Closed bdeonovic closed 7 years ago
A few comments on the pull request:
n_sims
argument in the discretediag()
function to nsim
to match the argument in ABC()
which similarly controls the number of generated datasets, as opposed to introducing a new argument name. Update code and documentation files accordingly.discretediag()
attempts to automatically subset its AbstractChain argument to only those parameters with discrete support. The package cannot do this reliably, and may include parameters that are not discrete (pollution.jl example) or miss ones that are. The user will be able to make the proper distinction and can subset chains in a straightforward fashion (e.g. discretediag(sim1[:, :gamma, :])
). Consider removing the automation. Update code and examples accordingly.Issues addressed
Squashed and merged into master now.
Here is the discrete diagnostic code Brian. Let me know what you think needs fixing up/cleaning up. I didn't notice any tests or docs for the other diagnostics so I did not include any either.