Horseshoe prior for logistic regression in brms

  • Operating System: windows 10
  • brms Version: 2.3.1

I have been reading this paper, along with the supplement to try to understand the horseshoe prior for logistic regression in a case where I have around 30 predictors. I am having trouble understanding how to implement this prior in brms (even though the documentation for brms is excellent). I’m very new to using brms and Stan (and modeling in a Bayesian framework) in general.

Would someone be able to specify the prior in brms in the above paper for the the classification example? This might help me understand how to use brms to implement the same kind of model and prior. Here is the supplementary section with Stan code

I was thinking the prior in brms would look something like this: horseshoe(1, scale_global = 0.5). scale_global would be obtained by plugging in sigma=2 and then the appropriate D, n, and p into equation 16 from the paper, correct?
Sorry for the simple question. Thanks for the help!