Skip to content
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

Recover from PIRLS failure by returning finitial #616

Merged
merged 12 commits into from
Aug 9, 2023
Merged

Recover from PIRLS failure by returning finitial #616

Show file tree
Hide file tree
Changes from 9 commits
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
23b4813
attempt rescaling of poorly scaled problems
palday May 18, 2022
3c97596
recover from PIRLS failure by returning finitial
palday May 18, 2022
ed75bc8
fix try-catch
palday May 18, 2022
bbc0ca0
catch DomainError
palday May 18, 2022
0b4c410
Update src/generalizedlinearmixedmodel.jl
palday May 18, 2022
9853a8a
Merge branch 'main' of github.com:JuliaStats/MixedModels.jl into pa/p…
palday May 20, 2022
f5df1bd
Merge branch 'main' of github.com:JuliaStats/MixedModels.jl into pa/p…
palday Oct 21, 2022
7f196bb
Merge branch 'main' of github.com:JuliaStats/MixedModels.jl into pa/p…
palday Oct 21, 2022
6b3db82
Merge branch 'main' of https://github.com/JuliaStats/MixedModels.jl i…
palday Aug 9, 2023
5608759
NEWS
palday Aug 9, 2023
413d2f3
version bump
palday Aug 9, 2023
8cbaadf
try a different test
palday Aug 9, 2023
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
10 changes: 9 additions & 1 deletion src/generalizedlinearmixedmodel.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -284,7 +284,15 @@ function StatsAPI.fit!(
fitlog = optsum.fitlog
function obj(x, g)
isempty(g) || throw(ArgumentError("g should be empty for this objective"))
val = deviance(pirls!(setpar!(m, x), fast, verbose), nAGQ)
val = try
deviance(pirls!(setpar!(m, x), fast, verbose), nAGQ)
catch ex
# this allows us to recover from models where e.g. the link isn't
# as constraining as it should be
ex isa Union{PosDefException,DomainError} || rethrow()
iter == 1 && rethrow()
m.optsum.finitial
end
iszero(rem(iter, thin)) && push!(fitlog, (copy(x), val))
verbose && println(round(val; digits=5), " ", x)
progress && ProgressMeter.next!(prog; showvalues=[(:objective, val)])
Expand Down
6 changes: 5 additions & 1 deletion test/pirls.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -162,7 +162,7 @@ end
center(v::AbstractVector) = v .- (sum(v) / length(v))
grouseticks = DataFrame(dataset(:grouseticks))
grouseticks.ch = center(grouseticks.height)
gm4 = fit(MixedModel, only(gfms[:grouseticks]), grouseticks, Poisson(); fast=true, progress=false) # fails in pirls! with fast=false
gm4 = fit(MixedModel, only(gfms[:grouseticks]), grouseticks, Poisson(); fast=true, progress=false)
@test isapprox(deviance(gm4), 851.4046, atol=0.001)
# these two values are not well defined at the optimum
#@test isapprox(sum(x -> sum(abs2, x), gm4.u), 196.8695297987013, atol=0.1)
Expand All @@ -173,6 +173,10 @@ end
@test sdest(gm4) === missing
@test varest(gm4) === missing
@test gm4.σ === missing
gm4slow = fit(MixedModel, only(gfms[:grouseticks]), grouseticks, Poisson(); fast=false, progress=false)
# this tolerance isn't great, but then again the optimum isn't well defined
@test gm4.θ ≈ gm4slow.θ atol=0.05
@test gm4.β[2:end] ≈ gm4slow.β[2:end] atol=0.1
end

@testset "goldstein" begin # from a 2020-04-22 msg by Ben Goldstein to R-SIG-Mixed-Models
Expand Down