Skip to content

Change PoissonADSampler threshold to 10 #2060

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
devmotion merged 3 commits into from
Jun 19, 2026
Merged

Change PoissonADSampler threshold to 10 #2060

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
2805535
Raise `Poisson` sampler dispatch threshold from 6 to 10
k88hudson Apr 25, 2026
0d098bf
refactor: fix up tests
k88hudson Apr 29, 2026
b022b0e
Add code comment
devmotion Jun 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion src/univariate/discrete/poisson.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -135,7 +135,8 @@ fit_mle(::Type{<:Poisson}, ss::PoissonStats) = Poisson(ss.sx / ss.tw)

## samplers

const poissonsampler_threshold = 6
# PoissonADSampler is biased for μ < 10
const poissonsampler_threshold = 10
Comment thread
devmotion marked this conversation as resolved.

function sampler(d::Poisson)
if rate(d) < poissonsampler_threshold
Expand Down
10 changes: 10 additions & 0 deletions test/univariate/discrete/poisson.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,16 @@ test_cgf(Poisson(1 ), (1f0,2f0,10.0,50.0))
test_cgf(Poisson(10 ), (1f0,2f0,10.0,50.0))
test_cgf(Poisson(1e-3), (1f0,2f0,10.0,50.0))

# Regression: PoissonADSampler is biased for low values of μ
# because the Ahrens-Dieter algorithm is only designed for μ ≥ 10;
# Threshold now requires μ ≥ 10 for AD; μ < 10 uses PoissonCountSampler.
@testset "Poisson sampler bias for μ < 10" begin
n = 100_000_000
μ = 6.0
tol = 5 * sqrt(μ / n)
@test sum(rand(Poisson(μ)) for _ in 1:n) / n ≈ μ atol = tol
end

@testset "Poisson suffstats and OffsetArrays" begin
a = rand(Poisson(), 11)
wa = 1.0:11.0
Expand Down
Loading