Moved mortality to the end of processes to resolve order of competing… #292
Conversation
There was a problem hiding this comment.
Thanks Richard. This ordering makes much more sense to me. It appears to be having very little impact on transmisison as we hoped.
Not sure why the blank lines are showing up as differnces in this PR.
One suggestion would be to run a quick check on the resulting age distributions at the end of the runs, before and after this change. That would be the other obvious place mortality processes would show up potential differences?
There was a problem hiding this comment.
Looks OK to me! The plots look reassuring too! The only other check beyond @pwinskill's suggestion would be to perhaps look at the immunity profiles before/after changes to see whether they remain similar given your point about more susceptibles without immunity?
I think it's because Richard had deleted spaces on these lines? |
|
Tom is right re blank lines. I have a setting in RStudio that deletes trailing spaces when you save. I feel like it tidies things up a bit. Here are the age structure plots @pwinskill, they're pretty similar again (averaged over the last year for each age group). Although, now I've done it, I'm not sure the differences would be obvious here. Regardless of the position of mortality, the other processes don't interfere with the age, so an age won't be reset... I think it's the immunity profile that might look slightly different...? |
|
Thanks. Agree I wouldn't expect them to be different, but good to check! |
|
Thanks for approving, for fun I'll also just share the difference in immunities. |
Merge branch 'dev' into competing_hazards_mortality_order # Please enter a commit message to explain why this merge is necessary, # especially if it merges an updated upstream into a topic branch. # # Lines starting with '#' will be ignored, and an empty message aborts # the commit.
I have moved the mortality process to the end and created a couple of plots to show the simulation differences:
I ran 10 simulations for ten years over EIR =1,5,10,50,100 (human_population = 100) for each version.
Average with 95% CrI
competing_hazards_mortality_process_position.pdf
Timesteps with smoothed lines
competing_hazards_mortality_process_position_timesteps.pdf
Prevalence may be a little higher with mortality at the end due to more deaths leading to more susceptibles without immunity leading to more infections? but there's not much in it.
Any thoughts, please let me know!