Closed aksarkar closed 4 years ago
Estimating the mode assuming g is a point mass, and fixing it in ash does better:
> ashr::ash_pois(dat$x, dat$s, mixcompdist="halfuniform", mode=sum(dat$x) / sum(dat$s))$loglik
[1] -137.1668
i tried this - i get different results. I will update ashr in a minute to make sure I have latest version, but want to record this before I do.
> library(ashr)
> dat <- readRDS("~/Downloads/pois-ex2.Rds")
> ashr::ash_pois(dat$x, dat$s, mixcompdist='halfuniform')$loglik
[1] -136.6626
The sessionInfo says I have: ashr_2.2-39
ok, when i went to update it says i have the latest version. What version of ashr do you have? what version of mixsqp?
oh, sorry i just noticed your sessionInfo.
can you update your mixsqp? I have mixsqp_0.2-2
OK, updating mixsqp fixed this example.
Test data is available here https://users.rcc.uchicago.edu/~aksarkar/pois-ex2.Rds