I am trying to generate toys from my signal like pdf constructed from the pyhf model. Following the example Running Toys by Hand here https://scikit-hep.org/pyhf/examples/notebooks/toys.html this seems to be straight forward by simply using mc_sig = pdf_sig.sample((n_samples,)).
However, I am unsure how the sampling is done exactly and the source code is not really helpful. Are the toys the result of only statistical fluctuations of the signal pdf or are the NPs (systematics) varied as well?
Thank you for your help!