Thank you for making this repository available. I have a question regarding the projection function argument projfn.
The description from section 3.1 (Positive Definite) in your paper corresponds (I think) to the argument "PSD-REHU" (line 203 in stabledynamics.py):
elif props["projfn"] == "PSD-REHU":
V = MakePSD(ICNN([lsd, ph_dim, ph_dim, 1], activation=ReHU(float(props["rehu"]) if "rehu" in props else 0.01)), lsd, eps=projfn_eps, d=1.0)
but the default setting provided in train_many uses NN-REHU. Can you please clarify which projection function you used for the results in the paper? Thanks.