Add statistical tests for infection loop. Closes #52. #60
Conversation
ChiragKumar9
force-pushed
the
ckk_infection_loop_statistical_test
branch
from
26ba937 to
c3d42ca
Compare
src/infection_propagation_loop.rs
Outdated
| context.add_plan_with_phase(1.0, ixa::Context::shutdown, ExecutionPhase::Last); | ||
| // Add a bunch of people so we mimic an infinitely large population. | ||
| for _ in 0..100 { | ||
| context.add_person(()).unwrap(); |
There was a problem hiding this comment.
Why do you need to unwrap() here?
There was a problem hiding this comment.
context.add_person(_) returns Result<PersonId, IxaError> to return an error in case there are person properties that need to be initialized at start-up that aren't specified!
src/infection_propagation_loop.rs
Outdated
| // population) but those people themselves cannot transmit secondary infections. We're going | ||
| // to stop those simulations at the end of 1.0 time units and compare the number of infected | ||
| // people to the infectious rate we used to set up the simulation. | ||
| // We're also going to check the times at which they are infected -- since we only record |
There was a problem hiding this comment.
Can you elaborate here why infection times are expected to be uniform?
| context.add_person(()).unwrap(); | ||
| } | ||
| // We don't want infectious people beyond our index case to be able to transmit, so we | ||
| // have to do setup on our own since just calling `init` will trigger a watcher for |
There was a problem hiding this comment.
Tbh, I don't like this but can't think of anything better. Is the goal to keep track of the number of infection attempts as in the expected number of infections in a homogeneous and large completely susceptible population? What's the difference with having two people, one infected, the other one recovered. Then infectious person would keep trying to infect the recovered person over and over again, but they wouldn't be able to transmit to others. Not suggesting to implement this but wondering if there's a more explicit/straightforward way of measuring infection attempts.
There was a problem hiding this comment.
Hmmm, I think this is a great point. You are absolutely right about the goal. I really like your suggestion instead, and that's what I implemented before. I am happy to switch to that implementation (and I think it will be faster because the population size is smaller) but do you think the code is cleaner?
ixa-epi-isolation/src/transmission_manager.rs
Line 355 in 9ee4b69
There was a problem hiding this comment.
This ended up being so much faster, thanks Guido!
| } | ||
|
|
||
| #[test] | ||
| fn test_number_timing_infections_one_time_unit() { |
There was a problem hiding this comment.
I like the idea of this test and would also like to see it as a plot outside of ixa, which I think we can recreate with the transmission chain report.
There was a problem hiding this comment.
Love it. Would love to maybe talk with @eqmooring about this.
ChiragKumar9
force-pushed
the
ckk_infection_loop_statistical_test
branch
from
1ea49ae to
21d0306
Compare
This is inspired by the tests we had the previous transmission manager (see #51's deleted files) but welcome suggestions for other statistical tests to include.