value by using sp. Instead, I wish to adjust the prior shrinkage before fitting, similar to how k works. I can set gamma to a value >1, but that will affect all smooths in the model. I can set min.sp, but that does not work with method='fREML' for bam() (I have a lot of smoothing parameters). Are there any additional options for gam(), gamm(), or bam()?