search

epinowcast / primarycensoreddist

Primary event censored distributions in R and Stan.
https://primarycensoreddist.epinowcast.org/dev/
Other
7 stars 1 forks source link

Check why it works for maths lines we can do without #84

Open seabbs opened 1 week ago

seabbs commented 1 week ago

I.e in the analytical solutions there are a few substitutions where we first write out the full equation vs just subbing in our already solved component chunks. The fear is this might be confusing to readers and might not really add much. Based on #68

seabbs commented 1 week ago

Refers to this: https://github.com/epinowcast/primarycensoreddist/pull/68#discussion_r1765260060

and similar.

Potentially this can be reviewed along side #72

seabbs commented 1 week ago

@SamuelBrand1 did you catch this in your PR? I think no?

SamuelBrand1 commented 1 week ago

No.

Maybe @parksw3 could give a bit of feedback? I'm possible a bit too close to decide what reads well/ is useful vs not!

parksw3 commented 1 week ago

Sorry for the confusion, but should I read the math or code? If you could point me to where to look, that would be helpful... the title sounds like simplifying math but the comments sound like simplifying the code...

seabbs commented 1 week ago

The maths. The link is to the Rmd doc so we can get the line number 2017

seabbs commented 1 week ago

Essentially it boils down to should we get rid of the first term in 3.9 because we actually arrive at the second line by subbing the components we have just derived into the last line of 3.5

and the same point for the lognormal

seabbs commented 12 hours ago

Lack of traction here. As this is relatively easy to close shall we either PR a quick change or close this as won't do

SamuelBrand1 commented 12 hours ago

Hey @seabbs . I think there is a tension here because we also want to be clear... replication not necessarily bad in the maths.

seabbs commented 12 hours ago

My point is that I think in these instances it is unclear as it adds additional lines of maths for readers to take in that don't actually advance the argument and aren't clearly signposted. So from my view solutions would be either to more clearly signpost the steps/links or to remove.